Suitability Evaluation And Prediction Of The Potential Distribution Of Pinus Koraiensis Based On Random Forest Model

Pinus koraiensis is a kind of important economic and ecological tree species in our country,while is also one of national reserve forest tree species.Therefore the studying of the Pinus koraiensis is necessary.At present,the forestry resources data is large,and the relation among them becoming increasingly huge,the close link between also more and more closely.Using the method of modern science and technology for the traditional forestry data analysis mining is imperative.To solve above problems,in this paper,Pinus koraiensis was taken as research object,we used the random forest model and the data of Pinus koraiensis distribution points information from national forest continues inventory and forest management inventory,combined with the climate,topography,soil and other environmental data,to predict the potential distribution in our country and analysis the effects of environmental factors on distribution of Pinus koraiensis at the same time.In addition,according to the sample point of Pinus koraiensis,the suitability evaluation model of Pinus koraiensis based on random forest was established in Liaoning Province.Finally,taking Fushun city as a test region,the suitability and adaptability of Pinus koraiensis by random forest model were compared with the results obtained,and studied the integration of the two models.The main content of this paper is as follows:1 We used PRISM spatial interpolation model,combining with DEM and stations climate data,to get temperate climate factors,including the annual average temperature,the annual maximum temperature,the annual minimum temperature and annual precipitation.Compared with traditional interpolation method,PRISM interpolation model had a higher accuracy,and could get more accurately meteorological data.This part laid a foundation to evaluate the quality of site more accurately.2 This research uesd random forest model,DEM,Worldclim climate data,HWSD soil data,and the data of the existence or non-existent of Pinus koraiensis under various site conditions to establish the prediction model of the potential distribution of Pinus koraiensis.The calculation results showed the TSS is 0.9574,AUC is 0.9809,and Kappa is 0.9597.And we got Pinus koraiensis potential distribution map after inputing environment data In order to make the result more reasonable,we calculated the correlation coefficient between the factor,and removed some fators that high correlation coefficient.Then we repeated previous calculation.The calculation results showed the TSS is 0.976,AUC 0.9881,Kappa is 0.9749.The accuracy was improved so that we got more precise and reasonable Pinus koraiensis potential distribution map.This part also used partial correlation function for environmental factors change curve,analyzing the specific influence mechanism.3 The selection criteria for sampling points had two methods.One was river basin,and the other is grid.We got different Pinus koraiensis sampling points in different methods,the combined with meteorological data,soil data and DEM to calculate model accuracy.The results showed river basin sampling points had higher accuracy,which was more suitable for establishing the model.The five best factors were elevation,temperature seasons standard deviation,annual average temperature,average annual precipitation,and annual maximum temperature.Then the model output the map of Pinus koraiensis suitability in Liaoning Province,and analyzed the different suitable area of Pinus koraiensis in Liaoning province.4 Fushun city was selected as the study area of potential distribution of Pinus koraiensis.and a part of Xinbin Manchu Autonomous County as the suitability evaluation studying area.After that,we studied the integrated research of suitability and adaptability.The suitable areas with high levels of Pinus koraiensis has a good match with potential distribution region.Most suitable areas fall in the potential distribution region,while most lowersuitable areas do not fall in the potential distribution region.This article has the following main innovation points:(1)Providing the theoretical method for the prediction distribution and suitability evaluation of the reserve tree species,constructs the relevant data index,and puts forward the scientific work flow;(2)Introduced the random forest model,using the high precision of the regression and classification characteristics,in different sizes to get the correctness and suitability of pinus koraiensis;(3)This paper not only considers the modeling precision of the model,but also compares the simulation results with the actual data in comparison with the actual data,and explains the applicability of the model more rationally.