"Probably going to snow" is not a snow day prediction. That's a forecast. Snow day prediction is a completely different problem — and it's harder than it looks.
Weather models tell you how much snow will fall. Snow day prediction tells you whether that amount will actually close your school. Those are two entirely different questions.
The Two Data Sources Every Prediction Needs
1. Weather Data Modern snow day calculators pull from multiple meteorological sources to build a complete weather picture:
- •National Weather Service (NWS): Hourly forecast grids for the continental US
- •Open-Meteo: High-resolution European weather model data (ECMWF)
- •NOAA HRRR: High-Resolution Rapid Refresh — the gold standard for short-range winter storm forecasting, updated hourly
No single model is right all the time. Combining multiple sources reduces error and catches when models disagree (which is itself a signal of forecast uncertainty).
2. Regional Tolerance Data This is what separates a snow day *predictor* from a snow day *calculator*. Raw weather data tells you the storm. Regional tolerance data tells you how your district responds to that storm.
- •Key regional variables:
- •Historical closure thresholds for that geographic area
- •Available plow infrastructure (trucks, routes, staffing)
- •Elevation and road grade profiles
- •District's history of closures vs. delays
The Four Factors SnowSense™ Weighs
Factor 1: Snowfall Accumulation (Weight: ~30%) Not just total depth — but accumulation rate. **2 inches per hour** is operationally very different from **0.5 inches per hour**, even if they produce the same total.
Factor 2: Ice Risk (Weight: ~25%) Freezing rain, sleet, and black ice are more dangerous than snow. The model analyzes: - Surface temperature vs. dew point - Precipitation type probability - Freeze-thaw cycles overnight
Factor 3: Temperature / Wind Chill (Weight: ~20%) Extreme cold (wind chill below -20°F) can cancel school independently of snow. Bus engines fail. Students at stops face exposure risk.
Factor 4: Storm Timing (Weight: ~25%) **The most underappreciated factor.** Snow falling between 2 AM and 7 AM has maximum disruption impact. The same total accumulation falling between noon and 6 PM barely affects the next school day.
How Probability Is Calculated
Each factor produces a risk score from 0–100. These scores are weighted and combined:
Raw Score = (Snowfall × 0.30) + (Ice Risk × 0.25) + (Temp × 0.20) + (Timing × 0.25)
- •That raw score is then calibrated against the regional tolerance modifier:
- •Boston: raw score × 0.6 (requires more to close)
- •Atlanta: raw score × 1.5 (closes more easily)
The output is a probability percentage — not a yes/no, but a nuanced estimate of likelihood.
No prediction model is 100% accurate because the final decision is made by a human — the superintendent — who has access to information no algorithm does. They might know a specific bus route is on an unplowed road, or that their drivers union has been pushing back on early morning runs. The best a prediction model can do is give you the statistical likelihood based on conditions.
⚡ The Trench Truth
Why Predictions Update Every 30 Minutes
Winter storms move and change. A storm predicted to arrive at midnight might stall and hit at 4 AM instead — which dramatically changes the school closure probability. SnowSense™ re-fetches and recomputes predictions every 30 minutes so you're never working with stale data.
Check Your City's Current Prediction
Use our snow day calculator by location to see a real-time probability score for your city, including a breakdown of all four factors.