Estimating forest parameters from top of atmosphere multi-angular radiance data using coupled radiative transfer models

Abstract

Traditionally, the estimation of forest parameters using physically-based canopy radiative transfer models (RT) requires correcting the remote sensing data to top-of-canopy (TOC) level by inverting an atmosphere RT model. By coupling the same canopy and atmosphere models, it is possible to simulate the top-of-atmosphere (TOA) radiance and to work directly with the measured TOA radiance data, thus avoiding the correction to TOC level. Many studies discussed the increased potential of multiangular data for parameter estimation, especially for forests, which have strong directional properties. These studies, however, were based on TOC data. In this study, we investigate the potential of multiangular data at TOA level, based on a case study for three Norway spruce stands in the Czech Republic, using multi-angular CHRIS data and the coupled SLC-MODTRAN model. The coupled model provided satisfactory TOA simulations of spectral and angular signatures, and the dimensionality of the parameter estimation problem increased with increasing angular sampling. Canopy cover, fraction of brown material, leaf chlorophyll and leaf dry matter content were estimated using all possible angular combinations. No combination was best for all parameters.