How accurate is the model?

A probability isn't a claim that something will happen — it's a statement about likelihood. A model that says "70% chance of above-normal temperature" should be right about 70% of the time it makes that prediction. Verifying this is the only honest way to know whether KLIMAT-P's numbers mean anything.

The standard metric for probabilistic forecast accuracy is the Brier Score: for each prediction, it computes (predicted probability − actual outcome)². A perfect score is 0.0. A completely uninformative model that always predicts 50% scores 0.25. Lower is better.

This page updates automatically every Monday. The system looks back at the previous week's IMGW station data, determines what actually happened for each indicator, compares it to what the model predicted, and records the result. No manual entry required.