How Salt Radar estimates salt risk and seasonal driving dates

Weather data and location

Current conditions, thirty past days, and up to sixteen forecast days come from Open-Meteo. Seasonal dates use up to five complete seasons of ERA5 historical reanalysis at the selected location. Model grids cannot resolve every road, hillside or neighborhood.

Northern seasons run July–June, southern seasons January–December. Complete seasons must have usable daily temperatures and precipitation data. At least three qualifying seasons are needed for a historical date; history is cached for up to 24 hours.

What each date means

Car storage

Seven days before the median first wintry event: snowfall of at least 0.1 cm, or at least 0.5 mm precipitation with a minimum temperature at or below freezing.

First snow

The median first day with at least 0.1 cm snowfall. Forecast snow can replace a future estimate; earlier snow this season is not ruled out by that forecast.

Clear roads

The median date after the final wintry event with fourteen nights above freezing and at least 10 mm of subsequent rain. This is not a measured salt-removal date.

Winter tires on

Seven days before the first seven-night spell at or below 7°C, or an earlier wintry event. Forecast cold can bring installation forward. The date is clamped to season start in persistently cold locations.

Winter tires off

Fourteen nights above 7°C after the season’s final cold or wintry day. Forecast cold at or after the historical removal date makes that date unavailable.

Uncertainty and community reports

Ranges show the earliest and latest dates in qualifying historical seasons. Confidence labels reflect sample counts and whether a near-term forecast is used; they are not calibrated probabilities. Regional road treatment, mountain conditions, tire condition and local regulations require separate checks.

Salt risk also considers recent nearby reports. The map shows up to 50 observations from the past seven days; the risk model weighs reports from the last 48 hours within 10 km. Reports are public observations and are not independently verified. If the database is unavailable, the risk assessment uses weather only and says so.