We had a PostGIS database which was initialized quite some time ago using an osm.pbf file from Geofabrik, without a specified coordinate system. It defaulted to using EPSG 900913 (Google Maps Global Mercator)

Recently, we initialized a new database the same way, but this time we used the flag -E 31370, which initalized the data to EPSG 31370 (Belge 1972 / Belgian Lambert 72).

When we compare the two databases, we can see a very clear coordinate drift (see pic)

enter image description here

Red = EPSG 31370 (Belge 1972 / Belgian Lambert 72)

Blue = EPSG 900913 (Google Maps Global Mercator)

Geodesy is really not my specialty. A colleague and I spent some time going over it trying to figure out which one is right.

Obviously the blue polygons are in the right position according to the basemap. But the basemap is mostly likely also derived from a PostGIS database that has been initialized to a Web Mercator projection for the sake of visualization, whereas the database for the red polygons was initialized to a local projection immediately.

It is my understanding that the OSM dumps from Geofabrik are in WGS84 natively so I am confused as to why they are ending up in two different places.

This is quite a big issue for us because a previous analysis workflow did all the coordinate transformations manually, while a new workflow is trying to initialize everything in the same coordinate system and just use it for the entire duration.

How can we stop this coordinate drift?


It turns it it was a lot less complicated that I thought. The simple fact is that EPSG 31370 (Belge 1972 / Belgian Lambert 72) is just an old and outdated CRS. We re-initalized the database in EPSG:3812 (ETRS89 / Belgian Lambert 2008) and everything is exactly where it should be.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.