Board index FlightGear Development Weather

Advanced Weather debugging

Everything related to weather simulation, visuals should be discussed in the shader subforum.

Re: Advanced Weather debugging

Postby V12 » Thu Aug 22, 2019 7:18 pm

https://ulozto.net/file/reguUlBgLaqA/lsgs-lowi-ods
Another captured data with wind direction changed in 5 seconds over 120 degs, resulted into 10 kts overspeed caused AP off (IDG A320). Again 5 seconds after active METAR station switch. Is possible extend interpolation time from that 5 seconds to 30 or more seconds ? Why is this time always 5 seconds ?
Sample rate is standard 1 second, route from LSGS 25L STAR SPR2U to LOWI 08 SID XEBI1B approach RNVZ08
Last edited by V12 on Thu Aug 22, 2019 7:22 pm, edited 1 time in total.
Fly high, fly fast - fly Concorde !
V12
 
Posts: 2757
Joined: Thu Jan 12, 2017 5:27 pm
Location: LZIB
Callsign: BAWV12

Re: Advanced Weather debugging

Postby Thorsten » Thu Aug 22, 2019 7:20 pm

Is possible extend interpolation time from that 5 seconds to 30 or more seconds ? Why is this time always 5 seconds ?


if you want to change the interpolation algorithms because you believe you know better - first start to learn how they work, then start to understand why they might be implemented that way. Then make your changes and test them for multiple different use cases, deal with the complaints of your testers - then make a merge request and have a discussion about pros and cons based on test results - that's how development really works. Not by claiming 'this must work so and so' out of the blue.

If your AP can't deal with rapid changes - make it more resilient. Reality has gusts, turbulence, windshear and sensor glitches and APs need to cope - the FG Shuttle AP can deal with the jumps of a teleportation when the apporach trainer resets to a new approach as well as brutally selecting a new weather situation while in air - so it is quite possible to write stable APs.

Just don't make your problems my problems by assuming that constantly complaining will get you a fix
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Advanced Weather debugging

Postby wkitty42 » Thu Aug 22, 2019 7:36 pm

@V12: don't you get those in csv format and then import them into ods? cvs is a lot smaller and easier to handle... especially when analyzing them from command line instead of a huge spreadsheet app...
"You get more air close to the ground," said Angalo. "I read that in a book. You get lots of air low down, and not much when you go up."
"Why not?" said Gurder.
"Dunno. It's frightened of heights, I guess."
User avatar
wkitty42
 
Posts: 9125
Joined: Fri Feb 20, 2015 4:46 pm
Location: central NC, USA
Callsign: wk42
Version: git next
OS: Kubuntu 20.04

Re: Advanced Weather debugging

Postby V12 » Fri Aug 23, 2019 6:55 am

Chart in the spreadsheet is reason, why I uploaded that file in the .ods format. Here https://ulozto.net/file/AxUccYR3fjjb/lsgs-lowi-csv is link for CSV file.
Fly high, fly fast - fly Concorde !
V12
 
Posts: 2757
Joined: Thu Jan 12, 2017 5:27 pm
Location: LZIB
Callsign: BAWV12

Re: Advanced Weather debugging

Postby Thorsten » Fri Aug 23, 2019 8:15 am

Well, now compare the data with the result of the algorithm worked out in a pocket calculator and see if you get the same values - that's how you debug math formulae.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Advanced Weather debugging

Postby ludomotico » Fri Aug 23, 2019 11:54 am

SurferTim wrote in Thu Aug 22, 2019 12:58 pm:@ludomotico: I appreciate the weather lesson. I got that same lesson in 1969 when learning to fly helicopters with the US Army.


Nice! Flying helicopters close to the ground and between the mountains must be scarily awesome.
User avatar
ludomotico
 
Posts: 1269
Joined: Tue Apr 24, 2012 2:01 pm
Version: nightly
OS: Windows 10

Re: Advanced Weather debugging

Postby legoboyvdlp » Fri Aug 23, 2019 12:52 pm

SurferTim wrote in Thu Aug 22, 2019 4:11 pm:This is a pressure thing ADJUSTED FOR ALTITUDE. The altimeter setting at the base of the mountain and at the peak will be the same, unless there is a frontal passage expected.

Yep, however lower temperature for a start will cause the pressure to be somewhat different.
Regardless of the reason, the altimeter setting is almost always different at Mammoth and due to how interpolation works and how the METAR station loading works you'll always get a rapid rise (think of the difference between smoothing .10 inHg difference over seventy miles and .30 inHg over thirty miles!)
User avatar
legoboyvdlp
 
Posts: 7981
Joined: Sat Jul 26, 2014 2:28 am
Location: Northern Ireland
Callsign: G-LEGO
Version: next
OS: Windows 10 HP

Re: Advanced Weather debugging

Postby SurferTim » Fri Aug 23, 2019 1:02 pm

legoboyvdlp wrote in Fri Aug 23, 2019 12:52 pm:Regardless of the reason, the altimeter setting is almost always different at Mammoth and due to how interpolation works and how the METAR station loading works you'll always get a rapid rise (think of the difference between smoothing .10 inHg difference over seventy miles and .30 inHg over thirty miles!)

The problem is it doesn't happen in FG over 30 miles. It happens in one update.

But that doesn't seem to matter if it can't be changed..
Fly low. Fly slow. Land on a dime. Twin Otter. https://github.com/SurferTim/dhc6p
My other aircraft is a Citation-X https://github.com/SurferTim/CitationX
PirateAir videos at https://www.youtube.com/user/SurferTim850
User avatar
SurferTim
 
Posts: 1689
Joined: Sun Dec 09, 2018 6:49 pm
Location: Miramar Beach, FL
Callsign: Pirate
Version: 2020.4.0
OS: Ubuntu 18.04

Re: Advanced Weather debugging

Postby Thorsten » Fri Aug 23, 2019 1:16 pm

The problem is it doesn't happen in FG over 30 miles. It happens in one update.


No, it does not - you never see the exact air pressure reported by the new station unless you go there, and you never see any change in one update.

Perhaps you're running different code - but the AW interpolation code is conceptually incapable of doing what you claim.

(That's the problem in this discussion - we can sit down, work out an example with a pocket calculator, see that the code does interpolate - and then comes the next forum post claiming an instant jump update).
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Advanced Weather debugging

Postby SurferTim » Fri Aug 23, 2019 1:19 pm

Thorsten wrote in Fri Aug 23, 2019 1:16 pm:
The problem is it doesn't happen in FG over 30 miles. It happens in one update.


No, it does not - you never see the exact air pressure reported by the new station unless you go there, and you never see any change in one update.

Perhaps you're running different code - but the AW interpolation code is conceptually incapable of doing what you claim.

That is not what I am seeing, but whatever you say.

I am running the newest version 2019.2.0 freshly downloaded, compiled, and installed with the help of legoboyvdlp on a Ubuntu 18.04 box.
.
Fly low. Fly slow. Land on a dime. Twin Otter. https://github.com/SurferTim/dhc6p
My other aircraft is a Citation-X https://github.com/SurferTim/CitationX
PirateAir videos at https://www.youtube.com/user/SurferTim850
User avatar
SurferTim
 
Posts: 1689
Joined: Sun Dec 09, 2018 6:49 pm
Location: Miramar Beach, FL
Callsign: Pirate
Version: 2020.4.0
OS: Ubuntu 18.04

Re: Advanced Weather debugging

Postby Thorsten » Fri Aug 23, 2019 1:21 pm

That is not what I am seeing, but whatever you say.


Then log at 120 Hz (FDM rate) and show me an example of an instant jump from old to new air pressure. Or look into the code and explain how it should perform this miracle.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Advanced Weather debugging

Postby SurferTim » Fri Aug 23, 2019 1:44 pm

Thorsten wrote in Fri Aug 23, 2019 1:21 pm:
That is not what I am seeing, but whatever you say.


Then log at 120 Hz (FDM rate) and show me an example of an instant jump from old to new air pressure. Or look into the code and explain how it should perform this miracle.

I got the message. No problem here. I am hallucinating the radical altitude changes. Maybe a flashback? ;)

I did learn a very important lesson taught to me by my second company commander in Korea. If you know the person in charge has already stated he isn't changing anything, why go through the motions? Just say "yes, sir", strap on your aircraft, and fly.
Fly low. Fly slow. Land on a dime. Twin Otter. https://github.com/SurferTim/dhc6p
My other aircraft is a Citation-X https://github.com/SurferTim/CitationX
PirateAir videos at https://www.youtube.com/user/SurferTim850
User avatar
SurferTim
 
Posts: 1689
Joined: Sun Dec 09, 2018 6:49 pm
Location: Miramar Beach, FL
Callsign: Pirate
Version: 2020.4.0
OS: Ubuntu 18.04

Re: Advanced Weather debugging

Postby Thorsten » Fri Aug 23, 2019 2:03 pm

I am hallucinating the radical altitude changes.


I would not know about that. But this

The problem is it [the pressure update] doesn't happen in FG over 30 miles. It happens in one update.


is a very specific claim that the full pressure change between old station and new station happens in less than 1/120 of a second (that is one FDM update cycle).

It's not a claim that the change can be 'rapid' or 'radical' - not even a claim that there can in principle be jumps - but a claim that the new pressure is instantaneously applied.

Note that this scenario is very different from e.g. a rapid change across 5 seconds as claimed by V12 - 5 seconds are 600 FDM iterations, instantaneous is one (!) FDM iteration - so in the first case the code has a few hundred times less stress to cope with any change.

1/120 of a second is far too small to estimate by eye-balling, so you could not possibly know whether that actually happens without a log file obtained at high sampling rate.

If you just want to talk, you can be as imprecise as you like - but if you want to debug software or report issues, you need to learn to be precise about what the issues actually are - you can't simply magically turn an eye-balled observation into something much more precise.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Advanced Weather debugging

Postby Thorsten » Fri Aug 23, 2019 2:41 pm

I did learn a very important lesson taught to me by my second company commander in Korea. If you know the person in charge has already stated he isn't changing anything, why go through the motions?


Alas - that lesson has left you woefully unprepared for the OpenSource environment where there is no person in charge - you can decide to be interested in the problem, take the code, modify it to your needs and submit it for inclusion in FG - and then face the wrath of users who don't like your new feature.

There's in fact multiple things that could be done on the data front (pre-fetching METAR much earlier for starters, including aloft wind data as lego has done it, use an offline model for the aloft winds as a proposal a while ago had it,...). It just has to be done by someone who is interested in the feature and is willing to spend half a year with it to iron out the kinks.

Well, I'm not interested (for the reasons mentioned) - but if you change perspective and look at my involvement in this thread, I've tried over and over to help you see what the issues really are, what the relevant code parts are, how to approach such an issue - you're just too happy to complain in the hope that someone else will fix it for you and stubbornly refuse to really understand the issue to notice any of that.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: Advanced Weather debugging

Postby SurferTim » Fri Aug 23, 2019 4:32 pm

Thorsten wrote in Fri Aug 23, 2019 2:41 pm:
I did learn a very important lesson taught to me by my second company commander in Korea. If you know the person in charge has already stated he isn't changing anything, why go through the motions?


Alas - that lesson has left you woefully unprepared for the OpenSource environment where there is no person in charge - you can decide to be interested in the problem, take the code, modify it to your needs and submit it for inclusion in FG - and then face the wrath of users who don't like your new feature.

If you think you are the subject of my wrath, you misunderstand my motive.

That lesson has left me well prepared for modern society. You change what you can, the rest...

There is always a person in charge, even open source, or you would not need to submit the change.

I am NOT COMPLAINING! It is a great program. I just spent several days getting 2019.2.0 installed in my Linux box. If it crashes me once a month, so be it. I can handle that.

As I used to tell my ex-boss: "Let me check my giveashitometer...Nope, nothing."
Fly low. Fly slow. Land on a dime. Twin Otter. https://github.com/SurferTim/dhc6p
My other aircraft is a Citation-X https://github.com/SurferTim/CitationX
PirateAir videos at https://www.youtube.com/user/SurferTim850
User avatar
SurferTim
 
Posts: 1689
Joined: Sun Dec 09, 2018 6:49 pm
Location: Miramar Beach, FL
Callsign: Pirate
Version: 2020.4.0
OS: Ubuntu 18.04

PreviousNext

Return to Weather

Who is online

Users browsing this forum: No registered users and 3 guests