Unlike geopandas, mundi.py does not extend from pandas. This results in more Pythonic syntax and reduced memory consumption.
By default, mundi.py only loads relevant chunks of a dataset. This allows a developer to write scripts on 100GB+ of spatial data without manually chunking or sending the operation to a compute cluster.
In mundi.py, every geometry is a subclass of a shapely geometry. Features are accessed like an array:
Perform intersections, calculate areas, distances, and boundaries of shapes with shapely.
Datasets are represented as collections of features. To filter, we iterate over the dataset:
Mundi.py also automatically chooses a projection for geometric operations by selecting a projection that minimizes distortion for the geometries in the operation.