Share.

    4 Kommentare

    1. In this case you can, there is a cell that comes from south and will hit Luxembourg, South of Belgium and Saarland directly. Now it depends, which places will be hit harder or not, but I can guarantee you that at 9 p.m., you will become wet, if you are outside.

    2. Basically all commercial services use either the American climate model (gfs) or the European model (ecmwf) which are basically governmentally paid combinations of weather satellites and supercomputers for predictions of the cell movements. Then the commercial services use humans or computers to generate a prediction of the higher resolution localized prediction based on the models.

      The crux is that those major models are flawed in that their calculations base on a training set using the last decades of weather data to predict the future. Due to increased variability in the pattern, those predictions tend to either be completely wrong or mis-interpret the risks coming from a weather front as the energy sources/sinks and global patterns shift and change.

      As predictive ai is now „mainstream“ thanks to Chatgpt, the current weather situation is akin to giving chatgpt access to all texts from the renaissance period and then asking it to write a modern space action movie. It will write…… something. But it won’t be very useful most of the time.

      Human interpretation by a skilled person is still the best chance for good local predictions. Météo Boulaide does a terrific job but the language barrier might be difficult for some redditors.

      Ignore Météolux, they tend to somehow manage to get the predictions even more wrong than the source models. Although that means that if you invert the prediction it could be pretty accurate

    Leave A Reply