[OSM-dev] Lake data from SWBD

Ludwig ludwigbrinckmann at gmail.com
Sat May 10 13:29:36 BST 2008


The SWBD data, a data set outlining worldwide water features (coastlines,
lakes, and wider river sections) derived from the SRTM data (see
http://wiki.openstreetmap.org/index.php/Potential_Datasources#SWBD_-_Shuttle_radar_topography_mission_Water_Body_Data.),
has been discussed before, but AFAIK only with respect to coastlines.

The data however also contains inland lakes and waterways. While I find the
waterways a bit dubious as they are a bit intermittent when it comes to
rivers that are sometimes wide, sometimes not, the lakes in this dataset
are, at least for the regions I know, not bad. While they are not exactly a
substitute for mapping from good-resolution imagery, they are even for those
areas where this is available, a good starting point and for other areas
much better than nothing. Unless I am very mistaken, the data is in the
public domain, like all such US government agency produced data.

Is there a reason (other than nobody has done it) why the lake data in this
data set has not loaded into OSM? Are there objections to this data being
uploaded for lakes only?

Different to the SRTM contour data which for size reasons alone should not
be uploaded, the lake data is quite manageable - plus there is some
information that OSM can add to it, namely identifying the lakes in the
data.

There are of course areas where lakes have already been entered, by hand or
by some other means.
I have run a check against osmxapi for this. There are 12K+ tiles of 1
degree size in the SWBD set, for 1766 tiles there are some waterfeatures in
the OSM data already, for 10463 the OSM data contains no water features (as
in natural=water), so with the vast majority of the data there would be no
overlap with other OSM work.
For the rest of the tiles, some strategy has to be found not to mess with
other people's work. This could be in one of the following ways:

   - not upload any SWBD lakes from the tiles where there is a potential
   overlap.
   - upload SWBD data but mark it as 'visible=false' (do renderers honour
   this?), so the data is there an can serve as a reference point
   - derive a further test as to which an SWBD lake overlaps with a lake in
   OSM data, then either not upload it or mark it as invisible. This one is
   clearly the most work to implement.

Comments? Suggestions?

Ludwig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20080510/583723fb/attachment.html>


More information about the dev mailing list