Well, they didn’t dig too deep into technical details, but from my understanding the main difference with SU7 should have been the “integration” of Metar data directly into the Meteoblue weather model.
2 Likes
How many want’s to have metar integrated in a dynamic weather system? I think it only will conflict with the dynamic part. It’s not dynamic anymore if they add some static data in it. It’s like to watch a movie that has some spots in it with a picture from some other movie. How can we make that fit in to the movie 
Maybe make it invincible but then it makes no point to have it there.
1 Like
Well if it’s Metar that is reading what the sim is currently depicting that’s great and therefor creating the Strips according to what’s going on in the simulator. But I rather not have a patchwork of static metar weather with gaps of “filler weather” to be the new “old” norm. Especially considering there already are tools for that if people want that sort of thing.
3 Likes
This is such a complicated topic. I sympathize with the points @Aviara has been making, and I made a forum topic a year ago with similar thoughts here:
There will always be discrepancies between METAR and weather model data. Merging the two is a VERY difficult task, and it’s extremely difficult to avoid abrupt changes or unrealistic artifacts. On the other hand, the existing weather solution was also not adequate.
I’m a meteorologist, and I use weather model data for airport forecasting all the time. The reality is that weather models struggle to accurately depict IFR conditions. Global models such as the GFS model (run by the US National Weather Service) are notoriously inaccurate for low-level clouds and fog. To pick one example: in the Great Lakes region of the US, widespread stratus is common in the winter on the northwest side of storm systems. The raw cloud cover data from the GFS model is frequently useless for forecasting this type of weather.
I’ve seen similar problems in MSFS. Last winter, I began a flight at an airport in the Great Lakes region that was under a widespread (stretching for hundreds of miles) stratus layer in real-life. In MSFS, there was brilliant sunshine with only a few widely scattered clouds.
Ultimately, the discrepancies could be improved with higher-quality, higher-resolution weather model data. High resolution models usually do a better job with stratus clouds and localized effects. Combining high-resolution model data with METARs would likely produce a better weather experience in MSFS than trying to merge relatively low-resolution global model data with localized METARs. But there are problems there too: high-resolution models usually only cover a regional domain. The data is not available for the entire earth. If MeteoBlue stitched together multiple high-resolution forecasts, the transitions could be abrupt and frustrating. The way I see it, there is no perfect solution to this problem. Every solution will have trade-offs.
3 Likes
I hear you. I’d rather have some sort of tool to read the data as the sim is rendering the weather at any given destination airport for flight planning purposes and stick with the before-SU7 weather instead of beta testing this new approach. But I guess they want to appease all those community members that were asking to open up the weather system for the “old” Metar based tools and were denied.
And here we are, going down the same road as the “night lighting improvement” demands by the community: first “improvements” weren’t great (to say the least), but now it’s pretty decent. Took only one year.
Oh, btw. They’ve announced “something big” regarding the weather for the end of 2022. Guess how that will play out!
1 Like
Approaching KONT, where differences in reported METAR visibility cause an extremely harsh transition. This breaks the intent of the new volumetric weather system, but also ruins VFR flying at certain times and locations.
8 Likes
Wow, that is terrible. 
Instant IFR.
Then perhaps stop trying so hard to imitate the real weather 1:1. It’s impossible anyway for any computer on earth. I really liked their approach pre update 7. It needed a few tweaks here and there, but to the most part it was believable and highly immersive. Let the simulator weather do its own thing, and have it use real world weather data as a baseline. What I really loved about the weather as also the consistency in multiplayer. We all were able to see the exact same weather and clouds and cloud position. And the weather itself was seamless. That’s a real world weather imitation I want in a flight simulator. And over time continue to fine-tune the parameters. But let the weather ecosystem in the simulator be unaffected by real world event that will cause unwanted side-effects like harsh transitions, which includes wind as well. And again provide METARs defined by the simulator weather. I believe if you provide 24 hour old data you can get pretty accurate results as the weather engine will be able to anticipate and thus keep things smooth, seamless and accurate.
2 Likes
If you fly in the opposite direction, does the haze disappear again?
Yes it does
that’s right, thanks for the explanation !
I have once (at that time) despite my not so great programming knowledge to “create” a program that makes it possible to make the then at a certain height suddenly disappearing visibility layer in FSX “smooth” - there I have really learned a lot about “interpolation”, “haze layer” (which can be not only horizontal but also vertical, this has nothing to do with fog), fog layer etc…
But this is how I would proceed (for now in the MSFS and if it is so important):
Yes, the problem with forecast data is the lack of low lewel clouds and fog - BUT - exactly the values and only the values including wind data I would take from METAR (and not vice versa).
Wind data I would slowly “smoothen” to the values of the reported values e.g. below 3000 feet (or whatever) within a radius of one METAR station (is possible - FSUIPC has shown us this before, (20, 30 nm, depending on how far away the other METAR stations are or how much data is available).
Clouds and the other forecast data I would leave as they are, i.e. THE ENGINE AS IT WAS and then if possible access higher resolution data or use satellite-based evaluation (Jörg has already talked about this).
And now for the haze layer - that’s one thing - anyone who has flown more often knows that it is almost always “hazy” (this has nothing to do with the visibility on the ground). Once more, once less, in some regions more or less - but to get it so smooth and in the horizontal and vertical levels is almost impossible (if of course the other values are also taken into account).
I think this “aerosol” value should have taken over - also not a bad approach (as this value is also given in forecast data)
and as long as this can’t be done automatically - render a constant layer of haze (like XP11 does, but unfortunately a bit too much, but can be adjusted via DataRefs)
- with a slider 0%-100% or additional on / off and that’s it (for now).
And I am not interested in reproducing the “real weather” 100%, although I would like to - that is not possible anyway, with METAR, forecast or whatever - but something “credible” in terms of appearance and above all constant and without any abrupt changes and other anomalies (and that is what we had before SU7 !)
1 Like
Asobo should at least make a statement and give us some info about an eventual hotfix. Especially when something gets really broken in the sim
4 Likes
Yes, but not right away. If you turn around immediately, you usually have to fly for miles until it goes back. Like you either have to get closer to the next METAR or the next volumetric voxel. So it makes it difficult to avoid this issue as a result.
Part of the problem is a lack of communication/documentation on how the system is supposed to operate.
Most of the live weather issues that occurred late last year into the first half of this year was finally discovered to be an issue with the sim loading in the wrong forecast hour for a certain portion of the day. (This is why you’d have half the users complaining the weather was wrong while the other half said the weather looked okay to them). It took months for the community to determine when weather was being updated and how often because nothing is documented or explained anywhere. It all was discovered through observation and reporting on our part.
Has anyone noticed how the complaints about inaccurate weather dropped significantly after April/May this year? That’s when the issue was finally fixed and most people had mostly accurate weather since within the confines and limitations of the system. But due to a lack of explanation, by the time it was finally fixed, everyone had already jumped on the METAR bandwagon and used that as the sole explanation for why the weather wasn’t accurate.
Flight simulators are complex, and weather even more so. How the system works and interprets the data needs to be explained to us so we know what to expect and what to look for when things don’t seem right.
5 Likes
Personally I had no complain about WX after SU5. It was smoth and pretty accurate. What I don’t understand is that i believe these SU are tested for weeks before release. One of these 3 things here:
1: They don’t test
2: They are sooooo incredibly lucky not to spot the issue
3: They just don’t care
It was clear that most people would have complained about such a problem so I don’t understand why to complicate their life
4 Likes
I completely agree on the end goal: I would like to see credible-looking weather without abrupt, unrealistic changes. To be honest, it sounded like that’s what we would be seeing in this update. Before SU7, the weather engine used global model data from MeteoBlue and modified it based on the latest METARs. But the modification seemed to be weighted toward the model data, rather than the METAR. On one occasion, I flew into an airport where the model data indicated broken clouds, but the METAR reported clear. I was flying through clouds until I approached within a few miles of the airport, and then I suddenly broke out into a clearing around the airport. This was unusual before SU7, though. More often, the sim weather looked consistent, without any abrupt transitions, and it didn’t always match the METAR.
Leading up to SU7, the devs explained that MeteoBlue would be incorporating the METARs into their model system. That sounded like an improvement to me. The National Weather Service does this in their high-resolution weather analysis (known as the RTMA – Real Time Mesoscale Analysis). The analysis is based on a high-resolution model forecast adjusted to more closely match METARs and other observations. Overall, the RTMA produces a realistic-looking weather depiction without abrupt changes around METAR stations. That’s what I hoped to see in SU7. I’m not sure why they were unable to achieve it, but it’s worth noting that the RTMA has been under development at the National Weather Service for more than a decade. Producing a high-quality product that blends observations and models is VERY hard. With more time and effort, I’m sure MeteoBlue can improve the current product.
6 Likes
Perhaps then what we should all do to help troubleshoot this…
When we encounter sudden transitions:
- Take note of the real world time it happened (not the simulator time)
- Take note of which METAR station you were flying into and the general location where it happened.
- If you can, after a transition turn 180 degrees in the opposite direction and take note if it suddenly transitions back to what it was and how long it takes for that to happen.
For this issue it’s important to determine if this happens due to ■■■■■■ transition between METAR stations or ■■■■■■ transition between weather updates to the sim itself.
1 Like
My sense is that the testing period is too short. In order to fix all the bugs and issues identified during testing, they might need a few months to iron everything out. There isn’t enough time to address everything, so fixes are planned for the next update, rather than delaying the current update. So in my opinion they are testing, and they are likely identifying many of these problems during testing. But the timetable doesn’t allow for a proper fix.
2 Likes
Everyone was asking for METAR based weather, now we got it, with all the downsides of it, like appruplty changing weather.
I hope they can smooth it out, but then again, peoples will shout “IT IS NOT EXACTLY AS IN THE METAR”…
1 Like
It’s definitely a spatial/geographic transition (a temporal one might also be possible). You can indeed repeatedly cause the transition by flying back and forth between METAR stations.
An easy one to reproduce is to simply fly north and then south out of KONT. The Mojave Desert just to the north is almost always clear, and Ontario sits in the LA Basin which is inundated with the marine layer this time of year. Once you get over the mountains, the fog disappears suddenly. You have to get within a few miles of KONT to make it come back, and likewise fly all the way back to mountains to get it to go away. But the change is always instant and over the same location.
2 Likes