Study Of The Stand Growth Model For Eucalyptus Plantation

In recent years, owing to the characteristics involving in fast-growing, high yield, wide range of applications, high economic benefits, etc, Eucalyptus were introduced to palnt in north from south progressively in Fujian Province. Currently, Eucalyptus varieties introduction in Fujian Province had gradually formed scale, and its operating and management aroused more and more forestry researchers'attention. Stand growth rhythm of Eucalyptus were important theoretical basis for its operating and management. Therefore, the paper, toward Eucalyptus palnted in north from south in Fujian Province, took Eucalyptus in Yongan City located in northwest Fujian Province as research object, and analyzed and studied the stand growth rhythm and some operating issues in the hope of providing a theoretical reference for Eucalyptus's operating and management introduced to palnt in north from south in Fujian Province.Artificial Neural Network (ANN) was one of the modern intelligent algorithms. ANN needn't consider the interior structure of mathematical model, suppose premise conditions and confirm factors power artificially. It could map the integrity of research objects synthetically, and had abilities of large-scale parallel computing, adaptive, self-learning, fault tolerance capabilities. Consequently, ANN was widely used in the nonlinear and complex predictive modeling in biological system. Therefore, basing on previous studies for the stand growth rhythm, the paper used ANN principle to study stand diameter, tree height, two-dimensional distribution of diameter and tree height, whole stand model, and stand merchant ratio model of Eucalyptus plantations planted in north from south. The results indicated that:(1)Respectively by relative diameter and relative height as input variable, accumulating frequency in number of trees as output variable, logsig as the transfer function of hidden layer and output layer, the paper constructed 1:2:1 Eucalyptus plantation stand relative diameter and relative height accumulating frequency in number of trees BP neural network. The average forecasting precision value of relative diameter accumulating frequency BP neural network was 0.8819. The average forecasting precision value of relative height accumulating frequency BP neural network was 0.8810. The result indicated the two models effect were better and had a better adaptability for the distribution of Yongan Eucalyptus plantation stand relative diameter and height accumulating frequency in number of trees.(2)By relative height and relative height as input variables, accumulating frequency in number of trees as output variable, logsig as the transfer function of hidden layer and output layer, the paper applied the method of BP neural network to construct 2:3:1 Eucalyptus plantation stand relative diameter and relative height two dimension accumulating frequency in number of trees BP neural network.(3) By age, site type and stand number density as input variables, average breast-height diameter(DBH), average height and volume as output variables, tansig as the hidden layer transfer function, purelin as the output layer transfer function, the paper constructed 3:6:3 Eucalyptus plantation whole-stand growth BP neural network model. Using 38 sample plots to forecast with the constructed model, the results indicated the average forecasting precision value of average DBH was 0.8675, the average forecasting precision value of average height was 0.8789, average forecasting precision value of volume was 0.7869.(4) By diameter class and height processed for normalization as input variables, stand whole merchant ratio as output variable, logsig as the transfer function of hidden layer and output layer, the paper constructed 2:4:1 Eucalyptus plantation stand merchant ratio BP neural network model. Using 49 sample trees as forecasting samples to forecast with the constructed model, the result showed the err average value of forecasting merchant ratio and fact ratio was 0.0042, the average forecasting precision value of was 0.9481, indicated the forecasting effort of constructed model was better.All of the above study indicated that artificial neural network model had a good adaptability on Yongan Eucalyptus plantation stand growth rhythm study, the model forecasting accuracy can also meet the requirement, and its application would provide a new direction and method for management study of northward Eucalyptus plantation.