7+ days — read the spread, not the run

Past about a week, a single model run is one plausible future among many, and the honest question stops being “how much snow” and becomes “how many of the possible futures give me a good week”. Everything on this page answers the second question. There is deliberately no deterministic chart here — those live in 3–7 days.

Share of ensemble members reaching each snowfall total at the model grid point — not a resort forecast.

Synoptic maps — ECMWF ENS vs GEFS → The pattern behind these numbers: matched MSLP, 500 hPa height, 850 hPa temperature, precipitation, wind and ensemble spread, as a 120–240 h loop.
How to read this

Grid point, not resort. Every system is sampled at its own native grid cell, at the model's terrain height — not at a lift, village or snow stake. Those cells are not the same size: from approximately 20 km to GEPS's ~50 km grid, coarse enough that adjacent resorts routinely share one cell. Each panel states its own grid beside the member count, and the coarsest one carries a badge. The 2 m temperature plumes are raw model-grid temperature: not a resort-elevation temperature, not a wet-bulb temperature and not a snow level. Elevation-specific temperature and precipitation phase live in 3–7 days.

Provider snowfall. Every total is the provider's own snowfall variable accumulated across a member's series, in centimetres. It is not a settled-depth forecast, and nothing here is combined into a consensus between systems.

The comparison window is the denominator. The systems can run to different horizons. In the common window all are truncated to the shortest usable valid-through and every percentage, median and spread is recomputed over that shared period, so adjacent numbers answer the same question. Under full horizons each runs to its own end and totals between systems are not comparable. All charts share one time scale, so a shorter system simply draws a shorter plot and the gap on its right is the horizon difference.

Tint and strips. In the summary table the tint behind a percentage is that percentage on a fixed 0–100% scale — never a ranking against the other resorts — and the percentage itself is printed in every cell, so the colour is reinforcement rather than the encoding. Every spread strip is drawn on one shared centimetre scale within a window, read against the axis in the column heading.

Percentiles need members. A timestep with fewer than three reporting members yields no percentile at all rather than a thin one, and a member that misses an interval drops out of every total after it instead of freezing at its last value.

Show NSW VIC
Ensembles
Compare over

Common window · through Wed, 7 Oct, 11:00 AEDT — every percentage, median and spread below is accumulated over this shared period, so the systems answer the same question.GEPS runs 8 days further; that part is excluded here and shown under Full horizons.

Full horizons — each system accumulates to its own valid-through, so the totals are not comparable between rows. ECMWF IFS ENS through Wed, 14 Oct, 23:00 AEDT ECMWF AIFS ENS through Wed, 14 Oct, 23:00 AEDT GEFS through Sat, 10 Oct, 05:00 AEDT GEPS through Thu, 15 Oct, 10:00 AEDT MOGREPS-G through Sat, 10 Oct, 05:00 AEDT ICON-EPS through Wed, 7 Oct, 11:00 AEDT

Basic and grid styles are a fixed print look — palette and dark mode do not apply inside them.
All resorts — ensemble snowfall outlook ≥5 cm · ≥15 cm · ≥30 cm and percentile spread, all systems · common window through Wed, 7 Oct, 11:00 AEDT
Share of ensemble members reaching each snowfall total at the model grid point — not a resort forecast. ECMWF IFS ENS 51 members · 0.25° ≈25 km ECMWF AIFS ENS 51 members · 0.25° ≈25 km GEFS 31 members · 0.25° ≈25 km GEPS 21 members · 0.5° ≈50 km MOGREPS-G 18 members · ≈20 km ICON-EPS 40 members · 0.25° ≈26 km
Resort Ensemble ≥5 cm by Wed, 7 Oct ≥15 cm by Wed, 7 Oct ≥30 cm by Wed, 7 Oct Middle half cm 10th–90th cm Spread cm, shared scale
Perisher NSW · 1605–2042 m Perisher: systems agree Perisher, ECMWF IFS ENS 0% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher: systems agree Perisher, ECMWF AIFS ENS 4% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher: systems agree Perisher, GEFS 0% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher: systems agree Perisher, GEPS 0% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher: systems agree Perisher, MOGREPS-G 0% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher: systems agree Perisher, ICON-EPS 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo: systems agree Thredbo, ECMWF IFS ENS 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo: systems agree Thredbo, ECMWF AIFS ENS 4% 0% 0% 0–0 0–1
Thredbo NSW · 1365–2037 m Thredbo: systems agree Thredbo, GEFS 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo: systems agree Thredbo, GEPS 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo: systems agree Thredbo, MOGREPS-G 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo: systems agree Thredbo, ICON-EPS 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass: systems agree Charlotte Pass, ECMWF IFS ENS 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass: systems agree Charlotte Pass, ECMWF AIFS ENS 4% 0% 0% 0–0 0–1
Charlotte Pass NSW · 1755–1964 m Charlotte Pass: systems agree Charlotte Pass, GEFS 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass: systems agree Charlotte Pass, GEPS 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass: systems agree Charlotte Pass, MOGREPS-G 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass: systems agree Charlotte Pass, ICON-EPS 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn: systems agree Selwyn, ECMWF IFS ENS 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn: systems agree Selwyn, ECMWF AIFS ENS 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn: systems agree Selwyn, GEFS 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn: systems agree Selwyn, GEPS 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn: systems agree Selwyn, MOGREPS-G 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn: systems agree Selwyn, ICON-EPS 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham: systems agree Mt Hotham, ECMWF IFS ENS 0% 0% 0% 0–0 0–1
Mt Hotham VIC · 1450–1845 m Mt Hotham: systems agree Mt Hotham, ECMWF AIFS ENS 2% 0% 0% 0–0 0–2
Mt Hotham VIC · 1450–1845 m Mt Hotham: systems agree Mt Hotham, GEFS 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham: systems agree Mt Hotham, GEPS 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham: systems agree Mt Hotham, MOGREPS-G 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham: systems agree Mt Hotham, ICON-EPS 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek: systems agree Falls Creek, ECMWF IFS ENS 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek: systems agree Falls Creek, ECMWF AIFS ENS 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek: systems agree Falls Creek, GEFS 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek: systems agree Falls Creek, GEPS 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek: systems agree Falls Creek, MOGREPS-G 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek: systems agree Falls Creek, ICON-EPS 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller: systems agree Mt Buller, ECMWF IFS ENS 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller: systems agree Mt Buller, ECMWF AIFS ENS 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller: systems agree Mt Buller, GEFS 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller: systems agree Mt Buller, GEPS 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller: systems agree Mt Buller, MOGREPS-G 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller: systems agree Mt Buller, ICON-EPS 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw: systems agree Mt Baw Baw, ECMWF IFS ENS 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw: systems agree Mt Baw Baw, ECMWF AIFS ENS 0% 0% 0% 0–0 0–1
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw: systems agree Mt Baw Baw, GEFS 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw: systems agree Mt Baw Baw, GEPS 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw: systems agree Mt Baw Baw, MOGREPS-G 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw: systems agree Mt Baw Baw, ICON-EPS 0% 0% 0% 0–0 0–0
All resorts — ensemble snowfall outlook ≥5 cm · ≥15 cm · ≥30 cm and percentile spread, all systems
Share of ensemble members reaching each snowfall total at the model grid point — not a resort forecast. ECMWF IFS ENS 51 members · 0.25° ≈25 km through 14 Oct, 23:00 ECMWF AIFS ENS 51 members · 0.25° ≈25 km through 14 Oct, 23:00 GEFS 31 members · 0.25° ≈25 km through 10 Oct, 05:00 GEPS 21 members · 0.5° ≈50 km through 15 Oct, 10:00 MOGREPS-G 18 members · ≈20 km through 10 Oct, 05:00 ICON-EPS 40 members · 0.25° ≈26 km through 7 Oct, 11:00
Resort Ensemble ≥5 cm ≥15 cm ≥30 cm Middle half cm 10th–90th cm Spread cm, shared scale
Perisher NSW · 1605–2042 m Perisher, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 6% 2% 0% 0–2 0–4
Perisher NSW · 1605–2042 m Perisher, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 4% 0% 0% 0–0 0–2
Perisher NSW · 1605–2042 m Perisher, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 5% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Perisher NSW · 1605–2042 m Perisher, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 12% 4% 2% 0–2 0–5
Thredbo NSW · 1365–2037 m Thredbo, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 8% 0% 0% 0–1 0–4
Thredbo NSW · 1365–2037 m Thredbo, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 5% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Thredbo NSW · 1365–2037 m Thredbo, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 12% 4% 2% 0–2 0–5
Charlotte Pass NSW · 1755–1964 m Charlotte Pass, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 8% 0% 0% 0–1 0–4
Charlotte Pass NSW · 1755–1964 m Charlotte Pass, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 5% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Charlotte Pass NSW · 1755–1964 m Charlotte Pass, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 2% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 2% 0% 0% 0–0 0–1
Selwyn NSW · 1492–1614 m Selwyn, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Selwyn NSW · 1492–1614 m Selwyn, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 10% 4% 0% 0–2 0–4
Mt Hotham VIC · 1450–1845 m Mt Hotham, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 6% 0% 0% 0–1 0–3
Mt Hotham VIC · 1450–1845 m Mt Hotham, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Hotham VIC · 1450–1845 m Mt Hotham, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 2% 0% 0% 0–0 0–2
Falls Creek VIC · 1500–1780 m Falls Creek, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 2% 0% 0% 0–0 0–1
Falls Creek VIC · 1500–1780 m Falls Creek, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Falls Creek VIC · 1500–1780 m Falls Creek, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 4% 0% 0% 0–1 0–3
Mt Buller VIC · 1375–1780 m Mt Buller, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 2% 0% 0% 0–0 0–3
Mt Buller VIC · 1375–1780 m Mt Buller, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Buller VIC · 1375–1780 m Mt Buller, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw, ECMWF IFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 2% 0% 0% 0–0 0–1
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw, ECMWF AIFS ENS through 14 Oct, 23:00 (+8 days on ICON-EPS) 0% 0% 0% 0–0 0–2
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw, GEFS through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw, GEPS through 15 Oct, 10:00 (+8 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw, MOGREPS-G through 10 Oct, 05:00 (+3 days on ICON-EPS) 0% 0% 0% 0–0 0–0
Mt Baw Baw VIC · 1450–1564 m Mt Baw Baw, ICON-EPS through 7 Oct, 11:00 0% 0% 0% 0–0 0–0

Perisher NSW · 1605–2042 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Perisher →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Perisher · 51 members · 0.25° ≈25 km · grid elev 1333 m

Cumulative snowfall Perisher ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 17.4 to 17.9 degrees Celsius; by the end they give 4.8 to 11.7, with a 10th-to-90th spread of 3.7 to 13.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Perisher ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–2 cm · 10th–90th 0–4 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 2 cm, with a full spread of 0 to 4 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 4 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 1 2 3 cm
6% reach 5 cm
2% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 17.4 to 17.9 degrees Celsius; by the end they give 3.0 to 5.5, with a 10th-to-90th spread of 1.3 to 6.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 25% 100% 96% 49% 22% 69% 35% 18% 29% 59% 39% 69% 71% 53%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Perisher · 51 members · 0.25° ≈25 km · grid elev 1104 m

Cumulative snowfall Perisher ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
4% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.0 to 17.3 degrees Celsius; by the end they give 5.7 to 10.2, with a 10th-to-90th spread of 4.8 to 12.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Perisher ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–2 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 2 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 3 of them pass above the 4 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 1 2 3 cm
4% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.0 to 17.3 degrees Celsius; by the end they give 5.0 to 10.2, with a 10th-to-90th spread of 3.9 to 11.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 22% 100% 98% 51% 59% 43% 49% 29% 14% 27% 41% 63% 45% 41%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Perisher · 31 members · 0.25° ≈25 km · grid elev 1276 m

Cumulative snowfall Perisher GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 17.1 to 17.5 degrees Celsius; by the end they give 9.9 to 16.1, with a 10th-to-90th spread of 8.7 to 17.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Perisher GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 1 2 3 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 17.1 to 17.5 degrees Celsius; by the end they give 7.5 to 12.7, with a 10th-to-90th spread of 4.9 to 14.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 32% 100% 100% 58% 16% 48% 23% 10% 32% 19%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Perisher · 21 members · 0.5° ≈50 km · grid elev 961 m

Cumulative snowfall Perisher GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.3 to 16.0 degrees Celsius; by the end they give 7.9 to 13.2, with a 10th-to-90th spread of 6.7 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Perisher GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 4 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 1 2 3 cm
5% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.3 to 16.0 degrees Celsius; by the end they give 8.7 to 13.1, with a 10th-to-90th spread of 5.8 to 15.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
0% 10% 95% 90% 71% 10% 57% 43% 38% 14% 48% 52% 38% 43% 67% 33%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Perisher · 18 members · ≈20 km · grid elev 1128 m

Cumulative snowfall Perisher MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 14.5 to 15.3 degrees Celsius; by the end they give 8.5 to 11.4, with a 10th-to-90th spread of 6.1 to 12.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Perisher MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 1 2 3 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.5 to 15.3 degrees Celsius; by the end they give 8.7 to 11.7, with a 10th-to-90th spread of 6.9 to 12.3.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 0% 100% 100% 72% 0% 61% 28% 17% 11% 22%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Perisher · 40 members · 0.25° ≈26 km · grid elev 1072 m

Cumulative snowfall Perisher ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 18.4 to 19.1 degrees Celsius; by the end they give 10.6 to 15.0, with a 10th-to-90th spread of 8.8 to 15.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Perisher ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 1 2 3 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 18.4 to 19.1 degrees Celsius; by the end they give 10.6 to 15.0, with a 10th-to-90th spread of 8.8 to 15.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
0% 10% 100% 83% 25% 23% 28% 18%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Perisher · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.00.00.317.411.30.40.21.80.10.10.0
10th–90th 0.0–0.0 0.0–0.1 0.1–0.5 9.8–28.3 2.9–20.3 0.0–2.1 0.1–0.4 0.2–4.8 0.0–5.7 0.0–5.6 0.0–0.4
Chance ≥1 mm 0% 0% 3% 100% 97% 23% 3% 65% 32% 26% 10%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.

Model run Tue 29 1600hrs AEST · 0600Z · 19 h ago · fetched 11 h ago

Calculation method

Every member of NOAA AIGEFS is retrieved direct from NOMADS as GRIB2 (surface APCP, six-hourly accumulations) and sampled at the resort point. Member totals are summed to AEST calendar days, and the percentiles and the ≥1 mm chance are computed across those per-member daily totals — never from the published mean and spread. No snowfall field is published for this system and no member pressure profiles are retrieved, so no snow depth, snow level or precipitation phase is derived from it. Background

NOAA/NCEP AIGEFS, via NOMADS

Thredbo NSW · 1365–2037 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Thredbo →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Thredbo · 51 members · 0.25° ≈25 km · grid elev 1450 m

Cumulative snowfall Thredbo ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 3 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 13.3 to 13.7 degrees Celsius; by the end they give 5.7 to 13.7, with a 10th-to-90th spread of 4.4 to 14.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Thredbo ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–2 cm · 10th–90th 0–5 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 2 cm, with a full spread of 0 to 5 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 5 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 2 4 cm
12% reach 5 cm
4% reach 15 cm
2% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 13.3 to 13.7 degrees Celsius; by the end they give 2.3 to 5.7, with a 10th-to-90th spread of 0.7 to 8.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 45% 100% 98% 84% 27% 69% 20% 24% 33% 57% 41% 61% 59% 43%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Thredbo · 51 members · 0.25° ≈25 km · grid elev 1194 m

Cumulative snowfall Thredbo ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–1 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 1 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
4% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 14.6 to 15.7 degrees Celsius; by the end they give 5.2 to 10.9, with a 10th-to-90th spread of 4.3 to 12.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Thredbo ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–1 cm · 10th–90th 0–4 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 1 cm, with a full spread of 0 to 4 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 5 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 2 4 cm
8% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.6 to 15.7 degrees Celsius; by the end they give 4.3 to 9.2, with a 10th-to-90th spread of 3.0 to 11.0.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 37% 100% 100% 67% 65% 51% 45% 25% 16% 27% 47% 65% 51% 39%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Thredbo · 31 members · 0.25° ≈25 km · grid elev 1316 m

Cumulative snowfall Thredbo GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.5 to 17.0 degrees Celsius; by the end they give 10.3 to 17.1, with a 10th-to-90th spread of 9.1 to 18.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Thredbo GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.5 to 17.0 degrees Celsius; by the end they give 7.9 to 12.7, with a 10th-to-90th spread of 5.6 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 48% 100% 100% 55% 19% 48% 23% 13% 32% 26%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Thredbo · 21 members · 0.5° ≈50 km · grid elev 961 m

Cumulative snowfall Thredbo GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.3 to 16.0 degrees Celsius; by the end they give 7.9 to 13.2, with a 10th-to-90th spread of 6.7 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Thredbo GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 5 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 2 4 cm
5% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.3 to 16.0 degrees Celsius; by the end they give 8.7 to 13.1, with a 10th-to-90th spread of 5.8 to 15.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
0% 10% 95% 90% 71% 10% 57% 43% 38% 14% 48% 52% 38% 43% 67% 33%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Thredbo · 18 members · ≈20 km · grid elev 1051 m

Cumulative snowfall Thredbo MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 12.4 to 13.1 degrees Celsius; by the end they give 8.7 to 11.7, with a 10th-to-90th spread of 5.3 to 13.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Thredbo MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 12.4 to 13.1 degrees Celsius; by the end they give 6.8 to 8.6, with a 10th-to-90th spread of 5.1 to 10.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 44% 100% 100% 89% 28% 67% 33% 28% 33% 22%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Thredbo · 40 members · 0.25° ≈26 km · grid elev 1090 m

Cumulative snowfall Thredbo ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.9 to 16.3 degrees Celsius; by the end they give 11.3 to 15.4, with a 10th-to-90th spread of 9.3 to 16.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Thredbo ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.9 to 16.3 degrees Celsius; by the end they give 11.3 to 15.4, with a 10th-to-90th spread of 9.3 to 16.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
0% 50% 100% 85% 40% 18% 25% 15%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Thredbo · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.00.00.719.99.10.30.32.40.10.00.0
10th–90th 0.0–0.0 0.0–0.1 0.3–1.3 13.7–30.5 1.4–18.4 0.0–1.6 0.1–0.8 0.4–8.3 0.0–4.3 0.0–1.9 0.0–0.6
Chance ≥1 mm 0% 0% 32% 100% 90% 16% 3% 71% 32% 16% 3%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.

Model run Tue 29 1600hrs AEST · 0600Z · 19 h ago · fetched 11 h ago

Calculation method

Every member of NOAA AIGEFS is retrieved direct from NOMADS as GRIB2 (surface APCP, six-hourly accumulations) and sampled at the resort point. Member totals are summed to AEST calendar days, and the percentiles and the ≥1 mm chance are computed across those per-member daily totals — never from the published mean and spread. No snowfall field is published for this system and no member pressure profiles are retrieved, so no snow depth, snow level or precipitation phase is derived from it. Background

NOAA/NCEP AIGEFS, via NOMADS

Charlotte Pass NSW · 1755–1964 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Charlotte Pass →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Charlotte Pass · 51 members · 0.25° ≈25 km · grid elev 1450 m

Cumulative snowfall Charlotte Pass ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 3 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 13.3 to 13.7 degrees Celsius; by the end they give 5.7 to 13.7, with a 10th-to-90th spread of 4.4 to 14.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Charlotte Pass ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–2 cm · 10th–90th 0–5 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 2 cm, with a full spread of 0 to 5 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 5 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 2 4 cm
12% reach 5 cm
4% reach 15 cm
2% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 13.3 to 13.7 degrees Celsius; by the end they give 2.3 to 5.7, with a 10th-to-90th spread of 0.7 to 8.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 45% 100% 98% 84% 27% 69% 20% 24% 33% 57% 41% 61% 59% 43%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Charlotte Pass · 51 members · 0.25° ≈25 km · grid elev 1194 m

Cumulative snowfall Charlotte Pass ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–1 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 1 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
4% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 14.6 to 15.7 degrees Celsius; by the end they give 5.2 to 10.9, with a 10th-to-90th spread of 4.3 to 12.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Charlotte Pass ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–1 cm · 10th–90th 0–4 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 1 cm, with a full spread of 0 to 4 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 5 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 2 4 cm
8% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.6 to 15.7 degrees Celsius; by the end they give 4.3 to 9.2, with a 10th-to-90th spread of 3.0 to 11.0.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 37% 100% 100% 67% 65% 51% 45% 25% 16% 27% 47% 65% 51% 39%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Charlotte Pass · 31 members · 0.25° ≈25 km · grid elev 1316 m

Cumulative snowfall Charlotte Pass GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.5 to 17.0 degrees Celsius; by the end they give 10.3 to 17.1, with a 10th-to-90th spread of 9.1 to 18.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Charlotte Pass GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.5 to 17.0 degrees Celsius; by the end they give 7.9 to 12.7, with a 10th-to-90th spread of 5.6 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 48% 100% 100% 55% 19% 48% 23% 13% 32% 26%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Charlotte Pass · 21 members · 0.5° ≈50 km · grid elev 961 m

Cumulative snowfall Charlotte Pass GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.3 to 16.0 degrees Celsius; by the end they give 7.9 to 13.2, with a 10th-to-90th spread of 6.7 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Charlotte Pass GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 5 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 2 4 cm
5% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.3 to 16.0 degrees Celsius; by the end they give 8.7 to 13.1, with a 10th-to-90th spread of 5.8 to 15.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
0% 10% 95% 90% 71% 10% 57% 43% 38% 14% 48% 52% 38% 43% 67% 33%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Charlotte Pass · 18 members · ≈20 km · grid elev 1066 m

Cumulative snowfall Charlotte Pass MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 11.9 to 12.3 degrees Celsius; by the end they give 8.6 to 11.9, with a 10th-to-90th spread of 6.3 to 14.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Charlotte Pass MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 11.9 to 12.3 degrees Celsius; by the end they give 5.8 to 9.2, with a 10th-to-90th spread of 4.8 to 10.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 44% 100% 100% 78% 39% 67% 17% 28% 33% 22%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Charlotte Pass · 40 members · 0.25° ≈26 km · grid elev 1090 m

Cumulative snowfall Charlotte Pass ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.9 to 16.3 degrees Celsius; by the end they give 11.3 to 15.4, with a 10th-to-90th spread of 9.3 to 16.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Charlotte Pass ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.9 to 16.3 degrees Celsius; by the end they give 11.3 to 15.4, with a 10th-to-90th spread of 9.3 to 16.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
0% 50% 100% 85% 40% 18% 25% 15%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Charlotte Pass · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.00.00.719.99.10.30.32.40.10.00.0
10th–90th 0.0–0.0 0.0–0.1 0.3–1.3 13.7–30.5 1.4–18.4 0.0–1.6 0.1–0.8 0.4–8.3 0.0–4.3 0.0–1.9 0.0–0.6
Chance ≥1 mm 0% 0% 32% 100% 90% 16% 3% 71% 32% 16% 3%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.

Model run Tue 29 1600hrs AEST · 0600Z · 19 h ago · fetched 11 h ago

Calculation method

Every member of NOAA AIGEFS is retrieved direct from NOMADS as GRIB2 (surface APCP, six-hourly accumulations) and sampled at the resort point. Member totals are summed to AEST calendar days, and the percentiles and the ≥1 mm chance are computed across those per-member daily totals — never from the published mean and spread. No snowfall field is published for this system and no member pressure profiles are retrieved, so no snow depth, snow level or precipitation phase is derived from it. Background

NOAA/NCEP AIGEFS, via NOMADS

Selwyn NSW · 1492–1614 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Selwyn →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Selwyn · 51 members · 0.25° ≈25 km · grid elev 1223 m

Cumulative snowfall Selwyn ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.2 to 15.5 degrees Celsius; by the end they give 8.9 to 14.6, with a 10th-to-90th spread of 7.8 to 16.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Selwyn ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 0.5 1 cm
2% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.2 to 15.5 degrees Celsius; by the end they give 3.5 to 6.8, with a 10th-to-90th spread of 2.3 to 8.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 6% 100% 100% 61% 14% 55% 18% 22% 27% 59% 39% 55% 63% 43%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Selwyn · 51 members · 0.25° ≈25 km · grid elev 1151 m

Cumulative snowfall Selwyn ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 14.2 to 15.3 degrees Celsius; by the end they give 7.1 to 11.6, with a 10th-to-90th spread of 6.3 to 13.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Selwyn ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–1 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 1 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 0.5 1 cm
2% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.2 to 15.3 degrees Celsius; by the end they give 4.9 to 10.1, with a 10th-to-90th spread of 3.5 to 11.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
0% 16% 100% 100% 45% 61% 49% 37% 22% 12% 25% 39% 51% 51% 35%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Selwyn · 31 members · 0.25° ≈25 km · grid elev 1245 m

Cumulative snowfall Selwyn GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.0 to 16.5 degrees Celsius; by the end they give 11.9 to 16.7, with a 10th-to-90th spread of 9.9 to 17.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Selwyn GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.0 to 16.5 degrees Celsius; by the end they give 8.3 to 11.8, with a 10th-to-90th spread of 7.0 to 13.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 23% 100% 97% 58% 19% 52% 19% 16% 32% 23%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Selwyn · 21 members · 0.5° ≈50 km · grid elev 1115 m

Cumulative snowfall Selwyn GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 13.7 to 14.4 degrees Celsius; by the end they give 7.3 to 12.6, with a 10th-to-90th spread of 6.9 to 13.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Selwyn GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 13.7 to 14.4 degrees Celsius; by the end they give 9.4 to 12.3, with a 10th-to-90th spread of 6.3 to 14.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
0% 5% 95% 95% 57% 5% 57% 52% 33% 19% 48% 52% 33% 33% 52% 29%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Selwyn · 18 members · ≈20 km · grid elev 1163 m

Cumulative snowfall Selwyn MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 13.0 to 13.4 degrees Celsius; by the end they give 9.8 to 11.9, with a 10th-to-90th spread of 7.2 to 12.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Selwyn MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 13.0 to 13.4 degrees Celsius; by the end they give 7.4 to 10.4, with a 10th-to-90th spread of 5.7 to 11.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 0% 100% 94% 61% 0% 50% 6% 28% 22% 22%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Selwyn · 40 members · 0.25° ≈26 km · grid elev 1322 m

Cumulative snowfall Selwyn ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.4 to 14.6 degrees Celsius; by the end they give 10.4 to 13.4, with a 10th-to-90th spread of 8.2 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Selwyn ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.4 to 14.6 degrees Celsius; by the end they give 10.4 to 13.4, with a 10th-to-90th spread of 8.2 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
0% 8% 100% 85% 35% 15% 35% 23%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Selwyn · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.00.00.417.17.10.70.41.60.10.10.0
10th–90th 0.0–0.0 0.0–0.1 0.2–0.7 11.7–25.9 1.3–14.0 0.1–2.7 0.1–1.1 0.4–6.9 0.0–6.0 0.0–7.3 0.0–0.9
Chance ≥1 mm 0% 0% 3% 100% 90% 42% 26% 68% 32% 19% 10%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.

Model run Tue 29 1600hrs AEST · 0600Z · 19 h ago · fetched 11 h ago

Calculation method

Every member of NOAA AIGEFS is retrieved direct from NOMADS as GRIB2 (surface APCP, six-hourly accumulations) and sampled at the resort point. Member totals are summed to AEST calendar days, and the percentiles and the ≥1 mm chance are computed across those per-member daily totals — never from the published mean and spread. No snowfall field is published for this system and no member pressure profiles are retrieved, so no snow depth, snow level or precipitation phase is derived from it. Background

NOAA/NCEP AIGEFS, via NOMADS

Mt Hotham VIC · 1450–1845 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Mt Hotham →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Mt Hotham · 51 members · 0.25° ≈25 km · grid elev 1423 m

Cumulative snowfall Mt Hotham ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–1 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 1 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 2 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 14.8 to 15.1 degrees Celsius; by the end they give 4.8 to 13.9, with a 10th-to-90th spread of 3.4 to 15.0.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Hotham ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–2 cm · 10th–90th 0–4 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 2 cm, with a full spread of 0 to 4 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 4 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 2 4 cm
10% reach 5 cm
4% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.8 to 15.1 degrees Celsius; by the end they give 1.4 to 4.9, with a 10th-to-90th spread of -0.3 to 6.3.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
41% 98% 100% 100% 53% 24% 65% 16% 18% 45% 59% 43% 63% 59% 45%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Mt Hotham · 51 members · 0.25° ≈25 km · grid elev 1218 m

Cumulative snowfall Mt Hotham ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–2 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 2 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 2 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
2% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 14.4 to 15.5 degrees Celsius; by the end they give 4.4 to 9.7, with a 10th-to-90th spread of 3.5 to 12.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Hotham ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–1 cm · 10th–90th 0–3 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 1 cm, with a full spread of 0 to 3 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 4 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 2 4 cm
6% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.4 to 15.5 degrees Celsius; by the end they give 3.6 to 8.6, with a 10th-to-90th spread of 2.2 to 11.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
2% 90% 100% 100% 80% 57% 67% 35% 25% 10% 33% 45% 63% 53% 41%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Mt Hotham · 31 members · 0.25° ≈25 km · grid elev 1329 m

Cumulative snowfall Mt Hotham GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.8 to 17.1 degrees Celsius; by the end they give 9.8 to 16.4, with a 10th-to-90th spread of 9.0 to 17.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Hotham GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.8 to 17.1 degrees Celsius; by the end they give 8.8 to 12.4, with a 10th-to-90th spread of 4.0 to 13.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
23% 90% 100% 100% 39% 26% 58% 23% 16% 55% 32%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Mt Hotham · 21 members · 0.5° ≈50 km · grid elev 860 m

Cumulative snowfall Mt Hotham GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.0 to 15.9 degrees Celsius; by the end they give 11.0 to 15.0, with a 10th-to-90th spread of 7.2 to 15.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Hotham GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.0 to 15.9 degrees Celsius; by the end they give 9.6 to 12.8, with a 10th-to-90th spread of 7.2 to 15.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
0% 81% 100% 100% 57% 14% 48% 29% 33% 24% 62% 62% 33% 48% 43% 14%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Mt Hotham · 18 members · ≈20 km · grid elev 923 m

Cumulative snowfall Mt Hotham MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 13.4 to 13.9 degrees Celsius; by the end they give 9.6 to 13.6, with a 10th-to-90th spread of 6.7 to 15.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Hotham MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 13.4 to 13.9 degrees Celsius; by the end they give 7.7 to 10.8, with a 10th-to-90th spread of 5.9 to 12.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 94% 100% 100% 72% 39% 72% 6% 22% 44% 28%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Mt Hotham · 40 members · 0.25° ≈26 km · grid elev 1252 m

Cumulative snowfall Mt Hotham ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.7 to 15.2 degrees Celsius; by the end they give 9.9 to 13.7, with a 10th-to-90th spread of 7.2 to 14.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Hotham ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 2 4 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.7 to 15.2 degrees Celsius; by the end they give 9.9 to 13.7, with a 10th-to-90th spread of 7.2 to 14.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
0% 98% 98% 83% 45% 20% 28% 23%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Mt Hotham · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.10.14.024.38.20.40.62.70.20.10.1
10th–90th 0.0–0.1 0.1–0.2 2.7–6.3 16.3–33.9 0.9–20.1 0.1–1.4 0.2–1.1 1.0–8.6 0.0–5.5 0.0–2.0 0.0–0.4
Chance ≥1 mm 0% 0% 100% 100% 87% 16% 16% 90% 29% 13% 3%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.

Model run Tue 29 1600hrs AEST · 0600Z · 19 h ago · fetched 11 h ago

Calculation method

Every member of NOAA AIGEFS is retrieved direct from NOMADS as GRIB2 (surface APCP, six-hourly accumulations) and sampled at the resort point. Member totals are summed to AEST calendar days, and the percentiles and the ≥1 mm chance are computed across those per-member daily totals — never from the published mean and spread. No snowfall field is published for this system and no member pressure profiles are retrieved, so no snow depth, snow level or precipitation phase is derived from it. Background

NOAA/NCEP AIGEFS, via NOMADS

Falls Creek VIC · 1500–1780 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Falls Creek →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Falls Creek · 51 members · 0.25° ≈25 km · grid elev 1114 m

Cumulative snowfall Falls Creek ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.6 to 16.0 degrees Celsius; by the end they give 8.9 to 14.6, with a 10th-to-90th spread of 7.7 to 16.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Falls Creek ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–2 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 2 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 2 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 0.5 1 1.5 cm
2% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.6 to 16.0 degrees Celsius; by the end they give 4.0 to 7.5, with a 10th-to-90th spread of 3.0 to 9.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
45% 98% 100% 100% 75% 37% 65% 14% 24% 47% 65% 37% 59% 51% 37%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Falls Creek · 51 members · 0.25° ≈25 km · grid elev 919 m

Cumulative snowfall Falls Creek ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 2 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.9 to 16.9 degrees Celsius; by the end they give 7.2 to 11.8, with a 10th-to-90th spread of 6.7 to 14.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Falls Creek ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–1 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 1 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 2 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 0.5 1 1.5 cm
2% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.9 to 16.9 degrees Celsius; by the end they give 5.6 to 10.0, with a 10th-to-90th spread of 4.2 to 12.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
4% 90% 100% 100% 71% 59% 55% 33% 16% 8% 35% 45% 57% 43% 31%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Falls Creek · 31 members · 0.25° ≈25 km · grid elev 1034 m

Cumulative snowfall Falls Creek GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 17.4 to 17.6 degrees Celsius; by the end they give 11.9 to 17.4, with a 10th-to-90th spread of 11.2 to 18.3.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Falls Creek GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 17.4 to 17.6 degrees Celsius; by the end they give 9.2 to 12.6, with a 10th-to-90th spread of 5.7 to 13.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
10% 90% 100% 94% 29% 19% 45% 16% 19% 48% 29%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Falls Creek · 21 members · 0.5° ≈50 km · grid elev 929 m

Cumulative snowfall Falls Creek GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.2 to 16.2 degrees Celsius; by the end they give 7.3 to 14.4, with a 10th-to-90th spread of 6.5 to 15.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Falls Creek GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 2 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.2 to 16.2 degrees Celsius; by the end they give 8.2 to 13.3, with a 10th-to-90th spread of 6.9 to 16.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
0% 33% 100% 95% 71% 10% 33% 38% 29% 24% 57% 57% 43% 43% 38% 19%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Falls Creek · 18 members · ≈20 km · grid elev 972 m

Cumulative snowfall Falls Creek MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 13.7 to 14.6 degrees Celsius; by the end they give 10.6 to 13.8, with a 10th-to-90th spread of 6.7 to 15.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Falls Creek MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 13.7 to 14.6 degrees Celsius; by the end they give 8.5 to 11.6, with a 10th-to-90th spread of 7.3 to 13.0.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 83% 100% 100% 61% 33% 50% 6% 17% 39% 28%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Falls Creek · 40 members · 0.25° ≈26 km · grid elev 677 m

Cumulative snowfall Falls Creek ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 17.3 to 17.7 degrees Celsius; by the end they give 14.1 to 17.2, with a 10th-to-90th spread of 11.3 to 17.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Falls Creek ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 17.3 to 17.7 degrees Celsius; by the end they give 14.1 to 17.2, with a 10th-to-90th spread of 11.3 to 17.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
0% 78% 98% 78% 45% 23% 33% 18%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Falls Creek · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.00.14.923.85.20.10.62.80.00.10.1
10th–90th 0.0–0.1 0.1–0.3 3.2–9.2 15.2–37.4 0.3–12.8 0.0–0.6 0.2–1.8 0.7–10.9 0.0–3.9 0.0–0.3 0.0–0.4
Chance ≥1 mm 0% 0% 100% 100% 84% 10% 26% 84% 26% 6% 6%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.

Model run Tue 29 1600hrs AEST · 0600Z · 19 h ago · fetched 11 h ago

Calculation method

Every member of NOAA AIGEFS is retrieved direct from NOMADS as GRIB2 (surface APCP, six-hourly accumulations) and sampled at the resort point. Member totals are summed to AEST calendar days, and the percentiles and the ≥1 mm chance are computed across those per-member daily totals — never from the published mean and spread. No snowfall field is published for this system and no member pressure profiles are retrieved, so no snow depth, snow level or precipitation phase is derived from it. Background

NOAA/NCEP AIGEFS, via NOMADS

Mt Buller VIC · 1375–1780 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Mt Buller →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Mt Buller · 51 members · 0.25° ≈25 km · grid elev 1118 m

Cumulative snowfall Mt Buller ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.8 to 17.2 degrees Celsius; by the end they give 7.5 to 15.0, with a 10th-to-90th spread of 5.3 to 15.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Buller ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–1 cm · 10th–90th 0–3 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 1 cm, with a full spread of 0 to 3 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 3 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 1 2 cm
4% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.8 to 17.2 degrees Celsius; by the end they give 3.0 to 5.5, with a 10th-to-90th spread of 2.1 to 7.7.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
16% 100% 100% 100% 73% 27% 71% 16% 20% 49% 63% 45% 67% 57% 37%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Mt Buller · 51 members · 0.25° ≈25 km · grid elev 960 m

Cumulative snowfall Mt Buller ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 3 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.1 to 17.1 degrees Celsius; by the end they give 6.3 to 11.0, with a 10th-to-90th spread of 5.7 to 14.3.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Buller ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–3 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 3 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 3 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 1 2 cm
2% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.1 to 17.1 degrees Celsius; by the end they give 5.5 to 9.6, with a 10th-to-90th spread of 3.4 to 12.3.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
20% 100% 100% 100% 76% 55% 61% 33% 16% 8% 47% 49% 55% 51% 29%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Mt Buller · 31 members · 0.25° ≈25 km · grid elev 1063 m

Cumulative snowfall Mt Buller GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 17.4 to 17.9 degrees Celsius; by the end they give 10.9 to 17.1, with a 10th-to-90th spread of 9.5 to 18.0.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Buller GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 1 2 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 17.4 to 17.9 degrees Celsius; by the end they give 8.8 to 13.1, with a 10th-to-90th spread of 4.4 to 13.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
74% 97% 100% 100% 29% 16% 65% 16% 6% 58% 32%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Mt Buller · 21 members · 0.5° ≈50 km · grid elev 674 m

Cumulative snowfall Mt Buller GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 15.2 to 16.0 degrees Celsius; by the end they give 12.0 to 15.6, with a 10th-to-90th spread of 9.2 to 17.0.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Buller GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 1 2 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 15.2 to 16.0 degrees Celsius; by the end they give 10.9 to 13.3, with a 10th-to-90th spread of 8.7 to 16.0.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 1501020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
10% 95% 100% 86% 33% 10% 43% 19% 24% 38% 67% 71% 43% 38% 33% 10%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Mt Buller · 18 members · ≈20 km · grid elev 811 m

Cumulative snowfall Mt Buller MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 14.5 to 15.5 degrees Celsius; by the end they give 12.4 to 16.1, with a 10th-to-90th spread of 9.3 to 17.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Buller MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 1 2 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 14.5 to 15.5 degrees Celsius; by the end they give 8.9 to 11.4, with a 10th-to-90th spread of 7.1 to 13.6.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 1001020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 94% 100% 100% 44% 28% 56% 6% 17% 44% 22%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Mt Buller · 40 members · 0.25° ≈26 km · grid elev 1043 m

Cumulative snowfall Mt Buller ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.6 to 17.0 degrees Celsius; by the end they give 11.1 to 15.1, with a 10th-to-90th spread of 8.7 to 16.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Cumulative snowfall Mt Buller ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 1 2 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.6 to 17.0 degrees Celsius; by the end they give 11.1 to 15.1, with a 10th-to-90th spread of 8.7 to 16.2.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 701020 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
5% 100% 95% 75% 43% 15% 33% 15%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Mt Buller · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.10.39.019.33.50.30.63.70.10.10.1
10th–90th 0.0–0.1 0.1–0.9 6.0–11.6 13.1–31.5 0.3–17.2 0.0–0.9 0.2–1.4 1.2–11.0 0.0–2.3 0.0–0.7 0.0–0.4
Chance ≥1 mm 0% 10% 100% 100% 74% 6% 19% 94% 26% 3% 0%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.

Model run Tue 29 1600hrs AEST · 0600Z · 19 h ago · fetched 11 h ago

Calculation method

Every member of NOAA AIGEFS is retrieved direct from NOMADS as GRIB2 (surface APCP, six-hourly accumulations) and sampled at the resort point. Member totals are summed to AEST calendar days, and the percentiles and the ≥1 mm chance are computed across those per-member daily totals — never from the published mean and spread. No snowfall field is published for this system and no member pressure profiles are retrieved, so no snow depth, snow level or precipitation phase is derived from it. Background

NOAA/NCEP AIGEFS, via NOMADS

Mt Baw Baw VIC · 1450–1564 m

Separate ensemble perspectives; none is combined into a consensus forecast. Over the shared window: systems agree All panels for Mt Baw Baw →

10th–90th 25th–75th median members, when shown weekend

ECMWF IFS ENS

Mt Baw Baw · 51 members · 0.25° ≈25 km · grid elev 962 m

Cumulative snowfall Mt Baw Baw ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 1 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.7 to 17.4 degrees Celsius; by the end they give 6.0 to 14.2, with a 10th-to-90th spread of 4.8 to 15.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 70102030 °C

Cumulative snowfall Mt Baw Baw ECMWF IFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–1 cm
Cumulative provider snowfall across 51 ECMWF IFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 1 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 2 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 0.5 1 1.5 cm
2% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF IFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 01:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF IFS ENS members Model-grid 2 m temperature at the ECMWF IFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T14:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.7 to 17.4 degrees Celsius; by the end they give 3.7 to 6.2, with a 10th-to-90th spread of 2.8 to 7.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 150102030 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
6% 100% 100% 100% 88% 16% 76% 29% 22% 47% 59% 45% 73% 67% 55%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 01:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF IFS 0.25° ENS has 51 members and a 15-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • ECMWF publishes its long ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF ENS — © ECMWF, licensed CC BY 4.0

ECMWF AIFS ENS

Mt Baw Baw · 51 members · 0.25° ≈25 km · grid elev 837 m

Cumulative snowfall Mt Baw Baw ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–1 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 1 cm. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 5 of them pass above the 1 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 17.3 to 18.8 degrees Celsius; by the end they give 5.9 to 10.9, with a 10th-to-90th spread of 5.4 to 13.3.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 70102030 °C

Cumulative snowfall Mt Baw Baw ECMWF AIFS ENS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 14 Oct, 23:00 AEST

Middle half 0–0 cm · 10th–90th 0–2 cm
Cumulative provider snowfall across 51 ECMWF AIFS ENS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T12:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 2 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 51 individual member accumulations as faint unranked lines, none of which is a forecast. When those lines are drawn, 4 of them pass above the 2 centimetre axis top and are faded out at it rather than ending there. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ECMWF AIFS ENS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 04:00 AEST

Model-grid 2 m temperature spread across 51 ECMWF AIFS ENS members Model-grid 2 m temperature at the ECMWF AIFS ENS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T17:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 17.3 to 18.8 degrees Celsius; by the end they give 6.0 to 10.2, with a 10th-to-90th spread of 3.8 to 12.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 150102030 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-14
14% 100% 100% 100% 86% 43% 75% 45% 22% 12% 45% 53% 76% 55% 43%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 04:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 14 Oct, 23:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • ECMWF AIFS 0.25° ENS has 51 members and a 15-day, 6-hourly native source horizon; Open-Meteo interpolates the requested point series to hourly.
  • For AIFS 0.25° is the native model resolution, not a dissemination grid, so its terrain is flatter than IFS ENS at the identical grid cell: about 230 m lower at Perisher (1104 m against 1333 m) and about 260 m lower at Thredbo (1194 m against 1450 m). A lower surface runs warmer, so part of any AIFS-versus-IFS snowfall gap follows from grid elevation rather than from a genuine forecast disagreement.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • ECMWF publishes its long AIFS ENS runs off 00Z and 12Z, so the newest available run is routinely more than six hours old without anything being wrong.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 10 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECMWF AIFS — © ECMWF, licensed CC BY 4.0

GEFS

Mt Baw Baw · 31 members · 0.25° ≈25 km · grid elev 928 m

Cumulative snowfall Mt Baw Baw GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 18.5 to 19.0 degrees Celsius; by the end they give 10.1 to 15.4, with a 10th-to-90th spread of 8.5 to 18.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 70102030 °C

Cumulative snowfall Mt Baw Baw GEFS Tue 29 1800Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 31 GEFS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 31 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEFS Tue 29 1800Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 07:00 AEST

Model-grid 2 m temperature spread across 31 GEFS members Model-grid 2 m temperature at the GEFS 0.25° ≈25 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T20:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 18.5 to 19.0 degrees Celsius; by the end they give 9.1 to 14.3, with a 10th-to-90th spread of 4.9 to 16.4.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 100102030 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
45% 100% 100% 100% 32% 23% 68% 26% 6% 58% 32%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 07:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.25° grid (about 25 km); point values are not resort-scale terrain forecasts.
  • GEFS 0.25° has 31 members and a 10-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.

Model run Wed 30 0400hrs AEST · 1800Z · 7 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · NOAA GEFS — NOAA/NCEP, public domain

GEPS 0.5°

Mt Baw Baw · 21 members · 0.5° ≈50 km · grid elev 296 m

Cumulative snowfall Mt Baw Baw GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 20.8 to 22.2 degrees Celsius; by the end they give 12.4 to 16.3, with a 10th-to-90th spread of 12.0 to 18.1.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 70102030 °C

Cumulative snowfall Mt Baw Baw GEPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 21 GEPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 21 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 Sun 11 Mon 12 Tue 13 Wed 14 Thu 15 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature GEPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 15 Oct, 10:00 AEST

Model-grid 2 m temperature spread across 21 GEPS members Model-grid 2 m temperature at the GEPS 0.5° ≈50 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-14T23:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 20.8 to 22.2 degrees Celsius; by the end they give 11.9 to 17.6, with a 10th-to-90th spread of 10.8 to 19.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 10Sun 11Mon 12Tue 13Wed 14Thu 150102030 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-1010-1110-1210-1310-1410-15
0% 100% 100% 100% 62% 5% 57% 33% 29% 14% 52% 52% 48% 43% 38% 14%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Thu, 15 Oct, 10:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global 0.5° grid (about 50 km) — the coarsest grid of the systems shown beside it; adjacent resorts routinely share one grid cell.
  • GEPS has 21 members (control + 20) and a 16-day, 3-hourly source horizon; Open-Meteo serves the requested point series hourly.
  • GEPS publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · ECCC GEPS — © Environment and Climate Change Canada

MOGREPS-G

Mt Baw Baw · 18 members · ≈20 km · grid elev 584 m

Cumulative snowfall Mt Baw Baw MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and truncated to the comparison window rather than to this system's own horizon. At the start of that interval the middle half of members give 16.4 to 17.6 degrees Celsius; by the end they give 9.0 to 12.9, with a 10th-to-90th spread of 7.3 to 15.8.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 70102030 °C

Cumulative snowfall Mt Baw Baw MOGREPS-G Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 18 MOGREPS-G members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 18 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 Thu 8 Fri 9 Sat 10 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature MOGREPS-G Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 10 Oct, 05:00 AEST

Model-grid 2 m temperature spread across 18 MOGREPS-G members Model-grid 2 m temperature at the MOGREPS-G ≈20 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-09T18:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 16.4 to 17.6 degrees Celsius; by the end they give 12.3 to 15.5, with a 10th-to-90th spread of 11.2 to 16.5.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 7Thu 8Fri 9Sat 100102030 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-0710-0810-0910-10
0% 100% 100% 100% 89% 28% 78% 22% 6% 33% 6%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Sat, 10 Oct, 05:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • Global approximately 20 km grid; point values are at the native model grid point, not resort-elevation forecasts.
  • MOGREPS-G has 18 members (control + 17 perturbed). Open-Meteo currently returns about a 10-day point horizon; every displayed valid-through is derived from the returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · MOGREPS-G — UK Met Office; weather data served by Open-Meteo.com

ICON-EPS

Mt Baw Baw · 40 members · 0.25° ≈26 km · grid elev 748 m

Cumulative snowfall Mt Baw Baw ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 18.5 to 19.1 degrees Celsius; by the end they give 11.5 to 16.2, with a 10th-to-90th spread of 9.2 to 16.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 70102030 °C

Cumulative snowfall Mt Baw Baw ICON-EPS Tue 29 1200Z snowfall coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Middle half 0–0 cm · 10th–90th 0–0 cm
Cumulative provider snowfall across 40 ICON-EPS members Accumulated from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep. By that time, the middle half of members give 0 to 0 cm, with a full spread of 0 to 0 cm. The chart is drawn on the time scale shared with the other system, so it ends short of the full width where this system stops reporting. The percentile bands are the view; a control below the chart can additionally draw all 40 individual member accumulations as faint unranked lines, none of which is a forecast. Thu 1 Fri 2 Sat 3 Sun 4 Mon 5 Tue 6 Wed 7 0 0.5 1 1.5 cm
0% reach 5 cm
0% reach 15 cm
0% reach 30 cm

2 m temperature ICON-EPS Tue 29 1200Z temperature coverage · 30 Sept, 10:00 → 7 Oct, 11:00 AEST

Model-grid 2 m temperature spread across 40 ICON-EPS members Model-grid 2 m temperature at the ICON-EPS 0.25° ≈26 km grid point — not resort-elevation temperature, not wet-bulb temperature, and not a snow level. Drawn from 2026-09-30T00:00:00.000Z through 2026-10-07T00:00:00.000Z, using at least 3 members at each timestep, and stopping where member data stops. At the start of that interval the middle half of members give 18.5 to 19.1 degrees Celsius; by the end they give 11.5 to 16.2, with a 10th-to-90th spread of 9.2 to 16.9.Thu 1Fri 2Sat 3Sun 4Mon 5Tue 6Wed 70102030 °C

Daily chance of ≥1 mm liquid own horizon

09-3010-0110-0210-0310-0410-0510-0610-07
5% 100% 88% 75% 43% 25% 45% 13%
Coverage and model limits

Actual member-data coverage. Snowfall Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; temperature Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; liquid precipitation Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT; snowfall water equivalent Wed, 30 Sept, 10:00 AEST to Wed, 7 Oct, 11:00 AEDT. Percentiles and probabilities use at least 3 members per timestep; daily precipitation chances use only calendar days with at least 3 members reporting.

  • DWD ICON-EPS is served here on Open-Meteo’s approximately 0.25° (about 26 km) global point grid; values are at the native selected grid point, not resort-scale terrain forecasts.
  • A live Open-Meteo point probe returned 40 members (bare control + member01–member39) and about 180 hours of usable data; every displayed valid-through is derived from returned member data.
  • Provider snowfall is at the native model grid point. It is not snowfall at 1500 m, resort elevation or a physical snowpack depth.
  • No member-level pressure profiles are used or implied, so this cannot produce elevation-specific precipitation-phase probabilities.
  • The 180-hour ICON-EPS product served here publishes off 00Z and 12Z, so a run more than six hours old is normal, not a missed cycle.

Model run Tue 29 2200hrs AEST · 1200Z · 13 h ago · fetched 9 min ago

Calculation method

Each ensemble system runs many slightly different versions of its own forecast to sample how sensitive its outcome is to what we do not know about today. Each band above is a percentile range across those members: the 10th–90th is where four in five members land, the 25th–75th where half do. The probabilities are simply the fraction of members exceeding each total. None of it is a prediction of how much snow will fall — it is a statement about how much the models disagree, which at this range is the more useful thing to know. The snowfall being accumulated is each system's OWN snowfall field, taken from the provider as centimetres of snow and summed per member. It is not this library's wet-bulb phase logic, which is used on the shorter-range pages to decide where rain becomes snow and is applied nowhere on this page. What that means in practice is that the depth-to-water ratio here is the provider's fixed one, not a density that responds to the temperature of the storm, so these totals describe how much the members disagree rather than how deep the snow will lie. Every snowfall figure is accumulated over the comparison window chosen at the top of the page. In the common window all systems are truncated to the shortest usable valid-through and every percentile and probability is recomputed from the member series over that shared period — it is not a full-horizon number with its tail hidden. The temperature plume below each snowfall chart is built the same way from the same members, but it is the model-grid 2 m temperature: the value at the grid cell's own terrain height, not at a resort elevation, and not a wet-bulb temperature or snow level. It follows the same comparison window as the snowfall chart above it, because truncating a per-timestep plume simply stops it drawing and changes no value in it. The daily liquid-precipitation chances do NOT follow the window and always run to the system's own horizon: they are whole-calendar-day probabilities, and cutting one at 10:00 on its own last day would change what the number means. The snowfall chart can also draw every member's own accumulation as faint lines. They are all drawn identically on purpose — none is ranked, named or preferred, and none is the forecast. They are there only to show that the bands come from genuinely separated members. Those lines are plotted every six hours, matching each system's run cadence, while every total quoted on this page is accumulated over the full hourly series. Background

Weather data by Open-Meteo.com · DWD ICON-EPS; weather data served by Open-Meteo.com

NOAA AIGEFS

Mt Baw Baw · 31 members · 0.25° grid · precipitation only

Daily liquid precipitation own horizon

09-2909-3010-0110-0210-0310-0410-0510-0610-0710-0810-09
Median 0.10.49.917.95.40.30.43.90.20.10.0
10th–90th 0.0–0.1 0.2–0.6 7.2–14.6 9.6–30.4 0.6–23.2 0.1–1.3 0.2–0.8 1.6–8.9 0.1–4.5 0.0–1.3 0.0–0.3
Chance ≥1 mm 0% 0% 100% 100% 84% 13% 6% 97% 35% 13% 0%

millimetres of liquid, per AEST calendar day

Coverage and model limits

Every figure above is computed from the 31 members themselves. The artifact also carries NOAA's own published mean and spread, and neither is read here: a mean is not the middle of a skewed distribution, and a standard deviation is not a percentile range. Days are AEST calendar days, and a day appears only where enough members reported it.

No snow on this panel, and none is coming. NOAA AIGEFS publishes no snowfall field — its surface stream carries only 10 m winds, 2 m temperature, MSLP and total precipitation, exactly as AI-GFS and GEFS do not. No member-level pressure profiles are retrieved for AIGEFS, so it cannot produce elevation-specific precipitation-phase probabilities. So this system contributes no cumulative-snow plume, no temperature plume and no chance-of-5/15/30 cm figure — the questions the panels beside it answer. It is shown for the one question it can answer well.

Retention. NOMADS keeps two days of this product against a once-daily capture, so a missed capture is a permanent hole rather than a delay. When no cycle has been captured this panel is absent rather than empty.