Inversion of a radiative transfer model for estimating forest lai from multisource and multi-angular optical remote sensing data based on ANN

摘要

A new forest leaf area index (LAI) inversion method from multisource and multi-angle data combined with radiative transfer model and the strategy of k-means clustering and artificial neural network (ANN) was discussed. The four different temporal satellite images of Landsat-5 TM (L5TM) and Beijing-1 microsatellite multispectral sensors (BJ1) were selected to construct multisource and multi-angle data. Considering the vertical distribution of forest LAI for trees and understory vegetation, the hybrid model of the invertible forest reflectance model (INFORM) was used to support the retrieval forest LAI to eliminate the dependence of understory vegetation. Through the validation of inverted results with MODIS (Moderate Resolution Imaging Spectroradiometer) LAI product and field measurements, it can be concluded that the accuracy of inversion forest LAI can be improved through adding up observation angle data, if the quality of data were ensured. The inversion accuracy for multi-angle data was improved 20% than the average accuracy of single-angle data inversion of LAI.