ogr2ogr Cheat Sheet: The 20 Commands Every GIS User Needs

ogr2ogr is the Swiss Army knife of vector GIS. It ships with GDAL, reads more than 80 formats, and once you internalise the argument order — ogr2ogr [destination] [source] [options] — most data wrangling collapses into a one-liner.

This is the reference I wish I had three years ago. Every command below is real and tested against GDAL 3.8+.

1. Shapefile → GeoJSON

ogr2ogr -f GeoJSON out.geojson in.shp

The default. The output CRS matches the input .prj. If the input has no .prj, GDAL warns but writes the file.

2. GeoJSON → Shapefile

ogr2ogr -f "ESRI Shapefile" out.shp in.geojson

Watch the 10-character field name limit; -lco RESIZE=YES shrinks string columns to fit their content.

3. Reproject to Web Mercator

ogr2ogr -t_srs EPSG:3857 out.geojson in.geojson

-t_srs is *target* SRS. Use -s_srs if the source has no embedded CRS and you need to declare one.

4. Reproject UTM → WGS84

ogr2ogr -t_srs EPSG:4326 -s_srs EPSG:32633 out.geojson in.shp

Useful when a Shapefile arrives with a missing or wrong .prj.

5. Shapefile → GeoPackage (single layer)

ogr2ogr -f GPKG out.gpkg in.shp -nln parcels

-nln (new layer name) names the layer inside the GeoPackage.

6. Multiple Shapefiles → one GeoPackage

for f in *.shp; do
  ogr2ogr -f GPKG -update -append out.gpkg "$f" -nln "$(basename "$f" .shp)"
done

-update -append opens the existing .gpkg and adds a new layer per Shapefile.

7. SQL filter at read time

ogr2ogr -f GeoJSON out.geojson in.gpkg -sql "SELECT * FROM parcels WHERE area_ha > 5"

-sql runs an OGR SQL query against the source. Cheaper than reading everything and filtering after.

8. Spatial clip to a bounding box

ogr2ogr -spat 12.0 47.0 13.5 48.5 out.geojson in.geojson

Coordinates are minx miny maxx maxy in the source CRS unless you add -spat_srs EPSG:4326.

9. Spatial clip to a polygon

ogr2ogr -clipsrc mask.geojson out.geojson in.geojson

Clips features to the mask polygon. Use -clipsrcwhere to use only a subset of the mask.

10. Attribute filter

ogr2ogr out.geojson in.shp -where "landuse = 'residential'"

Lighter than -sql; uses simple WHERE syntax against the source layer.

11. Drop columns

ogr2ogr out.geojson in.geojson -select "name,population,geom"

Keeps only the listed attributes; everything else is dropped.

12. Rename a column

ogr2ogr out.geojson in.gpkg -sql "SELECT name AS feature_name, geom FROM places"

OGR SQL supports AS aliases — the destination column takes the alias.

13. PostGIS → GeoJSON

ogr2ogr -f GeoJSON out.geojson PG:"host=localhost dbname=gis user=me" -sql "SELECT * FROM public.roads"

The PG: connection string is the source; the SQL is evaluated server-side.

14. GeoJSON → PostGIS

ogr2ogr -f PostgreSQL PG:"host=localhost dbname=gis user=me" in.geojson -nln roads -overwrite

-overwrite drops the existing table; omit it to append.

15. KML → GeoJSON, flatten attributes

ogr2ogr -f GeoJSON out.geojson in.kml -dim XY

-dim XY discards the Z dimension if you don't need elevations. KML's ExtendedData attributes are preserved.

16. CSV with coordinates → GeoJSON

ogr2ogr -f GeoJSON out.geojson -s_srs EPSG:4326 \
  -oo X_POSSIBLE_NAMES=lon -oo Y_POSSIBLE_NAMES=lat in.csv

-oo (open option) tells the CSV driver which columns to use as coordinates.

17. Simplify geometries

ogr2ogr -simplify 0.0001 out.geojson in.geojson

Douglas-Peucker simplification with tolerance in source CRS units. 0.0001 degrees ≈ 11m at the equator.

18. Reduce coordinate precision

ogr2ogr -f GeoJSON -lco COORDINATE_PRECISION=6 out.geojson in.geojson

6 decimal places resolves to ~11 cm — more than enough for web maps and roughly halves the file size.

19. Append features to an existing layer

ogr2ogr -update -append out.gpkg new_features.shp -nln existing_layer

Schemas must match (column names and types). Use -skipfailures to keep going on mismatches.

20. Inspect without converting

ogrinfo -so -al in.gpkg

Technically not ogr2ogr, but the companion tool. -so is "summary only," -al is "all layers." Shows feature counts, CRS, attribute schema.

Closing notes

These twenty cover roughly 90% of day-to-day GIS data work. The remaining 10% is dialect-specific: M-values in Shapefile, GeoPackage attribute styles, GMLAS schema mapping. Reach for the GDAL driver docs when you hit them.