Hi Everybody,

Now that I have my night lighting adjustments just about done, thought I'd spend some time working reshade for day flight, during my editing I came across this amazing looking storm over Tulsa. 

https://drive.google.com/file/d/1qm9Y786E5jF790cvCGpuaMh-MHcmxjWU/view?usp=sharing

https://drive.google.com/file/d/1rW4AypAaj5GN39zIbxi_nTWnkpz1qk1B/view?usp=sharing

Yes I totally agree. The cloudscape in Xplane 12 is leaps and bounds above the competition. Last week I was flying over a frontal system....I was absolutely blown away by the various cloud layers... mostly stratiform (as it should be for a frontal cloud band) but with some incredible embedded convection. The realism is spot on. 

Yes XPlane has finally reached a very high level in the clouds and sky depiction indeed. For somebody like me who's been stuck on P3D and Xplane11 for what felt like ages, seeing such a sky in XPlane12 makes it feel like a game changer.

That being said, there's still one little thing that disturbs me: I can't help feeling the clouds are often depicted like upside-down... Don't you guys get the same impression ? Especially with the first shot above ? It's like, I'd really love to see the same picture, but edited by applying an horizontal mirror to the clouds 😄

 

It's indeed the case with Xp12 clouds, specially stratified and convective have gotten a lot better with time.

My only problem right now with Xp12 is when there are mid level clouds, above the METAR reporting altitude (5000' in my area).

Sometimes, specially if not nimbus or Cbs, Xp12 fails to represent it. I don't know if the problem lays on the way it get's the data from GRIB. I believe Xp12 uses GFS GRIB, not ECMWF? Both incorporate satellite model data to estimate cloud layers too.

OTOH, lately Ci are more acceptably represented than they were in the past!

Then there's lighting and colours wihch are really true to real in Xp12 !!! There are shots where I honestly have to look for a while and search for some details to tell it appart from a real world image 🙂

Edited October 25, 2025Oct 25 by jcomm

7 hours ago, jcomm said:

Both incorporate satellite model data to estimate cloud layers too.

Have you ever looked into the grib file to see if that information might be carried inside?

1 hour ago, blingthinger said:

Have you ever looked into the grib file to see if that information might be carried inside?

Yes, as part of my day job 🙂

BUT! I just checked LRCL, for instance, where medium clouds are Grib-reported, and METAR reads :

METAR LRCL 251600Z 05004KT 020V080 CAVOK 09/01 Q1010

and indeed Xp12 correctly displays those clouds with their base around 9500' !!!

Either the last time I made this type of accuracy tests there was some problem with real weather feeds, or it has been updated and is now even more realistic(?)

 

 

Edited October 25, 2025Oct 25 by jcomm

1 hour ago, jcomm said:

Yes, as part of my day job

XP uses the exact same gribs you'd use at the day job?

1 hour ago, blingthinger said:

XP uses the exact same gribs you'd use at the day job?

I'm not sure... I use primarily ECMWF, but doesn't differ in terms of data.

You have height of the cloud base for each level of the model as well as an estimate of the amount of cloud cover, but it requires additional processing.

3 hours ago, jcomm said:

Either the last time I made this type of accuracy tests there was some problem with real weather feeds, or it has been updated and is now even more realistic(?)

I noticed that too – during the period when METAR had upper hand (somewhere between updates), there were stations that were never correct – because here in the EU, METAR data above 5000 feet is not provided (as you said) – now the ‘weight’ seems to be better distributed again – and the whole weather forecast seems more authentic – hopefully it will stay that way - I would really like to have a slider where I could set the METAR/FORECAST division!

Edited October 25, 2025Oct 25 by AUA425

3 hours ago, jcomm said:

Yes, as part of my day job 🙂

BUT! I just checked LRCL, for instance, where medium clouds are Grib-reported, and METAR reads :

METAR LRCL 251600Z 05004KT 020V080 CAVOK 09/01 Q1010

and indeed Xp12 correctly displays those clouds with their base around 9500' !!!

Either the last time I made this type of accuracy tests there was some problem with real weather feeds, or it has been updated and is now even more realistic(?)

 

 

But is that cloudbase evdn accurate,  seeing many videos where pilots are reporting cloudbase on descent for ARC, provably because atis is not correct. Weather is only as accurate as the last report, which we all know could be completely wrong by the time we see it.

15 minutes ago, jcomm said:

I use primarily ECMWF

Do you get cloud type? Or just %coverage?

1 hour ago, blingthinger said:

Do you get cloud type? Or just %coverage?

ECMWF does not list cloud types, it lists clouds based on their altitude which integrates to a color model to show "depth" of the cloud layer

I use the same model at work and flightplanning IRL.

But GRIB2 has the category for it, but not sure who uses  it

Quote

# Code table 4.203 - Cloud type
 0 Clear
1 Cumulonimbus
 2 Stratus
 3 Stratocumulus
 4 Cumulus
 5 Altostratus
 6 Nimbostratus
 7 Altocumulus
 8 Cirrostratus
 9 Cirrocumulus
 10 Cirrus

 

1 hour ago, mjrhealth said:

seeing many videos where pilots are reporting cloudbase on descent for ARC, provably because atis is not correct

Our METAR /ATIS met reports (I work at the national Aviation Agency) are based on 8km/ 4,5NM  radius , or 8-16km for phenomena in "viscinity" .

Naturally, arriving crews will likely encounter variations in cloud base depending on which direction they arrive from. Right now one of our airports show 1400 SCT while the other is 4400 OVC. They are  located 15 NM apart.

Edited October 25, 2025Oct 25 by SAS443

14 minutes ago, SAS443 said:

I use the same model at work and flightplanning IRL.

Ultimately it is sounding like the gribs you and Jcomm encounter at work are not the same gribs as what LR is sending out for ingestion into the sim. Grib is merely a container format and the content itself can vary greatly. I'm wondering if LR is simply passing gribs along or if they are repackaging data into their own gribs.

Currently I am concluding that LR is forced to interpret the cloud type on their own. It sounds like they get a %coverage value in one square in the atmosphere (eg 1x1km) at a range of altitudes (from NOAA, so GFS not ECMWF) but it doesn't sound like there a cloud type carried along with the % coverage. Besides the fact that even that spatial resolution (1km square) forces a significant amount of creative interpretation on cloud type, they're doing remarkably well given the constraints.

3 hours ago, blingthinger said:

Do you get cloud type? Or just %coverage?

Percentage only. Depth is obtainable through additional processing... but if Cb or Ns is also specified.

Further tephigram/skew-T based post processing can be used to forecast depth for other types of nebulosity.

Grib2 is richer, but the exact classification of the forecast cloud types is not included, yet it has a greater vertical granularity. 

Resolution is yet another important factor and the free access GFS data that I believe is used by XP is probably 5 times wider than the restricted / paid access (0.1°, around 7.8 km at mid latitudes) available form ECMWF.

But from my various tests today, Xp12 did a remarkable job, and since it represents stratified clouds more plausibly than MSFS, the end result is really satifactory.

 

Edited October 25, 2025Oct 25 by jcomm

1 hour ago, blingthinger said:

Ultimately it is sounding like the gribs you and Jcomm encounter at work are not the same gribs as what LR is sending out for ingestion into the sim

No.

"We" use ECMWF, output its own GRIB (based on a nonhydrostatic model, "altitude" or how topography influence the weather). This is not a free product, but "some" subsets are available to the public. 

Whereas LR uses the free(?) GFS GRIB and is hydrostatic (based on pressure, simplifies equations for faster compute thus updates more often than the above)

ECMWF is particularly good at modelling weather over varying topography. Making it a good choice for northern Scandinavia flight planning (where I fly), the alps...or any mountainous region, really.