Training-free land-use / land-cover mapping

National mapping and statistics agencies need current land-use/land-cover without the cost and bias of supervised classifiers that need local labelled data and fail across regions. ETD's physics-based, training-free classifier discovers surface materials and spatial context directly, running anywhere with no calibration, with OpenStreetMap as a majority-vote label rather than ground truth — so high-confidence disagreements surface as candidate change rather than errors. The Land Cover Classification product is in development; today ETD delivers the enhanced multispectral and basemap imagery it runs on.