Very similar in motivation to the CoRE stack, to bring many different layers together - land use, hydrology, climate, socio-economic, etc. But they are doing it on rasters with the view to build foundation models that use data beyond just satellites like Sentinel-1/2, whereas we are doing it on vectorized spatial units with interprettable indicators that can be composed into understandable patterns of symbolic expressions like what we did with CoRE insights. Will be very interesting to see at what points these methods converge.
Adi
--
Aaditeshwar Seth Microsoft Chair Professor, Computer Science and Engineering, IIT Delhi