WCS primer

In this primer we will demonstrate how to obtain a small set of data from a large datasets with the OGC Web Coverage Service (WCS) protocol. We will show that ordering data through WCS is just as easy as buying an ice cream. Bear in mind the construction of a WCS request depends on the version supported by the data provider. Currently the versions are 1.0.0, 1.1.1 and 2.0.1. 

Find a WCS server

Find a data web source that hosts a WCS server (go to an ice cream vendor). We will use the Shuttle Radar Topography Mission datasets hosted at the Woods Hole Institute THREDDS OPeNDAP server as example in this primer. THREDDS OPeNDAP server can be configured to serve WCS for orthogonal datasets with proper geographic information. Here is a list with WCS servers. For the Netherlands search the PDOK WCS services for Nertherlanda, for instance AHN25.

Request an overview of the content of a WCS server

Ask for what the server has to offer (see which flavours he has and which kind of cups). You need to add the following mandatory <keyword,value> pairs to the base server url, separated by an &, e.g.: ?service=WCS&request=GetCapabilities.

This procedure works for all subsequent <keyword,value> pairs in this primer:
http://geoport.whoi.edu/thredds/wcs/bathy/srtm30plus_v6?service=WCS&request=GetCapabilities

This url will return an xml file that contains an inventory of the available datasets. (You can also request available datasets for one WCS version only by appending the optional version keyword.)

Example for 2.0.1 request on RWSProjectArchive geoserver.

https://rwsprojectarchief.openearth.nl/geoserver/mepduinen/wcs?service=WCS&version=2.0.1&request=GetCapabilities

Inspect the overview of the content of a WCS server

Look at what versions of WCS the server has to offer (check whether the ice cream is fresh). For each version there is a
tag WCS_Capabilities with attribute version, e,.g.

<WCS_Capabilities version="1.0.0">
...
</WCS_Capabilities>

For each version a number of datasets is hosted. Select one dataset (ice cream flavour) from the list through the ContentMetadata tag. The name of the dataset is the name tag, in this case there is only topo.

<ContentMetadata>
<CoverageOfferingBrief>
<description>topo meters true  Topography</description>
<name>topo</name>
<label>Topography</label>
...
</CoverageOfferingBrief>
</ContentMetadata>

Request some content of a WCS server

Now we can actually get a subset from the dataset want by using request=GetCoverage instead of the request=GetCapabilities we used above to obtain the meta-data (order ice cream). The following <keyword,value> pairs are mandatory.

These keywordd return data in the native resolution, some servers demand specification of a target grid:

Unlike WMS, to known which encoding formats are available, there is a separate request value DescribeCoverage
http://geoport.whoi.edu/thredds/wcs/bathy/srtm30plus_v6?service=WCS&request=DescribeCoverage&version=1.0.0
which is available when the request=GetCapabilities contains

<Capability>
...
<DescribeCoverage>
...
<OnlineResource xlink:href="http://geoport.whoi.edu/thredds/wcs/bathy/srtm30plus_v6" />
</DescribeCoverage>
..
</Capability>

In this case request=DescribeCoverage offers three formats

...
<supportedFormats>
<formats>GeoTIFF</formats>
<formats>GeoTIFF_Float</formats>
<formats>NetCDF3</formats>
</supportedFormats>
...

For 2.0.1 request this is:

https://rwsprojectarchief.openearth.nl/geoserver/mepduinen/wcs?service=WCS&version=2.0.1&request=DescribeCoverage&CoverageId=SpanjaardsDuinT0_50cm

Important parameters for subsetting are described in the output. Check for srsName and axisLabels. These need to be used in the getCoverage request.  

Summary

correct WCS request for two available formats are given below. You can copy these into your browser to obtain the data. We chose the same bounding boxes as in the OPeNDAP primer that access these same datasets via netCDF libraries in Matlab, python or R. We discussed the pros and cons of WCS vs OPeNDAP in a FOSS4G paper.

http://geoport.whoi.edu/thredds/wcs/bathy/srtm30plus_v6?request=GetCoverage
&version=1.0.0
&service=WCS
&format=geotiff
&coverage=topo
&BBOX=0,50,10,55

http://geoport.whoi.edu/thredds/wcs/bathy/srtm30plus_v6?request=GetCoverage
&version=1.0.0
&service=WCS
&format=netcdf3
&coverage=topo
&BBOX=0,50,10,55

for which the netCDF file look like

NetCDF-3 Classic srtm30plus_v6_netcdf3.nc {


dimensions:
	lat = 601 ;
	lon = 1201 ;

variables:
	int16 topo(lat,lon), shape = [601 1201]
		topo:_FillValue = -9999 s
		topo:grid_mapping = "GDAL_Geographics"
		topo:long_name = "Topography"
		topo:units = "meters"
		topo:coordinates = "lat lon lat lon"
	double lat(lat), shape = [601]
		lat:units = "degrees_north"
		lat:long_name = "Latitude"
		lat:_CoordinateAxisType = "Lat"
		lat:standard_name = "latitude"
	double lon(lon), shape = [1201]
		lon:units = "degrees_east"
		lon:long_name = "Longitude"
		lon:_CoordinateAxisType = "Lon"
		lon:standard_name = "longitude"

//global Attributes:
		:Conventions = "CF-1.0"
		:AREA_OR_POINT = "Area"
		:title = "UCSD SRTM30_v6 Global DEM (30 sec)"
		:History = "Translated to CF-1.0 Conventions by Netcdf-Java CDM (NetcdfCFWriter)
Original Dataset = bathy/srtm30plus_v6; Translation Date = Tue Nov 08 05:16:29 EST 2011"

}

You can readily open and visualize this file with for instance ncBrowse.

Some additional, optional WCS keywords are:

For 2.0.1 the full GetCoverage request is presented below (incl. subsetting)

https://rwsprojectarchief.openearth.nl/geoserver/mepduinen/wcs?service=WCS
&version=2.0.1
&CoverageId=SpanjaardsDuinT0_50cm
&request=GetCoverage
&format=image/tiff
&subset=Y(446730.5,447414.5)
&subset=X(68628.5,68998.5)

Accessing WCS by Python OWSLib library

Similar to WFS and WMS accessibility OWSLib has the ability to get data from a WCS service. The code below is a start and gives the contents of data accessible as a WCS.

import os
import tempfile
from owslib.wcs import WebCoverageService
url = 'http://deltaresdata.openearth.nl/geoserver/ows?'
wcs = WebCoverageService(url,version='1.0.0')
wcs.identification.type
wcs.identification.title
list(wcs.contents)

The code above results in a list of available rasters. For Sediment Thickness of the world some characteristics are retrieved via next code.

sed = wcs['global:Sediment Thickness of the World']
sed.keywords
sed.grid.highlimits
sed.boundingboxes

WCS enables you to get a specific part for a specific bounding box like in the following example. For a specified bounding box a tif will be written to a temporary file

requestbbox = (-6.283406057098745, 36.59251369598765, -6.266739197530853, 36.6050139853395)
requestwidth = 4
requestheight = 3
gc = wcs.getCoverage(identifier=sed.id,
                                  bbox=requestbbox,
                                  format='GeoTIFF',
                                  crs=sed.boundingboxes[0]['nativeSrs'],
                                  width=requestwidth,
                                  height=requestheight)
# random unique filename
tmpdir = tempfile.gettempdir()
fn = os.path.join(tmpdir,'test.tif')
f = open(fn,'wb')
f.write(gc.read())
gc.close()
f.close()