| Forest is the largest terrestrial ecosystem on earth and plays an important role in the global biosphere.The information of national forest resources is mainly obtained according to the comprehensive monitoring system of national forest resources and ecological conditions.The National Forest Inventory is an important part of this system.And the sample plot data of continuous inventories are of great significance for forestry research on a national scale.In this paper,the study of 10 tree species(groups)with the largest Area in China is obtained by using the sampling data of continuous inventory of forest sample plots,which are Quercus,Pinus massoniana Lamb.,Cunnminghamia lanceolata(Lamb.)Hook.,Betula,Larix gmelinii(Ruprecht)Kuzeneva,Larix gmelinii(Ruprecht)Kuzeneva,Pinus yunnanensis Franch.,Picea asperata Mast.,Cupressus funebris Endl.,Cupressus funebris Endl..And using one-way analysis of variance,stepwise regression analysis,partial correlation analysis and deep learning method,the tree growth data were combined with longitude and latitude,altitude,temperature,rainfall,slope gradient,slope direction,slope position and soil thickness in different regions.Thus,the suitability of trees growing under certain site and climate conditions and the future growth status of trees can be determined and predicted,providing references for afforestation planning.This paper mainly completed the following research:(1)One-way analysis of variance was used to obtain the distribution and growth rules of 10 tree species(groups)under different environmental factors,and the optimal growth status of each tree species in each factor was obtained.(2)Stepwise regression analysis was used to establish stepwise regression equations for the above 10 species,judged the effect of environmental factors on the growth of trees and obtained the order of importance of impact factors of each tree species and the factors with the greatest impact.(3)Partial correlation analysis was used to obtain the partial correlation coefficients between environmental factors and the growth of 10 main tree species,and the net correlation between each factor and tree growth was obtained.And compared with its Pearson correlation coefficient,there is a big difference between the two.(4)The fully connected neural network constructed by deep learning framework and combined with environmental factors were used to predict the DBH growth of 10 dominant tree species in China.The high accuracy obtained indicated that this method was feasible in tree growth prediction. |
