Estimating Forest Leaf Area Index Based on BP-Neural Networks

Leaf area index(LAI) is an important biophysical parameter of canopy structure that is related to biomass, carbon and energy exchange, and is an important input to ecological and climate change models. It is significance that estimating surface LAI for the study of earth ecological systems. How to quickly and efficiently estimate LAI of regional or global scale becomes a hot topic for research. The paper use HJ-1B CCD image, the remote sensing data for estimating forest LAI based on BP-Neural Networks in Beijing mountainous area. First of all, Based on Leaf Optical Properties Experiment (LOPEX93) database and simulations using the SAIL bidirectional canopy reflectance model coupled with the PROSPECT leaf optical properties model, the authors have established the Look Up Table (LUT) between the vegetation LAI and the canopy reflectance. Secondly, geometric correction and atmospheric correction were applied to the HJ-1B CCD. And the geometric rectification error was less than one pixel, and the 6S model is used to ensure the accuracy of atmospheric correction, the output is the four band reflectance. Thirdly, four kinds of vegetation index (VI) were established with the spectral reflectance both in the remote sensing data and the LUT. These Vls include DVI, RVI, NDVI, SAVI. Fourthly, BP neural network training with three layers, the input layer has eight parameters, that is four spectral reflectance corresponding with the HJ-CCD and four Vis, and the output layer is LAI. Then estimating of LAI use trained BP neural network. Finally validate the estimating result with the LAI in situ. The estimated results of LAI has a well accordance with the in situ LAI data. The precision of LAI estimating from BP neural network model was about 75%.