Forest Classification & Deforestation Tracking

Forest / non-forest at 2 m, and deforestation as canopy loss between any two dates — training-free, across areas up to 100,000 km².

Forest change is often reported late and coarsely. ETD classifies forest versus non-forest directly from the 2 m multispectral signal — no training labels — and compares two dates to surface canopy loss (clearing, logging, fire scars) and regrowth. It builds on the same physics-based land-cover representation, adding a canopy-specific layer, so a single run returns both a current forest map and a dated deforestation change layer over an OSM reference.