I guess you could make it work with airports producing a more higher resolution based on metar. But keep it with wind, visibility, pressure, temperature and dewpoint and leave the clouds be. Or I don’t know let meteoblue produce the clouds and have the metar only adjust them to their distribution (BKN, FEW, etc…).
As you say with the wind i want to know if the vatsim would be broken if the wind would inject it as a between value instead of a fixed rate from METAR. Example: If METAR says 10kt, could it be injected like this 4-19 with a random frequenzy? Or will that be inaccurate. Do it need to stay at 10kts all the time?
I’d rather get some dynamics in the wind than the steady blow like it is now. Wind gusts and variability would most definitely add to the immersion, I don’t think it would break anything while flying at Vatsim, most probably it would be even miles better and more challenging than the rails we have now.
2 Likes
Agreed. So a combination of perhaps the dynamic implementation of clouds and wind they had before where possible. And have the rest like visibility, pressure and temperature and perhaps upper winds be determined by static data. Or I don’t know. It seems like we won’t satisfy everyone. And not having the option to chose can’t be any excuse. There are options for the flight model as well.
Yes, that’s what I’ve been saying since the first “posting” - my point in all the “discussions” was not that we don’t need VATSIM/weather/METAR (although I personally would not have needed this change and of course I never implied it, I am there too) but that what we have here is not usable for VATSIM or anything else. and I don’t think the “bugs” can be solved so quickly:
- transitions METAR / Meteoblue (with correct cloud types)
- haze/visiblity jumps (this feature is of course very important for this sim)
- the total missing of the different cloud types (only then it could work slowly)
- no more cloud shadows (with these fluffy cumulus clouds)
- circular weather expansion known from Metar
- no upcoming weather fronts
- no real high level clouds
- this AGL/MSL “error” (almost don’t think it’s because clouds come out of the sea and that’s MSL and not AGL)
And above all and this is the most important POINT EVER: 0: plausible and realistic representation of the weather (even if not exactly accurate) !
Let’s see how this can be achieved, I am very looking forward to it
!
And briefly to your screenshots above about LKMT (I know it is not your intention to really praise it):
But that’s exactly what I hear (us) people complaining about, it may be that the values fit better now or it’s correct according to METAR but it doesn’t look realistic nor believable nor like anything we would see in real (probably, there are always some surprises in nature) , these flat clouds laid over this cumulus mishmash looks just like Active sky in the early days with FSX or P3D (with 3D clouds, can only repeat myself constantly, some people don’t care, but I do, because it was better before! ). And that’s what I’ve always complained about the weather engines - it’s not whether the “METAR” was accurate but the representation of it. Now of course nothing personal against you, people can show me so many times such screenshots but they won’t convince me of this “improvement”. This approach is not useful for a next-gen simulation unless there are real improvements. Almost everyone is in favour of improvements, but they should come when they really work. In this state the SIM is not usable for me (at the moment) and how long should I wait until the end of 2022 ? I don’t think so.
3 Likes
I have stopped waiting already! I have waited so much on those updates already to fix things in the weather but they only make things worse. I regret that i waited and did not enjoyed the sim more when the weather actually was better.
but have only now seen a few screenshots over there - in the “Weather regression - a broken mess” forum - looks better or not ?
The “mishmash” you’re talking about was there even before this METAR/Meteoblue blending came into play. The previous Meteoblue-only approach wasn’t any more realistic than what we see now. Right now we can at least have more realistic weather at and around the airport.
The weather aloft (clouds particularly) need work in terms of cloud types and graphical representation. Right now it looks like all we get are cumuliform clouds at different altitudes. Even a cloud layer at FL350 looks like a cumulus. It’s just a little more flat and translucent, but that’s all.
We need altocumulus, altostratus, cirrus and cirrostratus clouds. That would help the game a lot.
1 Like
I thought about a thing you said. If meteoblue/prediction or METAR old weather data is not good to use to inject weather what should be used instead? The real current weather is impossible to get into the sim i believe. It sounds like a feature way far into the future.
Xplane 12 will have prediction and METAR injected too. Maybe they will face the same problem as MSFS do right now. The announcement video of xplane 12 showed that metar dots not matching the prediction too. The prediction showing red where it was added circles of blue dots on top of it where it doesn’t match. I think on those dots we will see strange weather behaviour too.
1 Like
Does someone know for sure what prediction model is in use in MSFS? Isn’t that real important to know if they are using prediction? I heard NEMS global but is that confirmed somewhere? In that case where is the source for that statement or confirmation? I have wondered since release what model to choose when planning on those forecast websites.
1 Like
METAR has been the source of airport wind in MSFS since December of 2020 - long before SU4.
Whenever I fly, I always have the SDK utility program SimVar.exe open on a second monitor. This program allows one to monitor over 100 different internal simmconnect variables in real time. Since I am very I interested in weather and flight dynamics, I monitor 12 variables related to weather, altitude and airspeed, including the ambient wind direction and velocity, ambient pressure, equivalent sea level pressure, air density, static and true air temperature (which are identical when the aircraft is not moving) and others.
One other main reason I use the simvar utility is that I do not trust the in-sim ATIS. The utility program will show the actual weather conditions being injected into the sim environment at a given moment no matter what the source.
MSFS METAR winds are not absolutely static, even when the METAR reports no gusts. Watching the ambient wind direction and velocity over a period of time typically shows speed variations of 5 knots, and directional variations of around +/- 5 to 10 degrees. Now these variations are not “instant” as might happen in real surface winds - they seem to vary almost on a recurring sinusoidal curve - but they do vary.
The injected clouds from METAR in SU7 are not static either. In another post I recounted a flight I did yesterday afternoon departing from KDTW (Detroit). The actual METAR was giving 600 foot overcast, with two miles visibility. What I saw in the sim was very close to that. In this case, there was indeed a true overcast but the cloud layer was not just “sitting there”. The clouds were visibly moving with the strong southwesterly wind.
I spent 10 minutes sitting on the threshold of runway 22L in external view, just watching what the weather was doing. That runway is 12000 feet long, and I could just make out the far end of the runway, so the visibility was indeed very close to 2 miles. Sometimes it dropped down to less, sometimes it briefly improved to more.
Watching the overcast cloud layer (which was animated and moving), there were occasional breaks in the layer, which were moving with the clouds. When a “hole” would pass overhead, there would be a few moments of sunlight breaking through until the hole moved on.
It might interest you to know that weather models (such as the one MeteoBlue provides) are also static. I have been working with numeric forecast models for many years on a semi-professional level completely apart from flight simulation, and am very familiar with how they are internally structured, and what they data they typically contain.
A METAR is a “snapshot” of the weather at a particular location and time - but so is a weather model. In the case of the model MSFS is using - it contains a complete analysis of the atmosphere from the surface to (probably) 55,000 feet based on a 30 kilometer latitude/longitude grid. It is generated as 24 discrete forecast hours, and the data for any particular forecast hour and location is “fixed” for the entire hour. It does not vary in the sense that the model will say “At 22:10Z the weather will be this - at 22:15Z the weather will be that”.
No forecast model anywhere contains that level of time-varying detail minute-by-minute - if it did (for a worldwide model) the GRIB output file would probably be many terabytes in size.
There are some operational models like the US HRRR that can drill down to subdivide forecast hours into smaller segments, but those only cover a limited geographic area.
Yet, the model weather in MSFS is “dynamic”, so how is that being accomplished? Most likely by interpolating between adjacent static model forecast hours. In other words: if the model predicts a cold front will be located a location “x” at 22:00Z, and location “y” at 23:00Z, the weather engine moves the front over the course of the 22:00 to 23:00 time from.
If the MSFS weather engine can dynamically animate model-predicted weather at a given location between two adjacent static forecast hours, why could it not do the same for METAR? Based on what I saw yesterday at KDTW, it certainly appears to be doing that.
Don’t take my word for any of this. Try it for yourself. Spawn at an airport which is reporting bad/stormy weather, and spend some time just watching what the clouds and visibility do over time.
I am the first to acknowledge that the new METAR implementation has several flaws. The visibility (in the US) cannot always be exactly 10 miles when the METAR indicated “10SM”, and cloud heights definitely appear to be wrong more often than not. (although at KDTW yesterday they were correct).
3 Likes
I can understand how the model can animate between the forecast hours, but how is that possible with METAR? No one knows what the next METAR will be until it’s reported so how could the simulation know what conditions to animate towards?
Also, many thanks to you for constantly explaining the same thing over and over again to everyone.
agree with you. just looked at this video.
They never said it will be 100% accurate at the beginning either. It’s based of many forecasts models to be more accurate though.
I can’t see how to manage get METAR fit in those models though. It’s two different things.
But is it not very important to be able to trust ATIS in the sim? That they need to fix!
1 Like
They do, but that’s not necessarily an issue with the weather itself. That’s more of an issue with the ATC system being able to access that data. Or at least it was an issue before SU7. I don’t use in the in-sim ATC so I don’t know what the ATIS is reporting after SU7.
I absolutely agree!
1 Like
that’s all I meant ! 
Well there we misunderstand each other a bit, there is already a difference between “realistic”, authentic in the sense of “appearance” and correct - according to METAR "which does not mean that it is “realistic” - I would rather use the word “correct” than “realistic” when it comes to METAR, I think we mean the same thing anyway just with a different expression ! 
1 Like
It is, if atis not correct we compare with METAR and that does not match the prediction that atis should report correctly.
It depends on the location and time. IF the model accurately predicts what the weather will be at a particular airport at a particular time, the more likely the Model clouds will match the actual METAR clouds and the “blending” is likely to work well. If there is a significant difference, the more likely there will be visually jarring differences.
I have now done a total of 12 test flights over the last few days in various locations trying to analyze exactly how the METAR/Model mix is working, and it appears that the transition from METAR clouds to model clouds happens at between 6000 and 8000 feet AGL. If you are sitting on the ground, looking up, any clouds you see at altitudes higher than approximately 6000 feet will be from the model, while lower clouds will be from the METAR.
If you are in flight at 8000 feet or higher, looking down, all the clouds you see will be from the model and none from the METAR. At lower altitudes METAR clouds and visibility start to mix in.
2 Likes
Please bring BACK the old weather system. MSFS weather now reminds me of the sims of the past.
And why did you change what was almost perfect?
3 Likes
I was talking about the actual way the systems are built and integrate with each other. Yes, ATIS is related to weather, no dispute there. But the fact that ATIS in the sim was not reporting the correct values was not indicative of an issue with weather itself, but was a bug (or limitation) of the ATC system.