My experience is GRAMET work very well for approximate IFR work, layers are roughly correct, winds and temps same. Also fronts appear pretty well. Very low level cloud or fog is poorly represented but this is, I believe, a limitation of the GFS model.
Airborne_Again wrote:
@achimha, do you have a reference?
My experience of GRAMET has been very good until this week, when it has been completely wrong every time I have used it, which is why I wonder whether it is broken in some way?
Today we made the same experience with the Gramet as @Timothy. Instead of the forecasted layers without icing over Spain and the forecasted posibility to fly most of the time above clouds in flightlevel 110, we had moderate icing and were flying in the clouds much longer then expected. It has been a good idea to fill the TKS reservoirs completely before flying to Marokko.
To be honest, that was the mostley misleading weatherforecast I have ever had with the GRAMET. Normally it was nearly accurate.
Same for me, quite accurate give or take 1-2K feet. Icing so far has been more conservative than experienced in reality.
Let me put it as strongly as this: If @achima were to discover that the incoming GFS feed server was down, and was regurgitating the same model every day, it would not surprise me and would seem the most likely explanation for what I have seen.
These have been my flights:
21/4 Biggin local
22/4 Biggin-Bristol-Biggin
24/4 Biggin-Calais-Biggin
26/4 Biggin-Cherbourg-Biggin
27/4 Biggin-Lydd-Biggin
28/4 Southend-Abbeville-Southend
29/4 Biggin-Hangelar-Biggin
They have all shown clear blue skies and they have all had SCT-BKN clouds from about 3500-6000.
Please @achima do not think that I am complaining. You provide the most fabulous free service in so many ways, and you have my utmost gratitude and respect. I just think that something has broken and I am trying to help.
achimha wrote:
http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-318805
A master’s thesis from my own university even… Thanks.
Timothy wrote:
to discover that the incoming GFS feed server was down, and was regurgitating the same model every day, it would not surprise me and would seem the most likely explanation for what I have seen.
Each GRAMET prints the model(s) used for the forecast so stale data can be easily ruled out. Our systems also monitor the GFS status and we’d get alerts in case of problems. I believe the most likely explanation is that the model simply got it wrong. Happens from time to time.
One of the unfortunate things is the long delay from start of the model until the data is available and downloaded (the servers are terribly slow). Obviously the model takes input data (i.e. weather observations) and then calculates for some hours and then starts distributing the result datasets. From that start until we get it is about 12 hours at best.
That’s strange. In this part of Britain where we have more than our fair share of high ground and low cloud, I use the Autorouter Gramet to help predict and prepare for possible pop-up flight over hills in IMC. It has been pretty accurate up here for the last week or two.
Peter.
Short of posting them all (which I am happy to do, but would be a bit of a faf) I don’t know what to say.
