0–3 days — vertical structure

Inside three days the argument stops being “will it snow” and becomes “what shape is the column”. A snow level is one number; these pages show the profile it came out of — temperature and wet-bulb against height, humidity beside it, and four restrained readings of what that shape is: the lowest-layer lapse rate, any layer where temperature rises with height, how deep the model stays humid, and whether the supplied levels reach the dendritic growth range at all.

Pick a resort. Each page carries a model selector — ECMWF IFS, ECMWF AIFS, GFS and ICON — and six-hourly valid times out to 72 hours.

Interval timeline → How much falls in each three, six or twelve-hour window, phase-split at low, mid or top, across every model that publishes a track.

Compare resorts → Align two to four resort profiles at the same model and valid instant.

Live ACCESS-G maps → Current BoM rainfall, pressure, 850 hPa temperature and humidity charts.

Show NSW VIC

Snowy Mountains

Victorian Alps

Before you read one

These are model profiles, not soundings. Six to nine requested pressure levels, of which the lowest are usually underground at an alpine grid point and already discarded. No radiosonde is drawn on them, and neither are the surface observations on /now — one instrument at one height is not a column.
Nothing is drawn above the requested 500 hPa level. Where the dendritic range (-12 to -18 °C) sits above the highest genuine level supplied, the page says the supplied levels do not cover it — not that there is no dendritic zone.
Moisture depth is descriptive. Contiguous supplied levels at or above 80% relative humidity. It says where the model is humid. It is not cloud, and it is not snowfall.
None of it becomes an amount. No centimetres, no snow-quality score. Those would be numbers this data cannot support.