Live ACCESS-G maps

What is—and is not—wired in

These maps restore current ACCESS-G guidance inside the hobby app without relabelling another model. They are genuine regional model products, not point samples. The app does not read a colour from the image and pretend it is a resort temperature or precipitation amount.

The Bureau publishes no chart cut for the alps, and none is invented here: nothing is stitched from two charts or resampled onto a new grid. The default view is the Bureau’s own Victoria-area product, which runs to 151°E and 40°S and so carries the Snowy Mountains 2.6° inside its eastern edge, drawn at roughly 51 pixels per degree against the NSW chart’s 35. One Bureau chart, both ranges, at the finer of the two scales. The NSW chart is kept for what it is better at — twenty degrees of approach for a system still out over the Tasman.

The second chart is free: any region against any field, so the pair can be one field at two zooms or two fields over one region. Both are driven by one clock, and the pairing is by valid time rather than by frame number — the rainfall charts start one step later than the 850 hPa ones, so a shared index would silently compare two different moments. A chart with nothing at the selected time says so instead of holding its last frame.

ACCESS-G runs four times a day, but not all four to the same length: the 06Z and 18Z cycles are short, reaching three days, and the 00Z and 12Z cycles carry on to seven. This page takes the Bureau’s combined timeline, which uses the short cycle for as far as it goes — three-hourly, and up to six hours newer than the long run — and the long cycle beyond it. That is two model runs on one clock, so it is stated rather than smoothed over: the stamp names the cycle each frame was produced by, read from the Bureau’s own image URL, and a frame that does not carry one is left unlabelled. Where the two overlap the fresher, finer-stepped run is the one shown; neither is blended into the other, and no frame is interpolated between steps.

Times are labelled in AEST, the zone the season is planned in, with the UTC stamp printed on the chart itself kept in parentheses beside it. The Bureau draws the UTC stamp into the image; that pixel text is untouched, and the caption is a conversion of it, not a replacement.

Each frame is a copy, taken when this page was built and stored under the cycle it belongs to, rather than a link followed when you open the page. The Bureau serves every chart from one rolling address whose contents are replaced as each new cycle publishes, so a link followed later would return a newer picture beneath a caption written for the older one — and the caption is a conversion of the stamp burned into the image, so the two would visibly disagree. Copying is what keeps a frame and its caption the same run. Nothing is resized, recompressed or redrawn; a frame that could not be copied is left out rather than shown under a label that might not describe it.

The dots are Perisher, Thredbo, Falls Creek, Mt Hotham, Mt Buller and Mt Baw Baw, placed from the map’s own latitude and longitude graticule. They are locators, not readings: nothing is sampled at a dot, and resort numbers come from the vertical-structure pages, which do pressure-aware wet-bulb work these charts cannot. If the Bureau reissues a product at a different image size the dots are withheld rather than placed on a guess.

The existing point comparison continues to probe Open-Meteo’s ACCESS-G domain on every build. When that machine-readable feed resumes, its resort series can return without changing this map route. Until then, ACCESS-G cannot supply this app’s pressure-level wet-bulb snow-level calculation.

850 hPa is a pressure surface, not a fixed resort elevation; its height varies with the atmosphere. Read temperature and humidity together for broad airmass structure, not as a thermometer at a lift.