As long as WCS runs in "raw mode", the original pixel values
should be preserved. Of course, we come full circle to the fact
that rawmode does not currently support colortables. There is no
compelling reason it could not, but it isn't currently part of
the raw mode data model. I believe someone else has already created
a bug report on that, though I'm not promising if/when it will be
dealt with.
Ok, Thank. I'll also look at the code to try and contribute.
Ah. The problem is that MapServer uses PROJ.4 format as it's internal
representation of coordinate systems, and it is very bad at identifying
datums. So when you say epsg:4283 it is internally expanded to:
+proj=longlat +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +no_defs
When mapserver goes to write this to TIFF it is written as ellipsoid GRS80.
You might want to file a bug on this. It *might* be possible for me
to find the epsg:n value in cases like this and generate a proper
GeoTIFF definition based on that. This isn't normally an issue in
mapserver since it doesn't normally produced products with embedded
coordinate systems. WCS is somewhat special in this regard.
Strangely enough the i found that on one of my other installs of mapserver (I belive it was the Feb-2006 release of MS4W) this problem didn't exist and the Geotiff created by WCS request had the correct Coord Sys. While my current development server is running 'fgs' with the cvs head of mapserver and its showing this problem. I'll do a few more tests though before filing that bug.
Thanks for all your help on this issue thus far...
These issues are of general enough interest that I have taken the liberty
of cc:ing my response back to the list.
I was thinking of doing that :)
shoaib
