|
We were given city administrative boundary data as .shp files with the permission to import those to OSM / use on AddisMap.com. I have used the Ogr2Osm scripts to convert this data to the OSM format. I tried all three versions (old SVN, UVM rewrite, pnorman's updated version) which are linked in the Wiki. All of them produce the same results, but when I load the data in JOSM and compare them with the existing city data (Addis Ababa, Ethiopia), the imported data seems to be shifted by 176m, 222° (I measured this by drawing a way from the point where it should be to the imported point in JOSM) When using Ogr2Osm it produces the following output: Detected projection metadata:
PROJCS["Adindan_UTM_Zone_37N",
GEOGCS["GCS_Adindan",
DATUM["Adindan",
SPHEROID["Clarke_1880_RGS",6378249.145,293.465]],
PRIMEM["Greenwich",0.0],
UNIT["Degree",0.0174532925199433]],
PROJECTION["Transverse_Mercator"],
PARAMETER["False_Easting",500000.0],
PARAMETER["False_Northing",0.0],
PARAMETER["Central_Meridian",39.0],
PARAMETER["Scale_Factor",0.9996],
PARAMETER["Latitude_Of_Origin",0.0],
UNIT["Meter",1.0]]
The .prj file we received has the following content: PROJCS["Adindan_UTM_Zone_37N",GEOGCS["GCS_Adindan",DATUM["D_Adindan",SPHEROID["Clarke_1880_RGS",6378249.145,293.465]],PRIMEM["Greenwich",0.0],UNIT["Degree",0.0174532925199433]],PROJECTION["Transverse_Mercator"],PARAMETER["False_Easting",500000.0],PARAMETER["False_Northing",0.0],PARAMETER["Central_Meridian",39.0],PARAMETER["Scale_Factor",0.9996],PARAMETER["Latitude_Of_Origin",0.0],UNIT["Meter",1.0]] The data was exported from ArcGIS. What can be the source for the error? How can it be fixed? EDIT: Same problem when I use ogr2ogr and shp2osm ogr2ogr -t_srs WGS84 -s_srs "ESRI::subcities.prj" "subcities_converted.shp" "subcities.shp" perl shp2osm.pl subcities_converted.shp > subcities.osm Semi cross-posted to Stackexchange GIS |
|
As found out on GIS.stackexchange I need to specify the projection string manually, because the ogr library does not detect it properly / does not do the datum conversion: I think we used to have a similar problem with EPSG:27700.
(14 Aug '13, 18:52)
SK53 ♦
@SK53: did you solve it the same way? Or another?
(18 Aug '13, 18:28)
AddisMap_Ale...
yes I added the explicit Helmert factors (the 7 towgs84 parameters), Chris Hill blogged about it at the time (http://chris-osm.blogspot.co.uk/2010/04/getting-boundaries-right.html), but IIRC there were still bugs in proj4 (and therefore ogr2ogr, PostGIS) relating to WGS84 and OSGB transformations in 2011. A useful test if you have postgis, but I expect can be done with ogr2ogr is to use a well-defined point and perform transforms between various projections to test accuracy (particularly if there is a non-porj4 transformation available for comparison, as was the case for OSGB).
(18 Aug '13, 19:36)
SK53 ♦
|