I just want to clarify that my answer below is specific to the THREDDS
Data Server's (TDS) implementation of WCS. Other WCS implementations
will behave differently and advertise different capabilities.
On 7/13/2010 12:23 AM, Joshua Vote wrote:
> I was wondering what datapoints get returned when making a Bounding Box
> request to a WCS
> Say I had a dataset referenced by EPSG:4326 (longs/lats) with a
> datapoints distributed at every whole longitude/latitude value and I
> wanted to make a Bounding box request. What would happen if I made a
> bbox request like so...
> Would the resulting Geotiff be generated from the data points (11,11) ->
> (20, 20)?
> Likewise, if I then take the same request and add...
> Would I be getting the exact same dataset as before but now it's been
> interpolated over 256 points on each axis instead of just 10?
A few details about the WCS implementation in the TDS:
1) Currently, the TDS only supports WCS 1.0.0
2) The TDS WCS does not perform any interpolation.
This is indicated in the DescribeCoverage documents
3) The reason for #2 is that the TDS WCS only returns
data points actually contained in the backing dataset,
in the CRS in which it is stored. This is, very loosely,
indicated in the DescribeCoverage document with
NOTE: It is on our list to work on more explicitly
representing the actual CRS in which the data is stored
4) The TDS WCS also, currently, only support requests in
a single CRS (OGC:CRS84, now urn:ogc:def:crs:OGC:1.3:CRS84).
Also described in the DescribeCoverage document:
5) The dataset returned in the TDS WCS response, in
either supported format (netCDF and GeoTIFF), contains
information describing the CRS of the data.
So, given that the request and response CRSs are not necessarily the
same, we extract the data for the response by mapping the four corners
of the bounding box into the data CRS and taking the minimum box in the
data CRS that contains all four points.
Hope that helps,
> Josh Vote