When I first heard “Vector Data Cube”, I thought of something different.
In Conservative Region Aggregation with Xarray, Geopandas and Sparse I had a slightly different use case. I wanted a 2D array where each value was a vector polygon (GCM grid cell), and I wanted to intersect these with another set of geometries. Because Xarray does not implement vectorized overlay (like geopandas), I had to convert the dataarray to a multiindexed geodataframe.
I’m wondering if there are things we could do to support both workflows. But I don’t want to hijack the thread…