ROBUST LEAF AREA INDEX RETRIEVAL USING SENTINEL-2 RED EDGE BANDS

摘要

Current satellite leaf area index (LAI) productsrnoften have three major limitations: the need forrnland cover specific tuning, signal saturation at highrn(>3) LAI, and sensitivity to acquisition geometryrnand soil reflectance. Empirical studies suggest thatrnspectral features derived from red-edge (660nm –rn760nm) reflectance can be robustly related to LAI.rnA theoretical basis for a physically based algorithmrnrelating LAI to the normalized difference (ND) ofrndirectional reflectances from Sentinel-2 spectralrnsampling is presented. The concept of canopyrnspectral invariants is used to relate the ND in thernred-edge to the canopy recollision probability andrnleaf single scattering albedo. This allows retrievalrnof LAI through a simple relationship between LAIrnand recollision probability. The approach isrnevaluated using imagery over a mixed land usernregion.