LAI inversion using a back-propagation neural network trained with a multiple scattering model

摘要

Standard regression methods applied to canopies within a single homogeneous soil type yield good results for estimating leaf area index (LAI) but perform unacceptably when applied across soil boundaries. In contrast, the neural network reported generally yielded absolute percentage errors of >30%. The network was applied, without retraining, to a sample of Landsat TM data for an agriculture/forestry study site.