Neural Network Estimation of Urban Leaf Area Index

Leaf area index (LAI) is a critical parameter for urban forest monitoring. The goal of this study in Terre Haute, Indiana, USA was to develop algorithms to model gap-fraction LAI measured on sample plots as a function of radiometric response measured by the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER). Two neural networks facilitated the modeling. The first was trained to detect sites dominated by bare ground using ASTER visible, infrared, and thermal channels. The second estimated LAI as a function of vegetation indices. When the field sample sites were resubmitted to the two networks, the resulting systemwide standard error of the estimate was 1.25 LAI units.