Evaluation of Reflectance and Canopy Scattering Coefficient Based Vegetation Indices to Reduce the Impacts of Canopy Structure and Soil in Estimating Leaf and Canopy Chlorophyll Contents

Abstract

Chlorophyll is of great physiological and ecological significance. Leaf and canopy chlorophyll contents can be retrieved from remotely sensed data based on vegetation indices (VIs). However, the impacts of canopy structure and soil remain open problems. VIs are typically calculated from spectral reflectance. In this study, we also constructed and examined VIs based on canopy scattering coefficients ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">λ</sub> ) from spectral invariants theory. Based on extensive leaf and canopy radiative transfer simulations, linear regression and artificial neural network models were built with reflectance-based and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">λ</sub> -based VIs to retrieve leaf chlorophyll content (LCC) and canopy chlorophyll content (CCC). The results showed that the canopy structure and soil significantly affected the retrievals. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">λ</sub> can effectively suppress the impacts of the leaf angle distribution (LAD) but not the leaf area index (LAI). The <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">λ</sub> , estimated as the ratio reflectance/directional area scattering factor (DASF), contained a large error when the soil effect was strong. The <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">λ</sub> -based VIs did not yield very accurate results in LCC estimation but exhibited higher accuracy for CCC estimation compared to reflectance-based VIs. Of all the VIs investigated, the best VI was D99 (( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">850</sub> - <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">710</sub> )/( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">850</sub> - <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">680</sub> )) for LCC and Wmul ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">749</sub> × <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">956</sub> ) for CCC. Compared to D99 for LCC, Wmul for CCC was less accurate, and the accuracy varied more among canopies with different LADs. The main reason was that CCC equals LCC multiplied by LAI, but LCC and LAI impact VIs in a similar manner.