[gdal-dev] warp from ch1903 to google maps(mercator) projection - what t_srs to use

[Vasile Cotovanu]

Hello all, this is my first post on this list, first of all I want to
congratulate anybody involved in this fantastic library !

I have an issue that is fixed with a workaround but I want to avoid this
workaround, here it is :

I'm using gdalwarp to reproject imagery from EPSG:21781 (CH 1903) to Google
Maps (mercator based) projection.

Currently I have been using for -t_srs the suggestion from this blog entry
http://www.sharpgis.net/post/2007/07/26/The-Microsoft-Live-Maps-and-Google-Maps-projection.aspx

so

+proj=merc +a=6378137 +b=6378137 +lat_ts=0.0 +lon_0=0.0 +x_0=0.0
+y_0=0 +k=1.0 +units=m +nadgrids=@null +no_defs

which defines a perfect sphere(not ellipsoid) but i am not happy with
the results because :
- if i am transforming the bounds of image from CH1903 to WGS84 and
then to Google Maps pixel coordinate system, the result will be a
parallelogram (it is normal, the image will be distorted from CH1903
to Mercator) that can be included in a box with a pair of width/height in
pixels
- if i am using gdalwarp the result will be a distorted image (again
it's normal) with another pair of width/height.

The expected result after tis 2 operations would be 2 pairs of width/height
WITH SAME ration width/height. However, this issue can be fixed by FORCING
the gdalwarp "-ts width
height" parameter and imagery it's just drawn fine.

So my question is, in order not to do this workaround, can you tell me
if you use another -t_srs value, and if yes, which one ?

I suppose that should be different for each zoom level, as GoogleMaps does
use different radius for each zoom level. (on zoom level 0 the imagery is
drawn in a single tile 256x256 pixels, so the radius will be
circumference/(2*pi()) pixels , on zoom 1 : circumference is 512 px, then
radius is 2 times bigger than zoom level 0, etc)

Thank you in advance,
Vasile Cotovanu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/gdal-dev/attachments/20080215/53ad75b4/attachment.html