Models need help
-
The raw data from the GFS model really is a waste of time without the error correcting measures that the NWS has developed over decades. A much more honest weather forecast would be GFS corrected in the near term with NWS adjustments. While the ECMWF model is pretty good near term, it also suffers from a wide standard deviation further out, even to the point of having wild temperature forecast swings intraday 10 to 15 days out. The point here is that the models are so deeply flawed beyond a 7 to 10 day horizon that they are mostly worthless, yet this extended horizon is what you charge for! Windy seems great for monitoring hurricanes and getting accurate rainfall forecasts, but beyond that, at least in the US, the NWS is a good as anything out there. Signed, "not a meteorologist"
-
Weather prediction can be hard. 1 small change in the 1st day can lead to major changes 10-15 days out. Forecasts past a week are unreliable but are useful to see what might pan out. To metrologist (Not Me), seeing the general picture is useful.
Even through ECMWF may show a colder period out 12-15 days, this shows that it might be cold but you can't take anything beyond 7 days. Anything beyond 7 days usually shows the general picture, to quite a few people (like me), I read it and "Okay, we might have high pressure next week so it will be mild".
In the UK, around mid-december I saw a colder period around New Year incoming. All I could see was that the temperatures were around 3-5c lower than usual. Not until around 3 days out I started looking at what might happen during this colder period. ECMWF beyond 7 days is good at see what might pan out but take anything it says with a pinch of salt beyond 7 days. Look at the forecast over 3-4 days and if it is saying the same thing then you can assume that but if it changing, there is usually something going on in the atmosphere that is leading to the "jumpy" models
(Experts - Feel free to jump in and correct me)
I hope this shows why forecasts jump a lot
If you have any follow up question, just reply :)
-
You are absolutely right. At a given location, a 15-day forecast, regardless of the model, has no value at such a time frame. Beyond 5 to 7 days, a local weather forecast will have less and less chance of coming true.
So why does Windy present these detailed 15-day forecasts? When these forecasts on Windy were only 10 days, many users asked for them to be extended to 15 days !
On the other hand, as you said, a 15-day forecast on a large scale (synoptic) allows you to get an idea of the trend for the next 2 weeks.
