Neural network-based aboveground biomass estimation in Honghe National Natural Reserve using TM data

In order to estimate the wetland vegetation aboveground biomass, the neural network models (BP, RBF) were established based on the Remote Sensing (RS) image of Honghe National Natural Reserve (HNNR) and 29 samples of biomass data. Through training, simulation, and comparing with the measured biomass data, the results show that the accuracy of the biomass estimation by neural network is relatively high. Furthermore the accuracy of the model of dry biomass is higher than that of humid biomass. By comparison between BP network and RBF network, it is found that the RBF network is the better method for estimating the wetland vegetation aboveground biomass with RS information. With the method of RBF, the mean relative error (MRE) of estimated dry biomass was 2.795% and the MRE of estimated humid biomass was 3.366%.