Faster I/O between pygeos and geopackage file format for use in GIS.
Geopackages are a modern, sqlite-based file format containing geospatial data. These are intended to overcome shortcomings in the shapefile format that is ubiquitous in the geospatial data world.
pygeos is a very fast Python wrapper around the GEOS library. It provides very fast geospatial operations on geometries, including checking for intersections between geometries or calculating the geometric intersection between them.
The goal of this library is to allow serializing pandas DataFrames containing pygeos geometry objects as fast as possible. We wanted faster GIS compatible outputs from our processing chains in other projects, which are starting to leverage pygeos heavily.
This is a shim until pygeos is fully integrated into geopandas.
Not available on PyPi yet, so clone the repository in Github and python setup.py develop.
This requires the latest available version for pygeos, right now from the master branch of that repository.
To create a geopackage, given a pandas DataFrame df containing pygeos geometry objects in the 'geometry' column:
from pgpkg import Geopackage
with Geopackage('test.gpkg', 'w') as out:
out.add_layer(df, name='Test', crs='EPSG:4326')
Note: only write mode is supported at this time.
According to the benchmarks in our test suite, we are seeing 2-3x speedups compared to writing shapefiles or geopackages in geopandas.
This package may change radically once pygeos is used internally within geopandas.
We are not focusing on full coverage of all of the different ways of writing data to a geopackage. If that is what you need, use ogr2ogr or one of the other Python packages available for geopackages.