The place for technical discussions about GA and flying.
Forum rules: Technical discussions about GA only, please.
#1858510
As part of my mid-range planning, like many I use Windy.com (free version). In other words, I use BBC Weather in conjunction with Windy to look ahead 5 - 10 days to try and see which days might be flyable, which I then check my calendar with.

One of the things I struggle with as a relatively low-hours pilot is that I prefer to err away from winds near the upper end of limits, quite sensibly.

What I find difficult though is that I believe - in general - Windy.com seems to really over-egg the wind gust forecasts to the point where - if I take it at face value - I'd never fly.

What I notice is that the default model - from the ECMWF - appears to always be the most conservative (e.g. shows the highest wind gusts). If I switch to MeteoBlue, the forecast always drops.

Similarly, the GFS and ICON-EU model seem to be more polarising; it projects the windy days as windier, but the calmer days as calmer.

So, my questions are really:

What's the difference between models, in laymen's terms?
What models do others use, or do you just use the default (ECMWF)?
Is there a preferential model for aviation that I should use?

I'm aware I could be over-analysing this, but even at an intellectual level, interested in the differences between the models.
#1858512
I use GFS

I use it because Dr Simon Keeling whose magnificent Weatherspoons I have attended uses it and seems to think it is most appropriate.

That's all.

Rob P

PS: Stop using the free version and subscribe to support a great service. :naughty:
User avatar
By Paul_Sengupta
FLYER Club Member  FLYER Club Member
#1858519
The first thing you have to realise is that anything more than a few days ahead is just a guess, You have to have a range of things...e.g. a particular system could move a bit further north of south or east or west and that will make the winds completely different..."The wind will probably be somewhere from the north to north east, somewhere between 10 and 20mph." All aspects of weather can be thus affected so anything more than 2 or 3 days ahead, I'd just look at the general synopsis and base it on that.

On Friday evening I was in a friend's garden. We had a bit of a BBQ. I was watching the rainfall radar and predicted possible rain around 22:30. The rain never came. A blob went up the east of us and a blob went up the west, but we remained completely dry. Try forecasting that! :D

Sorry I can't answer your question about the different models, but I expect with the randomness of the weather, some will be more accurate on some days, and others on other days. It's pretty much tossing a coin as to which will work, though as you say, some will give the extremes more than the average as part of their forecast.
#1859205
With the usual caveats about the vagaries of longer range forecasting mentioned above, I think it is generally accepted that ECMWF is statistically more accurate for our part of the world at least. It's what I use and I don't often find it far off.
User avatar
By Genghis the Engineer
FLYER Club Member  FLYER Club Member
#1860550
Windy/ECMWF is my favoured for mid-term forecasting.

Met Office issued TAFs and BBC/Meteogroup forecasts for local areas. Meteogroup's local forecasts published by the BBC use a more complex local knowledge database than the Met Office.

GFS is not my favoured product, it is optimised for North America, and is not as accurate in Europe as Met Office in the short term, or ECMWF in the longer term. (Actually the Met Office led "Unified Model" is consistently the best in the world, with active buy-in from dozens of national met organisations around the world.

It is worth looking at multiple products. If they generally all agree, the conditions are conducive to reliable forecasting, if there is significant disagreement, I would tend to distrust all of them, for the same reason.

GFS' outputs however are free, and readily usable, and I like their availability of cloud layering predictions, which are used by Autorouter, my current favourite IFR planning tool, and are pretty good. So I'd not discount GFS by any means, it's just not my first port of call.

And yes, pay for Windy premium, it's still cheap, and you're supporting a really excellent product.

Whilst my degrees aren't in meteorology, I have been studying it and working with some of the world's best research meteorologists since 2008 - I hope I was paying some attention.

G
#1860993
Ventusky free as well is mostly accurate and to check that yo can run it backwards to compare what it said against how you really experienced it
Either way an excellent tool. You can buy if you like it enough, but like many aviators one tends to study all forecasts to reach one's own decisions
#1861714
If you want to get a feel for whether you can trust the forecast - have a look at multiple models - if they are all different, they are all likely to be unreliable. If they agree (even several days out), then the forecast is more reliable. Usually you can see how far out the models are good for this way.
#1861850
Meteoblue multimodel forecast gives an average of some 20 forecasts. You can easily see how the different forecasts compare and decide whether they are likely to be accurate, given any/all the differences.

https://www.meteoblue.com/en/weather/fo ... length=168

Personally, I use Met Office and Windy for short term forecasting but do do a check with Meteoblue. The morning of any flight, it's Met Forms F214 & F215 together with Meteoradar rainfall forecast.

In my professional career (MN) I used to advise my passengers that it was just a pity that the weather didn't read the forecasts.