Spectrodirectional investigation of a geometric-optical canopy reflectance model by laboratory simulation

Abstract

Canopy reflectance models (CRMs) can accurately estimate vegetation canopy biophysical-structural information such as Leaf Area Index (LAI) inexpensively using satellite imagery. The strict physical basis which geometric-optical CRMs employ to mathematically link canopy bidirectional reflectance and structure allows for the tangible replication of a CRM's geometric abstraction of a canopy in the laboratory, enabling robust CRM validation studies. To this end, the ULGS-2 goniometer was used to obtain multiangle, hyperspectral (Spectrodirectional) measurements of a specially-designed tangible physical model forest, developed based upon the Geometric-Optical Mutual Shadowing (GOMS) CRM, at three different canopy cover densities. GOMS forwardmodelled reflectance values had high levels of agreement with ULGS-2 measurements, with obtained reflectance RMSE values ranging from 0.03% to 0.1%. Canopy structure modelled via GOMS Multiple-Forward-Mode (MFM) inversion had varying levels of success. The methods developed in this thesis can potentially be extended to more complex CRMs through the implementation of 3D printing.