Windy ECMWF meteograms are wrong, shows rain when well below freezing
Rain and convective rain mixed with snow, at -8C in places where it is clearly snowing. It is a regular error that I have seen on your forecasts for locations around the world.
It grid problem, because GFS with a 22km grid does not show rain at all.
Either ECMWF data is unreliable/wrong, or your computers are wrong.
This is a serious forecasting error.
Windy is not responsible for forecasting errors.. it visualize forecasting of the important weather models..not more.. not less
johnckealy last edited by
Just a quick comment on the models themselves (I'll have to be general as I don't know where these meteograms are for). Things aren't always so clear cut in terms of " < 0C temperatures equals snow!". Snow is a tricky beast to forecast. But liquid water while the 2m air temperature is -8C is perfectly feasible, albeit unlikely. Here are some things to consider:
- Rain accumulations may be due to snowmelt: Many surfaces in early winter hold warmer temperatures deep in the ground which can melt snow in direct contact with it. In the meteograms above, my money is on this (though I don't know for sure whether this is part of the ECMWF surface parametrization or not). Interestingly, in this situation the road surface temperature is always exactly 0C.
- Temperature inversions: The temperature can actually increase with height in some situations, allowing for liquid water and melting to occur aloft. This is the mechanism that can lead to freezing rain.
- Supercooled water: Supercooled water is water that exists in liquid form below 0C. In absence of aerosols this can technically exist down to -40C. Supercooled water is common in convective clouds this, in combination with a process called riming, is how hail forms.
It is actually quite likely that ECMWF are showing more sophistication in their cloud microphysics parametrizations than GFS. Having said that, all NWP models still have a long way to go with snow forecasting :(
John - I understand that reasonably well. But trust me, the forecasts are wrong. Just plain wrong. I work every day in the mountains where Windy says it will be raining, it is absolutely not raining, not even close. Perhaps the EC model is trying to be "too clever". But they are chronically wrong either way.
Tomber - Windy may not be responsible in a technical sense, but they publish a website with errors in the forecasts. So in branding and public perception, it is Windy who is wrong, not the model that windy is porting.
I like Windy, but I hit so many accuracy hurdles over the years. It is hard to commit to them. Speaking mainly in terms of Windy as a mountain snow forecasting tool.
Global models used by Windy are not at all accurate or useful to forecast the Rain/snow limit in mountains. If you work in mountains, you must know that conditions change totally with altitude.
Even with the ECMWF model, a resolution of 9km is too low to get a precise forecast of temperature and rain/snow limit in mountains. In a surface 9 x 9km the model provides only one figure, that doesn’t take into account the terrain and local altitude variations. And worst, the resolution of the orographic model included in the weather model is really bad too.
You must know that because you have already open several topics about the same subject and got a lot of explanation. In this one for example: https://community.windy.com/topic/4513/what-elevation-for-spot-forecasts
I suggest to consider the forecast in the Mont-Blanc area and to look at the Sounding forecasts which give the model altitude.
In the valley at 1050m, the model altitude is 2140m ! More or less the same as on top of Mont-Blanc, 2245m for 4808m !
I don’t know where you are living, but it’s probably the same. (Except if the model altitude corresponds to the ground altitude, by chance, where you live)
To have a better idea of the Rain/snow limit, you may consider the 0°C isotherm which is shown in the Airgram.
Then, I don’t know why ECMWF shows sometime a part of snow precipitation in rain, even with low temperature. But it’s generally a very small part (less than 1mm).
Gkikas LGPZ Moderator last edited by