Re: [Flightgear-devel] Scenery 0.9.10 First Impressions

[Paul S. <su...@te...>]

On Friday 20 January 2006 00:13, Robicd wrote:
> Well, I make use of Google Earth and atlanteitaliano.it for getting
> aerial pictures and populate the town around LICP, both sources give
> same coordinates so I assume the offset is real.
>
> Do you think I can take sources like GoogleEarth as correct references
> for coordinates? If not, do you suggest searching elsewere (in that case
> I will have a bad time positioning the buildings)?


If you get some decent georeferenced and ortho corrected imagery and load it 
in a GIS viewer you can read the co-ordinates off that.
I'm suspicious of viewing imagery with a web browser because I've seen it 
sometimes distorted due to the browser flattening projected imagery or maybe 
because the companies involved add a bit of inaccuracy to make sure people 
can't use it without purchasing it. (Like the civilian GPS issue where they 
add in some random inaccuracy.)

You can get 14.25 meter/pixel panchromatic (black and white) Landsat 7 ETM+ 
data from here http://glcfapp.umiacs.umd.edu:8080/esdi/index.jsp (GLCF: Earth 
Science Data Interface) and load it in QGIS, OpenEV, etc. and read the values 
off that. The panchromatic bands are the *nn80* files.
The positional accuracy of Landsat 7 ETM+ data should be about 50m RMS or 
better which is about the width of a runway so it should be good enough.

I corrected the positions of 484 airports around Northern Italy, South Western 
France, all of Switzerland, all of Austria and Southern Germany by using this 
method (entire Alps range). There are lots of airfields in the X-Plane 
database that are off by several kilometers including major airports like 
Sion (2km).
There was an airfield in France that was off by more than 10km - it was on the 
wrong side of a river and took me about 10 minutes to find it and ID it.  :)
Unfortunately I don't store all the airport info so I can't submit it back to 
Robin's DB without a bit of work involved.

Paul