FORECASTFLAKES
NWS National Blend of Models · Snow Situation Room
LIVE · NWS NATIONAL BLEND OF MODELS

The Snow Situation Room

Every National Blend cycle, read for plows, slopes and passes — and scored every morning against the NWS snowfall analysis

—
NBM cycle
—

The map

Every quantity the blend publishes for winter, and four it does not
How to read this room
  1. Snowfall. The blend's own six-hour snowfall summed to each horizon, six hours to ten days, and by climate day — 12Z to 12Z, the day the NWS analysis scores — so every day drawn here is a day that will be verified.
  2. The range and the chances. The blend publishes seven percentiles on each six-hour snowfall through f054. The 24- and 48-hour fans are those percentiles stacked and the chances are read off the stack; both are the widest range any dependence between windows could produce, and the page says so beside each map.
  3. Ice. Freezing-rain accretion the same way — totals, the 90th percentile, the chance of a tenth (roads) and a quarter (power lines) — plus the blend's own freezing-rain probability at any instant in 48 hours.
  4. Density and load. The snow-to-liquid ratio weighted by the snow that falls at each ratio, and from it the weight of snow on the ground in pounds per square foot. This is the slop factor made into a number: six inches at 5:1 weighs as much as two feet at 20:1.
  5. Snow level at each instant, masked to where precipitation is falling; trend against the runs six and 24 hours earlier for the same valid window; and verification — the most recent NOHRSC day against what the blend said for it one, two and three days out.

Every map is a georeferenced PNG drawn through a fixed palette that ships in the catalog, so the pointer readout inverts the exact bytes that were drawn. Hover anywhere for the band; turn on City values for the numbers over the towns. The boards below the map are the same numbers at 1,500 named places — cities, ski areas and mountain passes.

—

The boards

1,500 places, ranked — click a row for its six-hour series

Pick a place from the board, or search:

The ledger

What the blend said, and what fell, every day of the season

Nothing scored yet this season. Each morning after NOHRSC publishes its 24-hour analysis, the engine scores it against the forecasts it stored one, two and three days earlier — on the analysis's own grid, cell for cell — and adds a row here.

Where the numbers come from. The NOAA National Blend of Models, core CONUS grid, byte-ranged from AWS Open Data within the hour of each 00, 06, 12 and 18Z cycle. Snowfall and ice accretion are the blend's own six-hour accumulations, summed. The percentile fans are the blend's six-hour percentiles stacked — exact only if the windows are perfectly rank-correlated, and therefore the widest range any dependence could produce — and every chance map is read off that stack, not published. The snow-to-liquid ratio is the blend's, weighted by the snow falling at each ratio; the plow load is that snow's liquid equivalent at 5.2 pounds per square foot per inch of water. Verification is the NWS National Snowfall Analysis, 24 hours ending 12Z, on its own 2.5 km grid; the blend's grid shares that projection and spacing, so the forecast is moved onto it by whole cells and differenced without resampling.