Performance Comparison Artificial Neural Networks with Regression Analysis in Trees Trunk Volume Estimation

Nowadays، regression analysis is a common method to estimate trees stem volume. Althought trees trunk volume can be estimated with a certain accuracy، however there are many constraints such as normality of variables and homogeneity of errors variance، when foresters use this method. In this study، Artificial Neural Networks (ANN) as a subset of the technology of Artificial Intelligence (AI)، was used as a new method to estimate the volume trunk. For this purpose، 101 trees were selected. Marked trees were located in Research and Educational Forest of Tarbiat Modarres University (REFTMU). DBH، diameter at stump height، end diameter trunk، trunk height and total tree height were mesuared with high accuracy during tree marking. Two neural network models، multi-layer perceptron (MLP) and radial basis function (RBF)، were developed to estimate trunk volume. The results showed that with increasing the number of variables، that have more correlation with trunk volume،correlation coefficient of neural networks increased from 0. 80 to. 95. Also the RBF neural network was more accuracte in trunk volume estimation than to MLP neural network. Comparing evaluation criteria for ANN with stepwise regression showed that MLP and RBF neural networks had RMSE value 1. 18 and 1. 05، respectively while the RMSE value of regression was 2. 57. Also the regression correlation coefficient is less in compared with two models neural network.