STAVROCAST

All models are wrong; Stavrocast is useful. Its only purpose is to compare equally probable weather forecasts from the world’s most trusted medium range ensemble forecast, the European Centre for Medium-Range Weather Forecast (ECMWF-ENS). Simply put - when the “ensemble members” agree, we can feel relatively confident in their collective forecast. And when they diverge, we can't.

I use Stavrocast to plan outdoor activities (field work, construction, cycling, etc.), and I built Stavrocast because it shows uncertainty in weather forecasts more clearly than popular forecast tools. For example, popular weather forecasts often contain statements like “80% chance of 0.6” of rain on Tuesday.” But… what does this even mean!? Is it going to drizzle all day or will it be a passing storm? Or do we not know? What's the likelihood of 0” of rain? Alternatively, what's the likelihood of a severe washout? Stavrocast aims to quickly answer these questions in an intuitive graphical form.

Forecasts are available for any location on earth, and they update every day at noon and midnight GMT.

Like all models, the ECMWF-ENS forecasts shown here warrant endless caveats. Of most relevance to Stavrocast - the ECMWF-ENS is a coarse, global model that essentially averages forecasts over ~9km grid cells. The result? In times/places of variability, highs may be higher and lows lower than predicted by Stavrocast. This is especially true when you see ensemble members diverging, for spatially discrete storms like summertime thunder storms, and in areas of highly variable terrain. Yet it remains useful.

Many thanks to Open-Meteo for making this data available via their API.