This page shows how well our Sunrise/Sunset forecast score has actually performed, measured after the fact. We believe transparency is important. You've chosen to spend your hard earned money with us, and you should be able to see if it is worth it. If the score looks worse the further out it's made, that's expected, and it's shown here on purpose, not hidden.
Method, in one line: every forecast is verified against independent NOAA satellite (GOES) and surface-analysis (URMA) data, never against our own model's self-reported confidence. How this works →
of the time, our forecast rating matched exactly what actually happened.
of the time, it was within one tier of what actually happened.
Read each row as: "of the forecasts we rated this tier, here's what actually happened." A full bar reaching 100% would look reassuring, and would also be a sign we're overstating our own precision.
The score is necessarily less certain 72 hours out than 2 hours out. This is the shape you'd expect from any honest forecast, and it's the main thing this page exists to show.