Natural forest conservation hierarchical program with neural network

Abstract

In this paper, the implementing steps of a natural forest protection program grading (NFPPG) with neural network (NN) were summarized and the concepts of program illustration, patch sign unification and regression, and inclining factor were set forth. Employing Arc/Info GIS, the tree species diversity and rarity, disturbance degree, protection of channel system, and classification management in the Maoershan National Forest Park were described, and used as the input factors of NN. The relationships between NFPPG and above factors were also analyzed. By artificially determining training samples, the NFPPG of Moer-shan National Forest Park was created. Tested with all patches in the park, the generalization of NFPPG was satisfied. NFPPG took both the classification management and the protection of forest community types into account, as well as the ecological environment. The excitation function of NFPPG was not seriously saturated, indicating the leading effect of the inclining factor on the network optimization.