Weather info for outdoor people

At noon every day, our system stores what each of the 7 global weather models predicted for 100 locations across the UK and Ireland — mountains, cities, coasts, and islands. The next day, we compare those predictions against what actually happened using observed weather data.

Each model gets a composite score out of 100 based on temperature accuracy, wind speed accuracy, rain prediction (both detection and amount), and cloud cover accuracy. Daytime hours (6am–8pm) are weighted 1.5× more heavily, since that’s when people are outdoors. Scores are averaged over a rolling 30-day window.

Where does the “actual weather” come from? We use the Open-Meteo Archive API, which draws on ECMWF’s ERA5 reanalysis — the gold standard for weather verification. Reanalysis combines real weather station observations, satellite data, and radar into a complete hourly picture at every location. It’s the same benchmark method used by national meteorological services worldwide to evaluate their own models.

The HILLANDGLEN model is our own accuracy-weighted ensemble — it automatically gives more weight to whichever models are performing best right now. As far as we know, no other weather site publicly scores and ranks weather models like this.

This data feeds directly into your forecast. The 5-day forecast you see on hillandglen.com isn't from a single weather model — it's the HILLANDGLEN blend, with each model's contribution weighted by how accurate it has actually been over the past 30 days. When a model starts performing poorly, its influence is automatically reduced. When one improves, it earns more weight. The forecast learns and adapts every day.