2025-02-25, 09:45–10:15, Room 1
duckdb-geography introduces a "geography" type optimized for global vector geometry to DuckDB powered by Google's S2Geometry library.
DuckDB's spatial extension is an excellent wrapper around the planar spatial libraries that underpin the modern open source GIS stack. These libraries are well-suited to planar geometry; however, workarounds are required for global datasets that include polar regions and/or straddle the antimeridian. Google's S2Geometry library (http://s2geometry.io/) provides a robust mathematical basis for spherical operations on the sphere and its C++ implementation is well-suited to wrapping as a DuckDB extension. Using the basis for the spatial data science-focused s2 package for R (https://github.com/r-spatial/s2) and the Spherely package for Python (https://github.com/benbovy/spherely), this talk will showcase the process and result of porting a collection of vectorized geospatial geometry operations to DuckDB.