Thanks @WxMarc , really enjoyed your opening !!
the forecast model they use can be up to 12 hours old.
if you wanted to know the current weather at your local airport this morning, would you check the 1 hour old metar, or the forecast from yesterday afternoon?
Then you can realise how hard it is to blend those two together. If forecast is completely wrong and then blend in METAR the outcome will be something random in between those weather. Not accurate either.
I hope they can figure it out because before SU7 the weather felt more realistic to me.
Now we do have more accurate (compared to real life) local weather at airports. But now we also have the classic problem that you see clouds and fog spawning in like in the older simulators which ruins immersion for me.
The dynamic weather model made MSFS weather so good and realistic, and now we’re back to METARs.
Agree with you!
What have we actually gained from having METAR creates clouds in the sim? Maybe a bit more accurate when it comes to visibility but it’s less realistic to me.
Yeah i feel we are going back to an inferior system because that’s what the sim community is used to and comfortable with, not because it’s actually better.
Still hope they can make it work but it will be very hard to blend it all together and make it look believable.
If we could help them make it better how could we help them fix it? Some tips? I think the most difficult task is to blend in overcast and clear sky. How can they blend in that without make it a sudden transition?
you give weight to the more likely to be accurate condition.
ie. you’d use an actual observation instead of a forecast. you’d use a more recent metar rather than an older metar, etc. etc. etc.
if the current metar says its overcast but the forecast from yesterday says it should be clear, i think the choice is obvious.
Then put out that in a circle with clear sky around that circle?
metar covers like 8km radius
you can randomize it why would it have to be a circle
the forecast model itself has a resolution also, you know
Very difficult. I think if the METAR is very different from the meteoblue model they should fall back to the meteoblue model to prevent the sudden transition. If you prefer the METAR you will get a hard transition no matter how accurate the METAR is.
Maybe they should only get visibility / cloud level from the METAR but the actual coverage from meteoblue. This way you have somewhat accurate conditions but you see no sudden wall of clouds or hole in the clouds.
This will work better if they have more cloud types and better overcast visuals.
But weather does not happen randomly. It’s a global thing. It may work sometimes, sometimes not. It’s many variables to weather that needs to match to look natural. And the higher we are in the sky the more noticable those are if manual changed.
And those things may work in other sims with pictures as clouds. In this sim we have volumetric clouds.
there already is randomness in the model. the meteoblue model has a resolution itself (250M points), and is smoothed between those points
metar would just be overriding to whatever degree those points around an airport
btw the other sims have volumetric clouds too
as i’ve said before there are smarter people than any of us that can solve or have solved many of these problems
I have been watching videos on REX, and Metar itself seems very well capable of rendering any clouds layers if reported. But the thing is that with metar you get some stark transitions, therefore as people said, METAR is good if you are planning to fly in a specific area, not far from it. For long flights from point A to point B meteoblue will give you a more accurate and smooth transition
Yes, probably. I only know what i have right now though and that is not better than it was before. If we could planning with only the visibility, wind, temp and pressure from metar i would be happy if they not touched the clouds at all with METAR. Then instead improve the accuracy of those predictions in use. May take many years but that i think is the future.
OMG Yes! I remember think it was release version maybe SU1 then it all went to crap around SU3 or something. Clouds USED to be really nice!
Is there a reason why they need to use a forecast model that is up to 12 hours old? What would be wrong with getting updated forecasts every few minutes?
I think that issue was fixed already. Wasn’t that you that helped them find that issue @utd4life12?
the giant supercomputers that do the computations only run the simulation a couple times a day, or once a day, or maybe 4 times a day at best…
apparently the meteoblue forecast is twice a day (but i’ve also read once a day)
If you’re referring to the issue earlier in the year where it would load in the forecast around 12 hours in advance then yes as far as I’m aware it is fixed.
we discovered that an area of data from around 20:00z-08:00z was missing/corrupt, so as soon as it hit 20:00z each day it loaded in the next set of data from 08:00z the next day. That issue though was fixed