Growth Modeling With Divisions And Growth Factor Automatic Updating Of Cunninghamia Lanceolata Stand

Forest resources are one of the most important natural resources in China,which play an important role in the national economic development and ecological environment.In order to solve the problem of shortage of forest resources and realize the sustainable development of forest resources in China,it is necessary to explore the law of forest growth based on China's national conditions,conduct management of forests rationally,and improve the growth rate and output of forests.In order to realize automatic forest growth prediction,this paper proposes automatic forest growth prediction modeling method by studying tree species division modeling technology and growth factor automatic updating technology.In the tree species division modeling part,Cunninghamia lanceolata is regarded as the target tree species,and 16 provinces(autonomous regions)with Cunninghamia lanceolata distribution are used as the research area.Combined with the environmental factors of the study area,the random forest algorithm and the random forest recursive feature elimination algorithm are used to screen the environmental factors.Based on spatial clustering method,the study area is grouped and grouping model is established.In the part of automatic modeling method for stand growth prediction,the construction and management methods of model database and sample data,model fitting and model evaluation techniques are studied to realize automatic prediction of stand growth.Finally,based on the automatic modeling method of forest stand growth prediction,the sample database is constructed by using the group data of Cunninghamia lanceolata.The model in the model library is selected to fit the growth model of Cunninghamia lanceolata,and the model is selected according to the results of model checking,so as to realize the automated modeling and prediction of Cunninghamia lanceolata and the rapid updating of forest resources data.The main research contents are as follows:1.Screening of environmental factors based on random forest algorithm and the random forest recursive feature elimination algorithm.The environmental factors(32 factors of topography,soil and climate)in the study area are selected as the characteristic factors.The relevant environmental factors are removed by correlation analysis.The remaining 15 factors are integrated into the data set of Cunninghamia lanceolata sample plots.Random forest and random forest recursive feature elimination algorithm are used to select the environmental factors that had a great impact on the growth of Cunninghamia lanceolata.The results showed that Bio4(Standard deviation of seasonal variation of temperature),Elevation,Bio3(Isothermality),Bio8(The wettest quarterly average temperature),Bio1(Annual average temperature),Bio14(the most dry month precipitation),Bio12(annual average precipitation),Bio2(Monthly mean diurnal temperature variation),had the greatest impact on the growth of Cunninghamia lanceolata.2.Grouping of Cunninghamia lanceolata research areas based on spatial clustering.The eight environmental factors are selected as the characteristics,and the spatial research method is used to group the Cunninghamia lanceolata research areas.After analysis,the research is divided into 7 groups.The grouping and non-grouping growth rate models of Cunninghamia lanceolata is established and tested respectively.The results show that the accuracy of the grouping model is greatly improved compared with the non-grouping model.3.Research on automatic modeling method for forest growth prediction.Referring to the data and sorting out the stand growth models with wide application and high precision at present,study the model library structure and construction method;based on the modeling requirements and sample storage requirements,the construction method of sample database is studied;study model fitting technology and evaluation technology,propose automatic modeling method for forest growth prediction.4.Application of automatic modeling method for forest growth prediction.Based on the automatic modeling method of stand growth prediction studied in the previous paper,a database is constructed with grouped data of Cunninghamia lanceolata as samples,and a model base is built.The DBH growth,height growth and accumulative growth of Cunninghamia lanceolata are simulated by single factor model in two groups of data selection model base of Cunninghamia lanceolata in the sample database.According to the obtained model accuracy table,Richards model is selected for Cunninghamia lanceolata DBH and tree height model,and Logistic model is selected for Cunninghamia lanceolata accumulation model.The model results are written into the model results table in the model database respectively,and the data in the model results table is used to update sub-compartment data of Cunninghamia lanceolata forest resources.This paper has the following innovations:(1)A grouping method of Cunninghamia lanceolata distribution area based on environmental similarity is proposed.The grouping modeling method provides a new method for large-area high-precision prediction of main plantation species.(2)An automatic modeling method for forest growth prediction is proposed,which can achieve rapid update of small data of forest resources.