Many thanks for your suggestions Marcin. I think you have set me on
the right lines - I'll try warping the tif and report back.
Thanks again,
Philip
Sent from my PDA
-----Original Message-----
From: Marcin Rudowski <[email protected]>
Sent: 22 February 2010 19:36
To: [email protected] <[email protected]>
Subject: Re: [Mapnik-users] Raster layer position error mystery
W dniu 22.02.2010 19:12, Philip Howarth pisze:
> Hi
>
> I am using the Raster symbolizer to include a colour relief file as
> background.
>
> I can generate the georef'd background files from either SRTM3V4 tifs (5
> degree squares) or ASTER tifs (1 degree squares) using exactly the same
> python script.
>
> Viewing the resulting files in QGIS or other tools - the coordinates of
> boundaries and features are correct.
>
> Mapnik produces maps when I use either the ASTER or the SRTM derived tif
> as background. My problem is that the SRTM derived background is
> displaced by about 0.06 degrees to the North. Interestingly, East/West
> positioning appears to be correct.
>
In Your case, I'm not sure how it works, that You are using WGS84 image
and passing to mapnik coordinates in mercator. If raster is WGS84 and
You generate map in Mercator, then it needs to be warped at some point
and I believe mapnik can't do that with rasters.
Maybe Your problem is very simple: You are using raster in wrong
projection not matched with one used in mapnik. In effect on upper and
lower raster edge, everything should be quite fine, but in the middle,
there should be some misalignment in north/south direction. So You need
to transform Your tiff to Mercator before using in Mapnik, and then it
will be visible even when using gdal plugin. Now I thing it tries to
draw raster near: 10,50 but in mercaator, and this is near Africa.
Beside that, I had problem with strange offset when first started with
hillshading, but in my case at ~50 deg North it was: ~20km (0.2deg).
After some time I found out, that this kind of error is connected with
ellipsoid reprojection.
SRTM is in WGS84 that uses ellipsoid (6378137,298.257223563) and
spherical Mercator is using sphere, so besides simple converting
coordinates gdal also did ellipsoid->sphere reprojection and this gave
offset in south/north direction. So in short it works like that:
1. source coordinates in degrees converted to XYZ using WGS84 ellipsoid
2. XYZ converted to degrees using sphere parameters
3. degrees converted to Mercator
My gdal_warp magic command, to avoid such problem was:
> gdal_merge.py -v -o ../srtm.tif *.hgt
> gdalwarp -of GTiff -srcnodata 32767 -dstnodata 32767 -s_srs \
> "+proj=latlong +a=6378137 +b=6378137 +lat_ts=0.0 +lon_0=0.0 \
> +x_0=0.0 +y_0=0 +k=1.0 +units=m +nadgri...@null +no_defs +over" \
> -t_srs "+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 +nadgri...@null +no_defs +over" \
> -rcs -order 3 -tr 76.4370282852 76.4370282852 -wt Float32 -ot Float32 \
> srtm.tif warped.tif
> ./hillshade warped.tif hill_8.tif -z 2
So I force srtm to be treated as simple latlon on same sphere as output
mercator, to avoid step 1 and 2.
But it is quite different from Your case. I describe it, because it is
still quite confusing for me.
Some links:
http://lists.maptools.org/pipermail/proj/2007-March/002852.html
http://www.mail-archive.com/[email protected]/msg00564.html
http://www.mail-archive.com/[email protected]/msg01051.html
--
Marcin Rudowski
_______________________________________________
Mapnik-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/mapnik-users
_______________________________________________
Mapnik-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/mapnik-users
