@sasipm I don't think it is a very good or at least trivial idea to built a forecast on a mix of all models. Some are very good for short term, others for long term prediction, some are more precise and so on.
A simple mix of all of them would be a lie as it would just predict: "the weather will change somehow"
Of course it might be possible to do it right, but that is a really complex task:
- define the weight of each model on a time line
- only mix those that are regionally available
- add long historical data sets of similar situations
- post processing to remove forecast errors
- extract parameters that only exist in some models (e.g. thunderstorms)
- .... (many more I guess) ....
The german DWD is calculating what they call ModelMIX / MOSMIX and they use it e.g. for their weather warnings. The result -imho - is a lot of warnings that are not really relevant or being a lot less strong as predicted.
In my opinion the preselected ECMWF is a very good model for the average user.
I take a look at the Compare View of the models to check the reliability of the forecasts: If all models predict similar weather, than it seems to be a stable situation and you can count on that even for a few more days. On the other hand, if all models are drifting apart in the near future, it seems to be clear that we have a difficult situation and we have to be careful to rely on any prediction.
