I’ve spent the last few months in Sweden. The local official forecast entity here is the Swedish Meteorological and Hydrological Institute (SMHI). They provide local forecasts with a dash of uncertainty thrown in. So, in the picture below, you can see that today’s forecast is “fairly certain” while Friday’s forecast is “uncertain”. That’s pretty cool, although even to me, it’s not entirely clear what those assessments mean despite looking around SMHI’s website for definitions.
What I think it even more fun, is a product they have rolled out in a beta form. Below, you can see that they provide different possible forecasts. So, rather than just providing the most likely forecast (what we might traditionally think of as the best forecast), they provide a range of reasonable forecasts. It’s probably going to be partly cloudy today, but hey, it could be mostly cloudy, or possibly cloudy. Regardless, it’ll probably be ~7C. Tomorrow is more interesting. It will most likely rain, but the second most likely forecast is sunny. Without this kind of categorical forecast, we might be tempted to split the difference and provide some kind of intermediate precipitation probability and cloud cover. But doing so risks misrepresenting the forecast. Maybe the possible futures are 1) 100% cloud cover and extensive rain or 2) clear skies. Providing a forecast that weights by the probability of those outcomes might be A) 75% cloud cover and intermittent rain. But that’s not really a likely future because it’s not 1) or 2) so it’s wrong from the star