I don’t know whether the GFS weather model is completely broken for what is called low level cloud, or whether the Spanish Gramet site is broken, but that site never depicted low level cloud (basically the stuff you see on a nice day, base say 2000-3000ft, maybe 2000ft thick).
That is not “low level cloud” in the GFS model, but “boundary layer cloud”. Do you plot that, Achim?
I have to say I like Ogimet’s icing model a bit better. I don’t want the cloud layers below freezing highlighted. I can see that for myself. If we could do something like the NWS does for icing potential, that would be more useful.
(And in case I forget, nice job!)
That’s an interesting paper, bookworm.
I was not aware that anybody was able to usefully forecast icing conditions – despite many claims.
Where is the website which shows this information? I wonder if it works for Europe, because it mentions satellite image derived cloud top temperatures.
I have to say I like Ogimet’s icing model a bit better.
Huh? We use the same icing model as ogimet (corrected for a missing minus sign).
I wonder if it works for Europe
And it uses Pireps… how many Pireps do you get during a heavy icing day throughout europe? 2 maybe?
Huh? We use the same icing model as ogimet (corrected for a missing minus sign).
Then I’ve misinterpreted what you’ve written in you “GRAMET interpretation guide”. All you write is:
It is very difficult to predict when icing occurs. Most meteorologists say that icing is possible between 0°C and -20°C, below that temperature clouds usually do not contain supercooled water droplets that turn into ice upon impact. Based on the forecast moisture and temperature, we predict icing and mark it with a green shaded area.
Why not make the criterion for icing to be show more explicit?
It is a very simplistic model, based on temperature (with a peak at -7 degC) and cloud amount. But it’s a reasonable guide.
Why not make the criterion for icing to be show more explicit?
It is a very simplistic model, based on temperature (with a peak at -7 degC) and cloud amount. But it’s a reasonable guide.
I’ve updated the guide and added the formulae for icing and turbulence with a bit of explanation. As Tom mentioned, we use Ogimet’s model as a starting point and plan to refine it in the future based on experience. It’s not what I would call a simplistic model but obviously entirely based on atmospheric conditions as predicted by GFS.
That is not “low level cloud” in the GFS model, but “boundary layer cloud”. Do you plot that, Achim?
We now do but it needs more work because it shows fog almost everywhere at all times. That’s almost how the German Met Office does aviation weather predictions, always too dangerous to fly 
The site also doesn’t resolve waypoints which exist in more than one place, which is very common
That issue should be resolved now. We require an unambiguous waypoint for the first one and then use a nearest match.
when I first saw an Autorouter GRAMET a few days ago, I mistook the temperature figures for headwind/tailwind, so I’d recommend to add a degree sign to make the diagram more self-explanatory.
Done, there is now a degree symbol but we won’t tell you whether it’s Celcius, Fahrenheit or Kelvin 
One more feature present at Ogimet and lacking here is wind components along the route – either down below like at Ogimet, or next to the wind barbs (e.g. “R 2kt T 5kt” or “L 7kt H 3kt”).
We want to refrain from doing this at the moment. We don’t want the table due to space and not clutter the image any further. It already gives you a good estimation with the wind barbs and the magenta arrow. For exact values, you can use the PLOG. If we come up with a good way to display it, we will revisit this.
We now show TCU clouds with a TCU symbol. I know it’s not pretty that way and I promise we will have nicer cloud shapes in the future.
Thanks for the responses and the improvements. This will be an extremely powerful tool. A few more points to consider.
That is not “low level cloud” in the GFS model, but “boundary layer cloud”. Do you plot that, Achim?
…
We now do but it needs more work because it shows fog almost everywhere at all times.
If I understand correctly the output variables are:
hpblsfc: surface planetary boundary layer height [m]
tcdcblcll: boundary layer cloud layer total cloud cover [%]
Are you saying that tcdcblcll is usually high? The PBL height is not necessarily an indication of cloud tops (and gives no indication of base).
How do you determine convective cloud type?
You seem to be putting TS warnings on based on lifted index. That’s a nice idea. Consider CAPE as well.
The precip depiction is a bit difficult to follow (Ogimet is no better). How do you determine what to depict? Do you distinguish between convective and other precip?
In the US, there’s http://www.aviationweather.gov/icing/fip — current icing and forecast icing. The current icing product is probably what bookworm’s paper describes. As far as I can tell, the inputs are the GOES satellites, ground-based radar, a reasonably high-resolution model (RUC then, probably http://rapidrefresh.noaa.gov/ now), and pireps. The equivalents for all of those are available for Europe (though how much retuning of the algorithm would be required is not clear), but then of course one runs into the (potentially €€€) problem of getting real-time data from Meteosat and from the European weather services’ models.
Are you saying that tcdcblcll is usually high?
It was a matter of not using a reasonable threshold for TDCD.
How do you determine convective cloud type?
Not at all; GFS AFAIK does not generate that info. We just overlay TCU symbols on the convective cloud layer to distinguish it from the stratiform cloud layers.
Thunderstorm classification is currently as follows:
LI >= -0.5: unlikely TS
LI >= -3.5: possibly TS
LI >= -5.5: probably TS
LI >= -6.5: severe TS
LI < -6.5: violent TS
I’m open to changing the classification. Do you have a suggestion for a classification using lifted index and CAPE?
How do you determine what to depict? Do you distinguish between convective and other precip?
Precipitation uses the categorical flags, i.e. CRAIN, CFRZR, CECIP and CSNOW.
I just get a blank page when I go to router.euroga.org/gramet (after authorising the oAuth prompt and filling in the registration details).
Is it likely there will be a version of this that doesn’t require registration or signup? – I’m not really interested in the auto router having a strictly day VFR aircraft, but I do find the GRAMET stuff useful in weather planning.
