[Geodata] NZ geodata on OSGEO

[Hamish]

Brent Wood wrote:
> As an update on where this is at; there is one change to
> the LINZ scenario which impacts on this.
> 
> LINZ is now developing a website with the raster &
> vector topographic data online. I'm told it will also have
> maps/data available via WMS/WFS. This will
> be several weeks away, and largely replicates what I think
> we envisaged doing under OSGEO.

This is very nice to see and solves the stale data problem.


> The application is being developed with ESRI tools.

I do what I can to maintain the code for GRASS's r.in.wms module. It is a bit brittle and it is on the project TODO list to rewrite it in another language, but it mostly works quite well.

If I may provide some observations from recent fights with dataservers ... (I bring this up as my headaches usually involve talking to ArcIMS)

- The server should support sending the query by http POST data method. The OGC spec states that sending query on one huge GET line is ok, but http has a data query function, and it should be used. This is a data query after all!

- the WMS server should support GeoTiff export. Otherwise the server just ships pretty pictures which are fine for a webapp, but not usable data. png/gif format seems to go out as 8bit images which are a pain to tile as each image has its own unmatching color palette (maybe gdal_merge.py helps?).

As it happens I had to adjust the code in the last few days for the above two issues to be able to get it to work with Geoscience Australia's WMS server.

- ensure that layer names are not wordy descriptions. e.g. they should be 6 alpha-numeric chars with no whitespace or punctuation.
(NOAA's S-57 WMS server suffers from all three of the above issues)


WMS is not much good for vector data beyond pretty pictures. So I hope WFS is done well. I also look forward to trying GDAL 1.5.0's new WMS import tool, but that is very new code and not widespread yet.


> This does not preclude the OSGEO initiative moving forward,
> but from the perspective of actually getting these NZ geodata
> online it becomes somewhat redundant, although a FOSS equivalent
> may be a useful demonstration.

I think it may have stronger value than just as a demonstration. It is nice to have things locally at your finger tips; who knows what/when future gov'ts will decide to change IP policy? I wouldn't like to trust my geo-business on that risk.


> It does still leave the door open for other LINZ datasets,
> such as cadastral & road data to be set up under OSGEO with
> background topo data via OGC web services direct from LINZ.
> This approach would demonstrate the interoperability
> of ESRI GIS applications & data with FOSS applications,
> which may be just as useful.

Regardless of what LINZ decides to do themselves I think it is a strong message+demonstration to other gov'ts/agencies if OSGeo hosts a not-watered-down version of a country's geodata and it is clearly used for educational projects etc. There is a difference at international conference workshops between "the NZ makes their data available" and "here it is".


> I'm still uncertain as to whether the appropriate
> people in OSGEO:geodata have
> decided the LINZ licence for these data is acceptable for
> such use or not.

AFAIU we have a generous offer to supply the data from a "reseller" with the same terms as LINZ; a generous offer to host the data; the license terms are very clear and non-problematic (and need I say generous), and there have been no concerns raised about it.


go! go! go!

Hamish