RE: Fwd: [GMLJP2] RE: GML, JPEG2K, and WCS

• To: "Gerry Creager N5JXS" <gerry.creager@xxxxxxxx>, <Ben@xxxxxxxxxxxxxxxx>
• Subject: RE: Fwd: [GMLJP2] RE: GML, JPEG2K, and WCS
• From: "Ron Lake" <rlake@xxxxxxxxxxxxx>
• Date: Mon, 19 Sep 2005 20:29:04 -0700

Hi,

See comments below.

Ron

I'll admit to not closely following the JP2 work, but I've still got a 
question:  Last I checked, JPEG compression was lossy.  Has this changed with 
JPEG2000, and if not, how do we reconcile lossy compression with recapturing 
the pixel values/contents?

For instance, I have a particular interest in using WCS for sea surface 
temperatures, as one example, and content rich weather radar data as another.  
If we invoke a lossy compression scheme, I lose data.

Am I concerned about a solved problem?

Note that JPEG 2000 offers both lossy and lossless wavelet compression.  (see 
http://public.migrator2000.org/newto.xalter)

Also: in the examples, you cite basic surface met observations data. We already move that quite nicely via WFS, and I'm not sure that particular example has bearing.

How are you doing this? Using GML Observation features?  That is what they are for.  Of course 
using observations can work even for gridded data - since the <resultOf> property of the 
<Observation> can be a coverage.  Of course a collection of <Observations> can also 
be regarded as a coverage.

NOW: attempting to represent fine grid from a model where the output is the 
result of shading or contouring might be an example of a utilitarian 
implementation.  Pedantically speaking, of course ;-)

Thanks, Gerry