Evaluation Of Land Comprehensive Carrying Capacity Area Based On Machine Learning Algorithm In Dongting Lake

Land Comprehensive Carrying Capacity(LCCC)is an important indicator for quantitative evaluation and judging the degree of sustainable economic development and the harmonious coexistence of man and nature.United Nations Sustainable Development Goals(SDGs)have three goals directly related to land carrying capacity:zero hunger,Sustainable cities and communities,and Life on land.The research on land carrying capacity is also one of the important contents of China’s ecological civilization construction.Under this background,An accurate and objective evaluation of LCCC is of great significance to land resources planning and sustainable land development.This study taking Dongting Lake Ecological Economic Zone as the research object,construction of an index system for LCCC evaluation in study area based on DPSIR(Drivers,Pressures,S tates,Impacts,Responses)conceptual model,build index space database,and construction of Random Forest(RF)model and Support Vector Machine(SVM)for LCCC evaluation of study area.This paper analyzes the evolution patterns and characteristics of LCCC,subsystems and indicators in Dongting Lake Ecological Economic Zone in 2010,2015 and 2020,and finds out the important factors affecting LCCC,and puts forward suggestions for improving the LCCC of Dongting Lake Ecological Economic Zone.The main conclusions are as follows:(1)From the comparison of the effect of RF model and SVM model in the evaluation of LCCC in 2015:the AUC of the RF model test set is 0.925,the AUC of the SVM is 0.757,the accuracy of the RF model is better,because SVM is not suitable for multi-classification problems and large sample size modeling.RF model has advantages in evaluating complex nonlinear systems such as LCCC,and is suitable for LCCC evaluation.(2)In the LCCC evaluation based on RF model,the overall change of LCCC increased at first and then decreased.The LCCC score increased from 304.7 in 2010 to 335.1 in 2015,and then decreased to 328.1 in 2020.There is heterogeneity in spatial change,the carrying capacity of the middle,north and east increases,while that of the south and west decreases.(3)From the evolution of LCCC subsystem from 2010 to 2020,the carrying capacity scores of Drive(D),Pressure(P),State(S),Impact(I)and Response(R)subsystems are all increasing year by year,and the growth rate of Driving subsystem(D)is the most significant,with a growth rate of 54.1%.The average carrying capacity score of impact subsystem(I)is always at the highest level,with an average of 407.2.In terms of the contribution of the subsystem to LCCC,the Impact subsystem(I)has the greatest contribution,with an average contribution rate of 48.1%;The contribution rate of the response subsystem is the smallest,with an average of only 8.0%.(4)From 2010 to 2020,the main importance indicators affecting LCCC are:per capita grain output,vegetation coverage,urbanization rate,water conservation,per capita cultivated land area,the Mean Decrease Gini is more than 4000,while the importance of land use intensity,population density,sewage treatment rate and slope is unstable.The indicators affecting LCCC are:Total power of agricultural machinery,land per capita investment in fixed assets,GDP,per capita grain output,annual evapotranspiration,in which the per capita grain output and total power of agricultural machinery decrease most,and the Mean Decrease Gini decreases more than 4500;the index showing an increasing trend is habitat quality,and other indicators are unstable.(5)Based on the marginal effect analysis results of RF model,the index ranges that have obvious positive effects on LCCC are:per capita grain output is(0,5)t/person,urbanization rate is(60,80)%,population density is(0,1000)person/25hm2,and per capita water supply is(0,200)m3/person,the vegetation coverage rate is(40,50)%,the habitat quality is(0.8,1),the carbon reserves is(600,700)t/25hm2,water conservation of(1250,1500)mm/25hm2-a,per capita cultivated land area of(0,1500)people/25hm2.This paper uses index space visualization processing and machine learning algorithm to improve and optimize LCCC evaluation,and puts forward countermeasures and suggestions to improve the LCCC of Dongting Lake Ecological Economic Zone,which will provide decision-making reference for the development and protection of land resources.