ArchGDAL driver

ArchGDALDriver() is the default driver. It is always available and does not require an additional package import.

Extension mappings

GeoDataFrames maps these extensions to ArchGDAL/GDAL driver names:

Extension	GDAL driver
.shp	ESRI Shapefile
.gpkg	GPKG
.geojson	GeoJSON
.vrt	VRT
.sqlite	SQLite
.csv	CSV
.fgb	FlatGeobuf
.pq	Parquet
.arrow	Arrow
.gml	GML
.nc	netCDF

For other extensions, ArchGDAL performs extension-based driver detection.

Reading

read(ArchGDALDriver(), fn; layer, kwargs...) reads the first layer by default. Set layer to a zero-based layer index or a layer name to select a different layer. Other keywords are passed to ArchGDAL.read; use options for GDAL open options.

Writing

write(ArchGDALDriver(), fn, table; kwargs...) accepts these driver keywords:

Keyword	Effect
layer_name	Name of the output layer; defaults to "data".
crs	Output CRS; defaults to the table CRS.
driver	GDAL driver name, overriding selection from the filename.
options	Dict{String,String} of GDAL layer-creation options.
geometrycolumn	A geometry-column symbol or tuple; defaults to the table geometry columns.
update	Opens an existing dataset for update instead of creating one.

The ArchGDAL reference and GDAL vector-driver index describe the supported GDAL drivers and their options.

Examples

Reading a CSV

Some data is distributed as CSV files with WKT geometries in a custom column such as wkt. Something like

wkt,name
"POINT (30 10)","point1"

You can read such files as follows:

using GeoDataFrames
# Tell GDAL where to find the geometry column and not to keep it as a regular column (otherwise DataFrames will fail to parse it)
df =
GeoDataFrames.read("test.csv", options=["GEOM_POSSIBLE_NAMES=wkt", "KEEP_GEOM_COLUMNS=NO"])

We don't know the CRS, so it will be nothing by default. You can set it using setcrs! if you know it, but ideally we use better spatial file formats (such as Geopackage) that store the CRS natively.