How to read the Winter Storm Lab

Everything here is guidance, not a limit. The maps are our own diagnostics of each model run: RAP and RRFS on a common 0.3° grid (about 33 km), HRDPS on its own grid at about 5 km (every second 2.5 km point, computed there, not interpolated). The airport panel uses the model point nearest the airport. The Ensemble tab shows the REPS and GEFS members at the airport: raw ensemble, not calibrated. None of it feeds our TAFs or airport forecasts.

Model picks the run. RAP is the only model whose files carry vertical motion at every level, so the lift-in-the-DGZ view and the omega weighting of the physically based ratio are RAP only. RRFS and HRDPS weight by cloud depth instead. HRDPS columns are read every 6 hours (2.5 km files are large); its snowfall still adds up every hour.

Snow ratio is shown only where the column would give snow (snow, rain/snow or snow/pellets types). Snowfall adds up the model's own hourly snow water times each ratio, from the start of the run. Snowfall is new snow, not snow depth: settling, melting and drifting are not taken off.

Precip type uses the Bourgouin energy method on the model temperature profile, strong colours where the model has precipitation in the hour and pale where it has none (the type the column would give). Dark lines mark transition zones between types.

Spread compares our deterministic models and their previous runs over the same window. It is model spread, not a calibrated probability. The real ensembles (REPS / GEFS) are in the airport panel's Ensemble tab.

Heights are feet above ground. Times are UTC. Drag to pan, use the wheel or pinch to zoom, click the map to pick the nearest airport.