The Individual Tree Model Research Of The Quercus Aliena Being Based On The Thiessen Polygon And ANN

In the research,it models with the artificial neural network . According to the other peoples' research , the growth of the tree is mainly affected by the competitive tree and the age. So it takes the begining diameter D1, the beginning of the age of T1, the end age T2, the beginning point of Tyson density basal area polygon TGP, and the begining area of Thiessen Polygon TA1 as the input variables, and it takes diameter D2 at the end of the target variables.It trains the model with the software matlab to set up the TGP individual tree growth model. It trains and tests the model with 11 Quercus standards data of the survey, which include 3/4 are training sample and 1/4 is testing sample. The model can directly determine the growth of a tree and it's growth potential. We can also define the growth of the tree that were keep down in the thinning through the model, which is a very important in forest intensive management.Compairing with the whole stand growth model or diameter class distribution model it has its unique advantages being put in the practice forest yield. The research can be used widely.It achieve the dynamic prediction of the natural secondary forest resources in Baotianman and it's an important basic research for the forests scientific management. Furthermore ,It has an obvious biological value to take Thiessen polygon point density as an indicator in individual tree's growth , and It has great practical value and academic value for the stand growth model research that is based on GIS and GPS. The results are as follows:1.The mathematical expression of ANN model are: D2= logsig(-89.6246 -1.8871 -89.8187 +90.9554) h1= logsig(13.1917D1+0.15329T1-0.32171T2+0.20718TGP-2.3654TA1-8.7767) h2= logsig(-2.1533D1+4.6894T1-4.6884T2+0.0056299TGP-0.4482TA1+0.86102) h3= logsig(-13.1796D1-0.33872T1+0.50708T2-0.20682TGP+7.2201TA1+3.9124)2.Through the analysis of simulation results, we can see that the end diameter D2 increases with the increasing of the beginning age T1 ,the end age T2 ,the begining area of Thiessen Polygon TA1 and the beginning Tyson polygon point density of basal area TGP,and the relationship is almost linear as is shown in the construction of individuals-tree growth model with age, area of Thiessen Polygon, Tyson polygon point density of basal area ;3. Accuracy Analysis of model results show that the mathematical model of the fitting accuracy is 99.11%,and it's very high;4. The accuracy of the model is 99.99%.So,the predictive ability of the model is excellent.5.The model described above not only has high accuracy to the fitting samples, but also to the test sample , and the model has a good generalization ability.It's difficult to measure the horizontal distance and the azimuth because of the complex topography, and the error in measure is large which affects the precision of the area of the Thiessen Polygon. Besides,the fitting accuracy , the prediction accuracy of the model , as well as the correlation coefficient regression equation is high, but it's confidence is not high.We will have a fourther research in the future study, for the inadequacy of the present study.