Study on Site Quality Assessment of Afforestation Land Based on GA-RBF Neural Network

Abstract

The assessment of forest site quality at early stages of stand development is very essential for scientific afforestation and forest management. In order to enhance the accuracy of the existing models, a new site quality assessment model based on Genetic Algorithm-Radial Basis Function (GA-RBF) was proposed to predict site index (stand dominant height). Data used in this study came from 980 permanent sample plots for Chinese fir (Cunninghamia lanceolata) plantations in Fujian Province, China, which were randomly divided into the training dataset (786 plots) and the testing dataset (194 plots) with a ratio of 8:2. In this paper, the GA-RBF was compared with the radial basis function (RBF) and the traditional Quantitative Theory I (QT-I) method. The results indicated that the predicted accuracy was significantly increased by using the GA-RBF model. Furthermore, we used the existing site-specific site index table of Chinese fir to test the results of the GA-RBF and the agreement was 73.2%. Therefore, we recommend the GA-RBF for assessing site quality of afforestation land.