Tree-Bole Volume Estimation on Standing Pine Trees Using Cascade Correlation Artificial Neural Network Models

摘要: Total tree volume estimation is an integral part of forest growth and yield forecasting. Complex formulae are used to estimate bole by section, based on relationships proposed Huber, Smalian Newton. All these relationships require many measurements of diameters at certain heights that difficult obtain standing trees especially when diameter measurements have be taken several meters above ground. The common practice till now days face the problem application regression analysis for tree-bole estimation, but there problems solved assumptions be carefully selected etc. In this paper attempt was made overcome difficulties by indirect using necessary values certain heights Cascade Correlation Artificial Neural Network models (CCANNs). The cascade correlation algorithm accomplished training ANNs, which a feedforward and supervised learning algorithm. Adaptive gradient Kalman’s rules were modify artificial neural networks weights. rule found superior for tree-bole. The networks designed adapt weights synapses, hyperbolic-tangent function training. reliability developed CCANNs assessed validation on independent testing data set. Paired t-test 45-degree line test were also validation of CCANNs. system proposed in paper, can applied forest inventory calculations producing accurate any section volume. For example, total tree-bole resulted root mean square error value of 0.0054 m3 (9.2%). This only two diameter measurements (stump diameter, d0.3 breast height, d1.3) of total height (h), enough replace standard forestry measurement procedures.