Forest Resource Inventory Mapping

Generates high-resolution vegetation species distribution layers from temporal multi-sensor stacks — Sentinel-1 SAR and Sentinel-2 multispectral imagery fused with soil, slope, aspect, and elevation covariates. Random Forest models are hyperparameter-tuned across configurations for predictive accuracy and spatial robustness. Operational in Himachal Pradesh, Odisha, Andhra Pradesh, and Maharashtra, with a fifth state in progress and deep-learning models under evaluation on the same feature stack.