Abstract:

In order to provide accurate model for the forest timber volume estimation,using the data of 245 Cunninghamia lanceolata trees obtained on the standard sampling sites by the forestry table compilation work,both one variable and binary model,as well as multiple variables model based on diameter at breast height (DBH),tree height (H) and crown width (Cw) as independent variables were optimized and constructed.The 11 types of curve model were fitted firstly with one independent variable.And then binary model was established using power function.In the end,the DBH-H-Cw multiple independent variables volume model was established by nonlinear regression estimation method,which was regarded as the best model for the fitted and tested parameters.The results showed that: 1) The power function was regarded as the best model among the 11 types of curve model.2) As far as the determinant coefficient and significant test level were concerned,multiple independent variables model was superior to two independent variables model,and then which was superior to one variable model.The most optimized multiple independent variables model combined of DBH,H and Cw had the determinant coefficient of 0.988,the contrast error of 0.087%,the total error of 0.57%,and the prediction accuracy of 99.40%.3) The most optimized model can drag the maximum loading information of the data obtained from forest timber structure,and provide measurement basis for highly accurate estimation method to the forest volume.

Key words: nonlinear regression procedure, Cunninghamia lanceolata, volume model, Guangxi