Another great site is Wetterzentrale.de If you look at the GFS models for CAPE+Lifted Index and the Niederschlag charts (it’s too much hassle to post the actual pictures on this forum) you will see a map view of the very low LI and the lack of precipitation expected for your route/time….the other chart to look at is the Mittl. wolken chart which shows cloud density and in your case shadows that there is some cloud (low density though) between 1000 and 500 hPa…the Autorouter Gramet is brilliant, but this site also gives you the map view…
1200 MSLP:
This shows a cold front over southern Italy.
The troughs should translate to a standard PROB30 TEMPO TSRA. However the 0600 MSLP doesn’t show the troughs, as one would expect, because morning wx is normally less convective
So I would do the flight first thing in the morning. This time of the year, I always fly first thing. It makes life vastly easier. For example much of this stuff (current lightning)
wasn’t there 0600 this morning.
The forecasts don’t agree with any of that
TAF
LGKR 291100Z 2912/3012 29008KT 9999 FEW018 TEMPO 2920/3002 VRB03KT BECMG 3003/3005 15008KT
LIBR 291100Z 2912/3012 09015KT 9999 SCT020 BECMG 2918/2920 16010KT CAVOK BECMG 3004/3007 16015KT 9999 SCT020
and I would suggest that is because they think the convective stuff is not going to get going before 1200Z. So, even if they are spot on, you may be cutting it close. Anyway, this underlines flying early in the day.
On the morning of the flight, and it is a very short flight
get the sferics and get the IR image and tafs and metars and that’s it.
I think if you leave it till 1200Z you may find a lot of stuff building up. In that case, stay in VMC all the way, obviously. I don’t know if you have oxygen but you won’t be able to outclimb any proper stuff.
The cold front is probably not significant because there is almost no wind across it. This is confirmed by the TAFs.
Thanks for all of this. Lot of good tips and analysis!
Here’s a radar image for Italy I came across – not sure of its accuracy
I’m actually flying out to Corfu from Brindisi early morning to refuel and meet up with a friend and the GRAMET I posted is for the return flight after lunch. So I’ll have to make a call in the morning whether to chance it.
Make sure to get updated GRAMETs. GFS data is calculated every 6h but needs a few hours to propagate. On https://router.euroga.org/gramet you can see the timestamp of the current GFS data and it’s also printed on the charts at the bottom.
There is also the original Spanish GRAMET
I’ve used that for ages, referring to it in my writeups.
It is not bad but like all these GFS plots it usually fails to show all but higher altitude and nasty convective cloud. And GFS is the only weather model which is accessible without hefty payments.
And before that we have Meteoblue (much used by me, and also in my cloud tops writeups here). That site is now dead, I believe. It did great graphics but…….
So there is no free lunch, a forecast is a forecast, perfection is not achieved (except in the form of your GF/wife and then only if she reads EuroGA) which is why my preflight briefing has become much simpler over the years. Nowadays I look at the MSLP for a quick and dirty decision, and use the sferics, IR and radar images before the flight (or before cancelling).
And I cancel only on the basis of actual wx, never forecasts (well not if I actually want to go there). I might cancel on the basis of an MSLP if I have to decide between competing propositions…
Hi there,
I had the opportunity to use GRAMET a number of weeks ago as part of my ‘active consideration’ for doing a trip, as opposed to just playing with it. It predicted the route weather reasonably well. How are you guys finding the accuracy of the visual depiction vs. real life weather? Is there much more that can be done to develop with with current model data/technology?
I did note the GFS Update Times were older than I expected. For example, the current GFS Ref Time is 06:00Z just under 12 hours old.
DMEarc
Which implementation did you use? The one from Ogimet or the autorouter version? While they are very similar and based on the same GFS data, we tried to make it a bit more realistic, mostly by being less optimistic and regards to cloud formation.
The GFS data is calculated every 6 hours and it takes a bit more than 6 hours to make the data available so your 12 hour difference makes sense. We get it directly from the US NOAA server the minute it is released. Unfortunately the server is very slow and the download takes a while.
GFS is a global model and its resolution is rather coarse. I hope that we will eventually be able to offer GRAMETs using the more fine grained COSMO EU or its successor — for forecasts in the Euro region obviously. There are weather situations where the model works fine and there are situations where it’s not very useful. Match it with other data such as MSLPs, TAFs, sat images to get a better picture.
I used the autorouter version of course :)
Yes, I take your point about the datapoints being a little sparse on GFS. I do wonder if a hybrid GRAMET and TAF depiction would be a good compromise. I’m not certain how one would phase the terminal forecast into the higher altitude forecasts, but if the TAF predicts ‘SCT @ 800’, it appears sensible that this takes priority over GFS derived data.
In saying all of that, my question is open about whether the less optimistic Autorouter version is proving reliable in the field?
(fingers crossed it is – it would be an amazing tool)
Over the years, loads of people have tried to generate fancy graphics from GFS data – the only wx model which is freely accessible.
Here (written a few years ago) I look at e.g. Meteoblue which ran for a number of years, producing great graphics like this
which looked great but on the actual flight turned out to be largely fiction in so far as telling you something not blindingly obvious with a quick look at the MSLP (surface pressure and fronts) chart. Meteoblue was I believe a PhD project of a Univ of Basel student, and last I saw it he was turning it into a commercial site of some sort.
I think the biggest problem that has never been solved is forecasting the vertical extent of cloud away from fronts. When you see a nice warm front, you can reckon on low bases, tops ~FL200-250, drizzle, well the usual stuff. When you see a nice cold front, you can reckon on somewhat higher bases, more convective stuff, tops could be anything (CBs), but often quite broken up. If you see a load of troughs, you can expect scattered buildups. And GFS will more or less tell you this – but you already knew roughly what to expect.
What GFS mostly fails in is forecasting the other cloud – the sort of non-frontal wx in which most light GA flight is done. It may be that national wx models are better at this but access tends to cost a few k a year minimum, with the cost linked to your intended re-publication volume.
TAFs have great value for cloudbases and vis because they are produced by forecasters with access to the national wx models and hopefully with some local knowledge. TAFs are also substantially cheated because a TAF is not published until the METAR for the said airport has come out 
A METAR (or any other observation) trumps any forecast of course, and a fresh TAF trumps any forecast too because it is hacked to match the METAR 
This autorouter Gramet from EGNV to EDDP the day before yesterday was spot on :)
