KUDOS to weather people

[SurferTim]

I'm now using FG2020.4.0 with advanced weather selected. I'm not certain who is responsible for the "new" advanced weather, but to you: NICE!!

During a flight into and out of Chennai, I flew through what was reported to be thunderstorms. It was. The approach and landing was tough.
On departure, a thunderhead was directly in my path. I decided to see what it was like to fly through it. Orange sky, what appeared to be hail or freezing rain in abundance, and severe turbulence. IRL, I think that would have been fatal.

[Isaak]

Kudos to the weather people indeed!

I experience one issue though: when flying at altitude with 'aloft waypoints' wind model and live weather, when entering a new weather region, the winds get reset for a brief moment, instead of evolving to the new wind direction/speed. Say when flying heading 090 with 40kt winds from heading 260, the wind model resets to 0 kts at heading 0° for a brief moment, after which it stabilises to e.g. the new wind direction of heading 250 at 50 kts. This means that the autothrottle retards to idle because the aircraft is suddenly overspeeding, and then picks up again to adjust the air speed to the new (increased) tail wind. In FG 2020.3.x, the winds would shift gradually from the old regime to the new regime without this hard reset, resulting in a much smoother transition. That way there was a smaller chance of encountering overspeeding/stalling situations. Is this behaviour intended? Should I use another wind model? It's not a big deal, but it looks like a small regression.

Thanks a lot for the hard work!

[SurferTim]

I noticed that abrupt change also. The Citation and Twin Otter both do a little dance with a weather update.
I can live with that for the moment.

Edit: I highly recommend "checking the weather" just before final approach. It apparently won't check again for a few minutes. Don't want that change spasm happening on short final.

[merspieler]

Are you sure, this is only on the new version? I've always had those sudden wind shifts.

[wkitty42]

I'm now using FG2020.4.0 with advanced weather selected. I'm not certain who is responsible for the "new" advanced weather, but to you: NICE!!

FWIW: AFAIK there is no "new" advanced weather... i don't recall seeing any major updates or changes to FGData in weather related stuff at all in the last several months to a year... i mean, nothing substantial, really... there were a few updates this month but they only fixed a few things that were already in place... these updates were also applied to 2020.3...

18 Aug 2021 22:16:20 +0000
[FGData] branch next updated: Local Weather: auto-enable 3D clouds and disable the checkbox. http://sourceforge.net/p/flightgear/fgdata/ci/c9709e2a62848f99c0702b4e223beb810d1e42d8/
[FGData] branch release/2020.3 updated: Local Weather: auto-enable 3D clouds and disable the checkbox. http://sourceforge.net/p/flightgear/fgdata/ci/a8b3272e0accc69b418588f9ed6d4af33d516383/

19 Aug 2021 00:33:34 +0000
[[FGData] branch next updated: Weather dialog: stop resetting weather when Detailed Weather has been loaded but disabled (fix for Issue 1 in https://sourceforge.net/p/flightgear/codetickets/2486/) http://sourceforge.net/p/flightgear/fgdata/ci/d4dfafaac7a1b006de1441e7de9654a860044429/
[FGData] branch release/2020.3 updated: Weather dialog: stop resetting weather when Detailed Weather has been loaded but disabled (fix for Issue 1 in https://sourceforge.net/p/flightgear/codetickets/2486/) http://sourceforge.net/p/flightgear/fgdata/ci/e3db6a0f959b5c83098b3358bade44978675be37/

23 Aug 2021 19:22:03 +0000
[FGData] branch next updated: Local Weather: auto-enable 3D clouds and disable the checkbox. http://sourceforge.net/p/flightgear/fgdata/ci/d394edb0a28e41d4a2a3e0d4bcf9b09fd9b3e726/
[FGData] branch release/2020.3 updated: Local Weather: auto-enable 3D clouds and disable the checkbox. http://sourceforge.net/p/flightgear/fgdata/ci/380f3228efe83d2cdba65323d9e0bcbe0f990bd5/

previous to those, the next oldest weather related update was back in February of this year (2021) but it doesn't appear to have been applied to 2020.3...
08 Feb 2021 08:44:51 +0000
[FGData] branch next updated: Fix issue 2537, re-enable the wind interpolation when live data is (temporarily) unavailable. http://sourceforge.net/p/flightgear/fgdata/ci/0c07fce69f33dfe2d4cc9bed48b70fb8cd7fc861/

and then there's this one from back in November 2020 which was also only applied to the next branch and not the 2020.3 release branch...
09 Nov 2020 10:44:16 +0000
[FGData] branch next updated: Add climate selection options to set the environment and weather based on climate data. Setting the weather is not yet implemented in the code but eventually it will be. http://sourceforge.net/p/flightgear/fgdata/ci/7f66257bfb0ff1b9cd2b79e89bccf4efc2836af1/

those are all i see on a quick look over the subject lines in the 249 messages in my thunderbird's flightgear-commitlogs mailing list archive when filtering the subject and bodies with "[FGData] weather"...

[SurferTim]

I have noticed it too. My last crash in the Twin Otter (long time ago) was a weather change on short final. A 15 knot headwind shifted 180 degrees just like that.

@wkitty42: I'm going from FG2019.2.0 to FG2020.4.0. It appears there has been a change.

[wkitty42]

@wkitty42: I'm going from FG2019.2.0 to FG2020.4.0. It appears there has been a change.

i can see that, in a manner of speaking... those are the only ones since 2019.* was released that i can see ATM... i'd have to dig deeper to see when 2019.2 was released and what may have been added to next before 2020.* was cut...

note that i'm not saying there wasn't any noticeable change but these are all i see in my list... of course compositor makes a huge difference in the visuals, too

[Isaak]

There were always the sudden wind shifts, especially at high altitude, but there's definitely a change between 2020.3.8 (last stable I checked) and 2020.4.0 in that the weather doesn't jump from old weather to new weather as it used to be:
2020.3.8: current winds > shift > new winds
2020.4.0: current winds > shift > zero winds > shift > new winds

I have no idea if it's the weather code itself or something else that has changed, but it doesn't keep me from using 2020.4.0, so please don't dig into it if you weren't already planning to do so

[V12]

I'm complaining on abrupt pressure and wind changes 3 years, this is one of the main reasons of my switch to the commercial sim.

[GinGin]

I'm complaining on abrupt pressure and wind changes 3 years

You ? Complaining ? No way ...
Yet again, have you learned anything for the past 3 years dev wise that could have improved the points you were/are complaining about ?

[V12]

Isaak :
I analyzed my video at https://www.youtube.com/watch?v=LZQx9WHaAFY, there are 3 significant abrupt wind changes (at 27:15, 32:23 and 33:35), but no one with pattern current winds > shift > zero winds > shift > new winds, all has pattern current winds > shift > new winds It was recorded with 2020.3.11 on Win10.

[Thorsten]

I'm complaining on abrupt pressure and wind changes 3 years, this is one of the main reasons of my switch to the commercial sim.

Yes, and the problem is analyzed to death (a fact you seem to forget the minute the analysis is over), the underlying reason is

a) METAR data is fetched in a particular way from the closest station that can mean you've seen the last report 500 miles ago and the winds 'here' have nothing to do with the winds 'there' you've been using so far

b) METAR reports ground winds that are influenced heavily by terrain - so even for close-by stations reported winds can vary a lot

As you can easily verify using the (rather well-developed) offline weather system, the weather simulation does not create any surprising wind jumps - it's the nature of the METAR data which forces shifty winds.

And as we've also stated a gazillion times, the situation will persist until someone supplies different data. For instance there's an addon to fetch aloft winds (from a commercial service? I don't remember) and such data will not experience sudden issues.

You seem to expect that someone owes you the weather data quality of a commercial provider - just for free - but the rest of the world doesn't share that view. So please go playing with your commercial games and stop complaining.

[Hooray]

@V12: Both, the fgcommand system and Nasal support interpolation/extrapolation functions/tables, which could certainly be used to smoothly begin "fading in" the new METAR data - that way, "sudden jumps" won't be as noticeable. I am sure that Thorsten can provide the corresponding pointers - but speaking in general, you will want to search the sources for "METAR" and look for any places that fetch the METAR info, and then use a timer (maketimer) to set up a smooth interpolation of the new METAR data, you could make the thing configurable (say 5-120 seconds) or even use heuristics to compare two different METAR sets if they vary greatly, use a longer period to fade in the new METAR - basically "mixing" two different METAR sets over the course of several minutes, depending on how old the previous METAR was and how current/close the new one is, i.e. with less focus on the old one.

At that point, you'd have a "virtual METAR", i.e. by mixing multiple different stations - which is also what you could use to deal with different data sources.

[legoboyvdlp]

I'm now using FG2020.4.0 with advanced weather selected. I'm not certain who is responsible for the "new" advanced weather, but to you: NICE!!

Just to be clear it's not really new - it's been there since FlightGear version 2.4.0 (10 years ago in 2011)

As you can easily verify using the (rather well-developed) offline weather system, the weather simulation does not create any surprising wind jumps - it's the nature of the METAR data which forces shifty winds.

And as we've also stated a gazillion times, the situation will persist until someone supplies different data. For instance there's an addon to fetch aloft winds (from a commercial service? I don't remember) and such data will not experience sudden issues.

Indeed. I improved the situation somewhat by making the closest METAR check update faster - that helps when you have two different airports, close-by, with very different weather (generally in mountains), but does nothing for transatlantic changes.

To explain it in simple terms.

You're flying across an ocean which is 5000 miles wide from east to west. For the first 2500 miles of the crossing your only weather data is from the last airport on the east bank. At the 2500 mile mark, you finally load weather data from the west bank. Thereafter, the interpolation routines interpolate across the whole 5000 miles, resulting in a short-lived discontinuity as the new weather data is fed in.

The only fix here is to add more weather data.. Which is exactly what I did. I added the ability to add aloft wind points from .grb2 files ... that simply inserts more data, so that you don't go 2500 miles without a weather update.

@V12: Both, the fgcommand system and Nasal support interpolation/extrapolation functions/tables, which could certainly be used to smoothly begin "fading in" the new METAR data - that way, "sudden jumps" won't be as noticeable. I am sure that Thorsten can provide the corresponding pointers - but speaking in general, you will want to search the sources for "METAR" and look for any places that fetch the METAR info, and then use a timer (maketimer) to set up a smooth interpolation of the new METAR data, you could make the thing configurable (say 5-120 seconds) or even use heuristics to compare two different METAR sets if they vary greatly, use a longer period to fade in the new METAR - basically "mixing" two different METAR sets over the course of several minutes, depending on how old the previous METAR was and how current/close the new one is, i.e. with less focus on the old one.

@Hooray, you're preaching to the choir -- the weather system already does that, it already does interpolate it, all you say is already being done (except in a slightly more complex way) -- but it can't handle only receiving data across transatlantic distances very well. Again, the only fix is to add more weather data -- which currently can be done through parsing + injecting .grb2 files.

[V12]

Gentlemans, on my video is short flight over the Alps, not long haul over the Atlantic.