Board index FlightGear Development Weather

A unified weather system?

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

A unified weather system?

Postby Thorsten » Wed Jun 13, 2012 8:45 am

There has been talk on and off on the mailing list if we should have just one weather system in Flightgear which is customized by a series of options to run in different detail. We're considering doing that for a 3.0 release. What is needed however is some brainstorming. The technology is largely there, the question is more how to present it to a user and how to apply it.

So (without making any promise that it actually gets implemented that way), this thread is to gather some ideas and feedback: What would you like a weather system to be? In particular:

* What level of detail would you like to control? For instance, wind speed and direction are in general different at different locations as well as different in altitude - how would you like to set this? Should the system come with its own wind model, is there a need for the user to set up specific situations, ...?

* What level of control would you like to have over the environment runtime without changing the general weather situation? For instance, should it be possible to control turbulence by the user even if the modelled weather situation has a lot of turbulence to make things easier? Should it be possible to make fog disappear because you are lost and want to take a look around? Or should the environment rather be something one has to deal with without quick runtime access?

* How important is METAR parsing vs. offline weather modelling for you?

* Are there specific features of the environment which currently can't be controlled right now but should be made accessible?

* Are there weather effects which are currently missing but should be modelled better?

* Is there anything else (besides better framerates, we already know that one...) on your wishlist?

(Also, please note that in many cases there's a tension - more realism usually means less framerate. Better user-control in the GUI usually means a more complicated and less intuitive GUI. And so on. One has to be realistic.)
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: A unified weather system?

Postby islandmonkey » Thu Jun 14, 2012 7:03 pm

Well, for the GUI, I'd like a bit more of a less complex one (e.g. airmass amounts - what airmass? Cold airmass? Warm airmass?).

And yes, when I use METAR mode in advanced weather I'd like to see the raw METAR report.
Last edited by islandmonkey on Fri Jun 15, 2012 7:35 am, edited 1 time in total.
User avatar
islandmonkey
 
Posts: 786
Joined: Mon Jan 30, 2012 9:51 pm
Location: EGCN (uni), EGHI (home)
Callsign: G-MNKY
OS: Ubuntu 20.04

Re: A unified weather system?

Postby polly » Thu Jun 14, 2012 11:23 pm

First, thanks for the work already.
Level of Control: It's good if the drop down scenarios update wind and cloud tables which can then be tweaked before applying to the sim session. Perhaps the (Basic Weather) wind and cloud columns could be extended to include adjustments relevant to Advanced Weather, e.g. in addition to cloud layers, cloud types, type densities; and turbulence column in each wind level.
If the user could adjust wind and cloud after selecting a scenario and the scenario list includes 'Clear and Still' then individual wind scenarios could be set up. User-specified winds seem important for example specific cross-wind landings.
Turbulence: 0-9 selection in each Wind layer seems adequate.
Clear Fog ( cheat ) .. doesn't the 'Z' hotkey accomplish that ? It could be re-mapped to reduce all cloud density values in the cloud layer table.
METAR: Very important. I'm forever looking out the window and then comparing with what's on the screen. I like to see the effect of a METAR string on the cloud/wind tables. I like to be able to enter a METAR string and see the effect parsed into the tables. I use Live METAR most often of the scenarios.

General: I'd like to see Advanced scenarios included into what appears on the Basic screen as "METAR Source" to become one "Scenario" drop down with METAR string applied from the Scenario or User-Pasted. One slider to select "Fast" (Basic) or "Detailed" (Advanced). I'd like METAR data to show and to be parsed into cloud/wind whether it's selected as Basic, Advanced or pasted from another source.
Thanks !
User avatar
polly
 
Posts: 956
Joined: Thu Nov 04, 2010 3:45 pm

Re: A unified weather system?

Postby Thorsten » Fri Jun 15, 2012 8:31 am

Well, for the GUI, I'd like a bit more of a less complex one (e.g. airmass amounts - what airmass? Cold airmass? Warm airmass?).


I don't understand - which dialog/setting is that comment aimed at?

And yes, when I use METAR mode in advanced weather I'd like to see the raw METAR report.

I'd like METAR data to show and to be parsed into cloud/wind whether it's selected as Basic, Advanced or pasted from another source.


The talk has been to separate METAR into a different dialog so that it can be fetched and parsed and looked at before deciding to execute it.

One problem is that METAR can't be uniquely parsed in cloud and wind info, because it doesn't contain the relevant information.

Wind: A METAR string contains the wind on the ground at one location. That wind is almost certainly measured in the boundary layer, i.e. the wind at 1000 ft above the location may be up to a factor 2 stronger. At higher altitudes, the wind pattern may be completely different from what METAR has to say. In mountain regions, the ground winds will reflect the deflection of air by the terrain rather than the large-scale wind patterns. Without an online aloft wind pattern source, the higher-altitude winds based on METAR are just a random guess.

Terrain: A serious problem with manually entered METAR strings is that they don't contain the station altitude information. However, that is highly relevant because the actual cloud base is station altitude plus the cloud layer altitude reported in METAR. Not a problem in flat terrain since you can always use the local terrain elevation at aircraft position, but a problem in mountains.

Cloud info: METAR just tells you that there is n/8 cloud coverage at m feet. This doesn't specify what the clouds are (a 6/8 layer can be a Stratus with a visible edge, or a closing Stratocumulus development, and they look totally different) or how they're distributed - a 4/8 cloud cover can be an undulatus pattern, few large patches with gaps inbetween or many small individually separated clouds. Again, the system has to guess this.


It's good if the drop down scenarios update wind and cloud tables which can then be tweaked before applying to the sim session.


This only works for the Basic Weather setup: A gui to manually readjust the full cloud and wind information going on inside the Advanced Weather simulation is huge - the problem is that Advanced Weather changes that information in space as well as in time. You don't have one table of cloud configurations, you have one per weather tile (same with winds) which at startup would give you 9 (!) times the size of the current Basic Weather interface, i.e. you'd end up with a gui spread over several pages. I judged that impractical.

Even worse with turbulence - turbulence in Advanced Weather is often tied to locations (a thermal, a thunderstorm,...) - so you'd need to edit about 1000 different locations to set the turbulence distribution of a Cumulus sky manually.

Turbulence: 0-9 selection in each Wind layer seems adequate.


No, it isn't - way not realistic enough, turbulence usually is not a layer phenomenon, it is either a vertical air motion phenomenon, created by an obstacle or occurs at the boundary of air layers with different windspeeds.

Clear Fog ( cheat ) .. doesn't the 'Z' hotkey accomplish that ?


You don't fly with Advanced Weather, do you :-) The key doesn't work when Advanced Weather is running. If you want to make clouds vanish, the cloud visibility range in the rendering dialog does the trick nicely.

Polly: In summary, you basically would like a Basic Weather ++, i.e. not really any of the features of Advanced Weather (like location-specific weather effects of spatial changes of... anything) but rather more control over what comes in from METAR before it is applied to the Basic Weather interface, so METAR should be fetched, parsed and passed into the table, then you'd like to edit the table, then you'd like to run the edited version. Is this a fair represenation of what you said?
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: A unified weather system?

Postby Gijs » Fri Jun 15, 2012 8:50 am

I'd vote for splitting the dialogs in a basic one and an advanced one. Just like we did for the shaders. The average user ( including me on most of my flights) isn't interested in finetuning the weather. He just wants to choose between various scenarios ( live weather, minimum VFR, fair weather etc.).

Until the systems are really merged, there could be a radio button in the simple dialog to choose between the systems. Depending on that selection, the "advanced settings" button either brings you to the global weather dialog, or the advanced weather's.

Just some early morning brainstorming :-)
Airports: EHAM, EHLE, KSFO
Aircraft: 747-400
User avatar
Gijs
Moderator
 
Posts: 9544
Joined: Tue Jul 03, 2007 3:55 pm
Location: Delft, the Netherlands
Callsign: PH-GYS
Version: Git
OS: Windows 10

Re: A unified weather system?

Postby Bjoern » Fri Jun 15, 2012 3:06 pm

Gijs wrote in Fri Jun 15, 2012 8:50 am:I'd vote for splitting the dialogs in a basic one and an advanced one. Just like we did for the shaders. The average user ( including me on most of my flights) isn't interested in finetuning the weather. He just wants to choose between various scenarios ( live weather, minimum VFR, fair weather etc.).


A basic selection box...I like that.

"Source:

Live METAR or User defined or Scenario"

If...
...the "live data" option is selected, display a box with nearest METAR data
..."user defined" is selected, display an input panel
..."scenario" is selected, display a scenario selection panel

I think MSFS does it just about the same way.
Bjoern
 
Posts: 484
Joined: Fri Jan 06, 2012 11:00 pm
Location: TXL (RIP)
Version: Next
OS: ArchLinux

Re: A unified weather system?

Postby islandmonkey » Fri Jun 15, 2012 3:31 pm

Thorsten wrote:I don't understand - which dialog/setting is that comment aimed at?


Note a slider at the bottom of the advanced weather panel labelled airmass. Now I realise that cold and warm air is controlled by the slider that says 'rough day' and 'low convection'. But that is unclear to the general new user.
User avatar
islandmonkey
 
Posts: 786
Joined: Mon Jan 30, 2012 9:51 pm
Location: EGCN (uni), EGHI (home)
Callsign: G-MNKY
OS: Ubuntu 20.04

Re: A unified weather system?

Postby polly » Fri Jun 15, 2012 3:56 pm

I just spent three hours honing a reply and lost it, ah well.
I think I was long windedly saying 'What Gijs says' .. I like all the things Advanced Weather does, it's most impressive but the way the menus are set up, there are three different drop boxes that affect scenario where there should be one, followed by a big Basic vs Advanced ( Fast + Global vs Complex + Local ) slider.
My other thought was that Basic does a good job of taking a scenario and showing on the cloud / wind lists how the guesses have been made in terms of cover and variance at layer altitudes. I don't need to be told that Advanced makes far more complex decisions to that effect, I suspected as much from the superb appearance and the frame rates; I'm thinking if Advanced could extend the cloud / wind lists with as much detail as it thinks the user can handle then it would make me a better man. I'm thinking, if it guesses Stratus when METAR says 8/8 at 250 ft or Cirrus for 1/8 at 18,000 ft then maybe a column showing [ St, Cm, Cb, St, Dynamic, Local ... etc] and for turbulence [ None, Lght, Mod, Strng, Svre, Bndry, Thrml .. etc ] i.e anything to show me what to expect on the screen.
METAR .. Meteorological .. Weather .. the selection is in the right place on the weather menu and is handled well in Basic Weather .. you select Live METAR or Paste a METAR string ( from the ghoul forum or wherever ) .. the string gets parsed and you get a freebie translation into cloud and wind setups that both educate yield a very believable effect. Why would you trash all that benefit to the user just because Advanced Weather does the wrong thing by declaring "METAR Disabled" and METAR string "NIL" and further persecute the innocent by removing METAR from the Basic Weather menu ?

Thanks.
User avatar
polly
 
Posts: 956
Joined: Thu Nov 04, 2010 3:45 pm

Re: A unified weather system?

Postby islandmonkey » Fri Jun 15, 2012 6:13 pm

polly wrote:from the ghoul forum or wherever


Never knew ghosts could be meteorologists :lol:
User avatar
islandmonkey
 
Posts: 786
Joined: Mon Jan 30, 2012 9:51 pm
Location: EGCN (uni), EGHI (home)
Callsign: G-MNKY
OS: Ubuntu 20.04

Re: A unified weather system?

Postby Thorsten » Fri Jun 15, 2012 7:38 pm

it's most impressive but the way the menus are set up, there are three different drop boxes that affect scenario where there should be one, followed by a big Basic vs Advanced ( Fast + Global vs Complex + Local ) slider


*sigh* You're basically thinking Basic Weather and asking 'Why isn't Advanced Weather structured the same way?'

Because... the philosophy is different. Basic Weather generates the wind and cloud structures based on user input and it allows for inconsistent input - you can select things which would never ever happen in reality, like few fluffy clouds in a low pressure zone.

Advanced Weather has (limited) understanding of what the weather situation is, and it builds all elements of the weather scene based on that understanding, so you can't get for instance pressure inconsistent with visible clouds. For example, the location of the lowest cloud layer has a deep meaning for the system, it's the location of the lowest inversion layer below which dust and moisture from the terrain exist, whereas above they do not. Visibility and turbulence adjust accordingly. For that reason, you can make it configurable on the physics level where you specify different airmasses and their properties, but not on a level where you select turbulence, clouds and winds independently.

Also, it doesn't aim at a global setting - basically a lot of what you could input at startup would be obsolete once you fly 20 km and reach the next tile. That's also the reason why you have two dropdown boxes influencing the weather scenario - one governs the behaviour at startup, the other the behaviour once you leave the starting position. You can't expect this to fit into the same GUI as Basic Weather.

Why would you trash all that benefit to the user just because Advanced Weather does the wrong thing by declaring "METAR Disabled" and METAR string "NIL" and further persecute the innocent by removing METAR from the Basic Weather menu ?


Advanced Weather neither declares METAR disabled nor the METAR string 'nil'... it has a working METAR mode and the string is fetched and parsed, it's just not shown in the GUI.

The rational for creating a separate menu item is that elements common to both approaches should be factored out. The translation into cloud configurations works rather different - Advanced Weather parses the METAR by first trying to understand what the overall airmass is like, and based on that info decides to generate clouds on a random per-tile basis, factoring in the terrain, so the decision is made in a different way for every chunk of weather you create. That's different in Basic Weather which makes one decision per METAR and sticks with the result till the next METAR comes in.

I acknowledge that Basic Weather ++ is what you'd like to run, but I can't see how Advanced Weather could possibly run with the GUI you have in mind. Think that you have to rebuild all elements of the weather every 40 km - how do you want to input that for a 1500 km flight from a GUI?
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: A unified weather system?

Postby polly » Sat Jun 16, 2012 5:30 pm

Try it this way: If it's true that Advanced METAR option sets initial weather parameters from the current local METAR string and thence generates weather, not from updated new local METAR messages but from the internal 'Advanced' engine then:
1) METAR option in Advanced should be in drop down: "Select Initial Weather Scenario" not in drop down: Tile Selection Mode.
2) That accepted then 'Basic' and 'Advanced' Initial scenarios may be unified into one superset drop-down. If you're prepared to add buttons for Core/Border then Advanced-High Pressure can co-exist with Basic-Fine, Advanced-Low could almost co-exist with Basic-Marginal VFR etc. There's sufficient line width in the drop downs to add needed explanations like: "Simple Mode: Marginal VFR High Detail Mode: low pressure system
Selection between Basic and Advanced is done according to the selection (4) , below.
3) METAR settings in the above initialisation drop-down may be: Live Data - Pasted - etc any of which will be parsed and used as _initial_ conditions by either engine as selected (below) with the METAR field available as currently in Basic.
4) Below that drop down is a 'Weather Evolves" drop down which contains options which determines whether Basic or Advanced engine is engaged. Drop-Down entries could be:
a) Static Global weather ( Low-Compute Weather )
b) METAR update from nearest station ( Low-Compute Weather )
c) Authentic Weather Develops in local tile only ( High Detail Weather )
d) Authentic Weather Develops with tiles repeated ( High Detail Weather )
e) Realistic Weather ( High Detail Weather ) [ the menu could be more helpful here ]
5) The METAR string is preserved but labeled as 'Staring Condition Only - No Local Updates' as long as the Advanced engine is active
6) It would be highly desirable if the Advanced engine, with its sophistication, were able to give at least some clue about its intended cloud and wind conditions, even if the Cloud/Wind list were marked "Approximate Starting Conditions"
7) The remainder of the present 'Advanced Menu", without startup and Evolution drop-downs, would be appropriate to the 'Advanced' ( or even "High Detail Refinements " ) sub-menu

I hope it's obvious, the above does not reflect my thinking regarding what each mode does or does not do; rather I'm presenting as someone who has become familiar with the weather engines' function solely though using the menus sharing the desire for a consistent interface.
User avatar
polly
 
Posts: 956
Joined: Thu Nov 04, 2010 3:45 pm

Re: A unified weather system?

Postby f-ojac » Sat Jun 16, 2012 6:10 pm

My way of thinking it would be exactly the way Gijs presented his own vision.
f-ojac
 
Posts: 1304
Joined: Fri Mar 07, 2008 10:50 am
Version: GIT
OS: GNU/Linux

Re: A unified weather system?

Postby Thorsten » Sat Jun 16, 2012 7:20 pm

Polly: You misunderstood.

METAR is (as I said above) incomplete information. Even taking atmosphere physics into account, there are many possible cloud and wind configurations matching a single METAR report. Basic Weather generates one of them and sticks with it till the next METAR report comes in, Advanced Weather generates one of them every tile (40x40 km area) till the next METAR report comes in. This gives you a much more natural looking sky, as cloud types and distribution isn't homogeneous but changes dependent on where you look.

Think this way: Basic Weather configuration centers around the effects of weather which the user can adjust, Advanced Weather configuration centers around the causes and generates the effects as consistently as it can once it understands the causes. That's the real fundamental problem your GUI has to address.

Also, currently METAR parsing is done independently of weather engine. Advanced Weather doesn't know if a current METAR is fetched, entered by a user or coming from any other source. It just works with parsed METAR information. So there is actually not an issue here.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: A unified weather system?

Postby polly » Sat Jun 16, 2012 8:09 pm

Both engines derive a picture from a METAR text. Both engines can derive a new picture from an updated METAR text. Both engines can use a text from a live METAR or from pasted / user-entered field. Both engines choose just one parameter set from from many that could match the METAR text.
Starting one engine, METAR text is hown for the user to compare the picture to the METAR text. Starting the other engine doesn't allow the text to show so the user can't compare the METAR text to the picture. One engine shows a list representing the parameter subset it has selected, the other engine makes a choice but doesn't expose even a hint. There are two places on the menus where initial starting conditions are selected; there only needs to be one. Refactoring the menus as I suggested above takes none of the function away, consolidates initial conditions for both engines, presents a unified interface to both levels of function and could educate the new user to these possibilities.
User avatar
polly
 
Posts: 956
Joined: Thu Nov 04, 2010 3:45 pm

Re: A unified weather system?

Postby stuart » Sat Jun 16, 2012 10:31 pm

One random thought I had on a Saturday night after a glass of wine:

What would happen if we made Advanced/Local weather the default, and offered what is currently known as Basic Weather as an alternative for those wanting to have specific control of cloud bases etc?

As I understand it, it would mean that the simulated weather wouldn't match 100% the METAR report, but we'd gain broadly more realistic weather masses, cloud placements, etc. Perhaps for more virtual fliers that's an acceptable compromise?

From personal real life flying experience, I don't expect the METAR to exactly match what I encounter. It's purely guidance from a sample taken at 15 minute intervals at the airport itself, and not usually relevant 10 miles downwind.

Doing this, we could simplify the GUI as follows:

1) Weather dialog allowing setting of Live METAR or specific scenarios. Uses Local Weather.
2) Weather System settings containing the more detailed Local Weather controls (tile types, airmass type)
3) Weather Settings dialog allowing users to switch to Basic Weather and have control of specific cloudbases.

The major problem I can see with this (other than convincing the majority of developers) is whether the Local Weather offers a reasonable experience with 2D clouds.

-Stuart

-Stuart
G-MWLX
User avatar
stuart
Moderator
 
Posts: 1629
Joined: Wed Nov 29, 2006 10:56 am
Location: Edinburgh
Callsign: G-MWLX

Next

Return to Weather

Who is online

Users browsing this forum: No registered users and 2 guests