I believe the weather reports are time stamped so we could ignore reports that are older than "x" hours. I don't think we are currently doing that.

Curt.


Chris Metzler wrote:

On Wed, 29 Dec 2004 13:16:48 -0600
Curtis L. Olson wrote:





I know different people will have different opinions on this, but I feel
that simply interpolating over time to the "closest" data is just as good as anything. Interpolating spacially between the closest 3 stations is attractive, but remember this data is already starting to get old by the time we get it so we will never be exactly correct with current conditions.


[ snip ]


Personally, I think it would be a *lot* simpler, and arguably just as accurate to do a temporal interpolation towards the latest data at the closest weather station.



Another issue is the fact that the data available from the METAR station seems sometimes *very* old (e.g. a day or more). I've flown (in FlightGear) around a metropolitan area where I know exactly what the weather's like (e.g. clear skies), and found it change from roughly correct weather to something *wildly wrong* (e.g. overcast down to 900 feet, which it was like earlier in the week but definitely not today) to something correct again (back to clear skies) as I fly 5 miles in a straight line over the metro area. So instead of spatial interpolation, one might consider weighted spatial averaging (e.g. a gather scheme with a broad Gaussian kernel or whatever) to lessen the effect of anomalous stations in densely sampled areas.




Lot's of fun to be had if someone had the time to play with it ...



I used to do stuff that bears some similarities to this for a living. Unfortunately, it was in FORTRAN. Heh.

-c



------------------------------------------------------------------------

_______________________________________________
Flightgear-devel mailing list
Flightgear-devel@flightgear.org
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d



--
Curtis Olson        http://www.flightgear.org/~curt
HumanFIRST Program  http://www.humanfirst.umn.edu/
FlightGear Project  http://www.flightgear.org
Unique text:        2f585eeea02e2c79d7b1d8c4963bae2d


_______________________________________________ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d

Reply via email to