Hello mappers, TL;DR: I am working on importing GNS data for Somalia into OSM for HOT, because there is a terrible famine on and our coverage is pants. I am trying to be meticulous about it. I welcome feedback. I will also be importing donated road data later in the week.
The Humanitarian OpenStreetMap Team has been asked by a couple of humanitarian organizations to work on improving OSM's coverage in the Horn of Africa, due to the famine currently worsening there. http://en.wikipedia.org/wiki/2011_Horn_of_Africa_famine I am currently working on preparing the NGA's GNS points of interest for Somalia for import into OpenStreetMap. Here is the feature type mapping I am using, which I worked out by hand: http://iconocla.st/hot/so/dsgfreq.txt Features that are not marked with OSM tags in the above file will not be imported. By and large, I decided to exclude linear features, for which a single node makes little or no sense. Granted the choices were somewhat arbitrary, but I did use the Map Features wiki page, as well as Taginfo, to try to determine the right combination of tags. I have pulled down the latest OSM data for Somalia, and identified all GNS points that match one of the following criteria: 1) closer than 100m to an existing named OSM point of interest 2) closer than 5km to an existing named OSM PoI with an 80% trigram similarity on at least one name variant This analysis yielded a list of 92 GNS points, which will be excluded from the automated import. I plan to extract these points and have someone conflate them with the remainder of OSM manually in JOSM. Here's a sample of the OSM XML I'm generating: <node id="-3" lat="6.540600" lon="47.660800"> <tag k="gns:ufi" v="-3228007" /> <tag k="gns:cc1" v="SO" /> <tag k="name" v="Qarsooni" /> <tag k="gns:dsg" v="WLL" /> <tag k="man_made" v="water_well" /> <tag k="gns:adm1" v="10" /> <tag k="gns:modify_date" v="2011-02-14" /> <tag k="gns:lc" v="som" /> <tag k="source" v="GNS" /> <tag k="alt_name" v="Qarsoon" /> </node> You'll note that I'm preserving as much of the GNS metadata as possible, for later identification. I am also preserving as many of the name variants as possible: <node id="-38 visible="true" lat="4.700556" lon="45.084167"> <tag k="gns:name:generic" v="Buur" /> <tag k="gns:ufi" v="-3219026" /> <tag k="gns:cc1" v="SO" /> <tag k="name" v="Buur Dameerlays" /> <tag k="gns:dsg" v="HLL" /> <tag k="gns:adm1" v="07" /> <tag k="gns:modify_date" v="2006-08-23" /> <tag k="ele" v="333" /> <tag k="source" v="GNS" /> <tag k="gns:name:variants" v="Dameerlays;Bur Damer Aleis;Bur Damer Alais;Damerlays" /> <tag k="alt_name" v="Damerlays" /> <tag k="natural" v="hill" /> </node> Anyway, unless there is grave objection, I plan to proceed with the import of 10,940 localities and points of interests in Somalia within the next day or so. I am going to set up a special account for this purpose. I would *greatly* appreciate any feedback that people with more experience doing OSM imports have. Later in the week, I plan to do an import of roads and rivers data provided to us by FAO Somalia. Here are some details about the road data from a post to the HOT list: http://lists.openstreetmap.org/pipermail/hot/2011-August/000897.html In case anyone is wondering, we have resolved the "geological trace" question, and will be importing them as graded tracks. Thanks! SDE _______________________________________________ Imports mailing list [email protected] http://lists.openstreetmap.org/listinfo/imports
