The data comes from a multiscale mesh of polygons generated by Arcgis Pro. Usually those polygons are used to create vector tiles optimized for feature density across the multiple levels of detail, but in these maps we are using them as a derived datasource that help us to summarize a big geospatial dataset.
Those summarizations could be used as a first step to implement the Shneiderman’s data visualization mantra for geospatial data:
“Overview first, zoom and filter, then details-on-demand“.
(Btw: Maps were created using Carto).