Hi. In summer 2025, I got a bit disappointed that many days ended up hotter than they were forecast, and decided to quantify this to see how "bad" is it. So I started recording.
I took forecasts for Bern and Zürich at the end of the day (~10PM), so that the forecast for the current day is not anymore "a forecast", and then also predictions of temperature and precipitation for up to 7 days.
A few weeks ago, I put a web app on top of it for easier navigation. You can check the history of forecast errors in these two/three (min/max temperature, precipitation) for Bern and Zürich in Switzerland. I may start recording other cities too.
I thought someone else may find this interesting. Any feedback, flaw, etc is more than welcome.
What I read from this is that their model lies within +/- 2 degrees most of the time. I assume therefore it's adjusted to reflect observations, and thats the error limit they set as a target.
I took forecasts for Bern and Zürich at the end of the day (~10PM), so that the forecast for the current day is not anymore "a forecast", and then also predictions of temperature and precipitation for up to 7 days.
A few weeks ago, I put a web app on top of it for easier navigation. You can check the history of forecast errors in these two/three (min/max temperature, precipitation) for Bern and Zürich in Switzerland. I may start recording other cities too.
I thought someone else may find this interesting. Any feedback, flaw, etc is more than welcome.
Best
What I read from this is that their model lies within +/- 2 degrees most of the time. I assume therefore it's adjusted to reflect observations, and thats the error limit they set as a target.
I don't think this is "bad" per se.