A few questions about MapServer WCS

[Frank Warmerdam]

Rahkonen Jukka wrote:
> I made a test and I can affirm that "PROCESSING BANDS" in layer 
> definitions is just discarded in WCS.

Jukka,

More properly, we could say that it is overridden.  :-)

>> Resampling types "NEAREST" and "BILINEAR" should be support 
>> via WCS but have to be passed in via the WCS request.  I 
>> don't think you can establish defaults in the mapfile.
> 
> MapServer WCS DescribeCoverage is listing nearest neigbour and
> bilinear as supported interpolation methods so this should really
> be possible.  From an OGC document
> 05-076_Web_Coverage_Service_WCS_Version_1.0.0_Corrigendum.pdf
> I found that different interpolations should be asken in WCS 
> request with key INTERPOLATION = interpolation-method.  I tried 
> to add &interpolation=bilinear to my request. Request was accected
> and I received three-band image as I was ordering, but it appears
> like greyscale image on screen and I believe that all three 
> channels are actually the same. I put my requests in the end of
> this e-mail so somebody can check if there is something wrong 
> it.

Hmm, this sounds a bit like a problem that might have been fixed
already.  What version of MapServer are you using?  If it is
recent (ie. 4.10.3 or 5.0) I'd appreciate your filing a bug to
my attention with details to reproduce the problem, including a
data file, mapfile and request url.

>> I suspect the LUT operations are processed only in the image 
>> modes other than "pure data".  That is, IMAGEMODE BYTE likely 
>> precludes their use as things stand.  I could look at 
>> altering this if you wanted.
> 
> I do not have any immediate use for that feature. But if somebody
> is utilising LUT operations and keeps unadjusted images on server 
> and delivers them through LUT processing to WMS, then images 
> downloaded through WCS service would not look the same. But anyhow,
> LUT processing works perhaps only for 1 of 3 band imagery and thus
> it would not be useful for multichannel images in any case.

Part of the problem is that "pure data" modes like IMAGEMODE BYTE
deliberately try to avoid doing any of the usual MapServer processing
that would alter pixel values.  Stuff like classification, stretching
and LUTs.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | President OSGeo, http://osgeo.org