Fusing Minnaert-k parameter with spectral unmixing for forest heterogeneity mapping using CHRIS-PROBA data

In this study we combine the spectral domain with the directional domain of hyperspectral CHRIS (Compact High Resolution Imaging Spectrometer) data to map 3D heterogeneity of an Alpine coniferous forest during wintertime. CHRIS mounted onboard the PROBA (Project for On-board Autonomy) spacecraft is capable of sampling terrestrial reflectance anisotropy over the visible/near-infrared region of the electromagnetic spectrum with high spatial resolution. In the spectral domain, linear unmixing of the nadir image led to a forest cover density map. In the directional domain, the Rahman-Pinty-Verstraete (RPV) model led to Minnaert-k parameter retrieval that provided information of surface heterogeneity at subpixel scale. Due to the bright snow background Minnaert-k yielded the best relationship with forest cover density at the 773 nm spectral band of CHRIS. Fusing both independently derived maps resulted in a forest cover map that includes information of 3D canopy heterogeneity at the subpixel scale.