How the Snow Day Calculator Works
Snow Day Estimator turns the National Weather Service forecast into a chance of a snow day by asking one question: is the snow coming tonight a lot for this town, or normal? Here is exactly how.
1. The places we track
We track 3,232 towns and cities: every place of about 10,000 people or more where snow is part of winter, the larger Southern metros where a little snow or ice can shut things down, and smaller towns that fill gaps in rural snow country. When you enter a ZIP code, we use the nearest of those towns and say so on the result. Your ZIP code never leaves your browser; the lookup table is a static file.
2. The forecast
For each town, we read the National Weather Service gridded forecast from api.weather.gov: forecast snowfall, ice accumulation and, in the upper Midwest, wind chill, hour by hour. A full update of every town runs every afternoon, and towns with snow or ice in the next three days, or inside a winter weather alert, update every one to two hours through the evening and early morning, when schools make their decisions. We also read every active National Weather Service winter alert once an hour.
3. The snow that matters
Districts mostly care about the roads at bus time. So for each school day we add up, in the town's own time zone:
- Overnight snow, 8 pm the night before to 9 am, counted in full.
- Snow the day before, 9 am to 8 pm, counted at 60%, because crews have time to clear some of it.
- Snow during school hours, 9 am to 3 pm, counted at 35%. It can prompt a closure if it's heavy, but it more often means early dismissal.
- Ice, converted at 25 inches of "snow effect" per inch of ice, so a tenth of an inch of freezing rain counts like 2.5" of snow.
We skip weekends, federal holidays and the day after Thanksgiving, so on a Friday you see Monday's odds instead of two empty days. From about December 19 to January 2 we show a winter-break reminder instead of guessing each district's calendar.
4. Each town's closing point
This is the part most snow day calculators skip. The same 4 inches that barely slows Buffalo can close every school in Atlanta. We estimate each town's closing point from its NOAA 1991–2020 normal annual snowfall (an inverse-distance average of up to three nearby NOAA stations), using a smooth curve:
| Normal snowfall | Closing point |
|---|---|
| 0" a year | 1" |
| 2" a year | 1" |
| 5" a year | 1.5" |
| 10" a year | 2" |
| 20" a year | 3" |
| 35" a year | 5" |
| 50" a year | 6" |
| 80" a year | 8" |
| 100" a year | 8" |
| 150" a year | 9" |
Formula: closing point = 1 + 11 × N1.3 ÷ (N1.3 + 601.3) inches, where N is normal annual snowfall in inches.
5. From snow to a percentage
We compare the effective snow amount with the town's closing point and run it through a logistic curve: at the closing point the chance is about 50%; at half of it, about 8%; at one and a half times it, about 90%. With no snow or ice in the forecast, the chance is under 1%. Then:
- An active Winter Storm Warning or Blizzard Warning during the decision window (3 pm the day before to 9 am) raises the odds; an Ice Storm Warning raises them a lot. Advisories and watches add a little.
- A tenth of an inch of ice or more raises the odds further, and a quarter inch more again.
- In Minnesota, Wisconsin, Iowa, the Dakotas, Michigan, Illinois, Nebraska and Montana, a separate cold check looks at the lowest wind chill from 5 to 9 am. Around −25°F the chance of a closure from cold alone is roughly 12%; around −35°F it's about 70%.
The delay chance uses the same inputs with a lower bar (40% of the closing point), minus the chance of a full closure.
6. Closure reports
Under every result you can tap Closed, Delayed, Open or E-learning. We count one report per device connection per day, store only the town, the date and your answer, and show the totals for your area ("58% of 214 reports near Rochester say closed"). No names, emails or accounts. See our privacy policy.
Accuracy record
This is the calculator's first season, so there is no accuracy record yet, and we won't claim one until there is. After each storm we'll compare our estimates with the closure reports people send in and publish the results here: how often a 70% turned into a closure, and where the closing points need adjusting.
Storms scored so far: 0. The first results will appear here after the first major storm of the 2026–27 season.
Limits
- Snow forecasts can be off by several inches, especially near a rain–snow line or under narrow lake-effect bands.
- Every district decides differently. Buses, hills, staffing, and how many snow days are left all matter, and we can't see those.
- We don't know your district's calendar, so check for teacher days and breaks.
- This is an estimate, not an official announcement. Your school district makes the call.