Simulation Analysis And Stability Evaluation Of Forest Landscape Patterns In Guangyun Forest Farm In Guangxi

Forest landscape is mainly composed of forest ecosystem.By studying and analyzing forest landscape,we can know more about the structure and function of landscape,and make it clear the objects of regulation and control to provide theoretical support for forest resource management.Based on the theory of forest management and landscape ecology,we took Guangyun Forest Farm in Guangxi as the study area,and classified the landscape types of the area with AcrGIS 10.2,Fragstats 4.2 and Python programming.On this basis,by using the landscape metrics method and multiple models,we analyzed the forest landscape pattern of the study area and the impact of natural environment on forest landscape pattern.Finally,the stability of forest landscape pattern was evaluated to provide a theoretical basis for forest management of the study area.The main conclusions are as follows:(1)In light of land types,dominant tree species and local situation of production,the forest landscape types in the study area were classified into nine categories:non-forest landscape,other forest landscape,broad-leaved forest landscape,fir forest landscape,pine forest landscape,unforested forestland landscape,shrub forest landscape,economic forest landscape and bamboo forest landscape.(2)A new method of selecting landscape index was proposed based on statistics,and achieved by Python programming.In addition,the number of type-level index and landscape-level index we selected in this study were 8 and 6,respectively.(3)The type-level index and landscape-level index were used to analyze the forest landscape pattern.At the level of type,the main landscape in the study area is broad-leaved forest landscape,which occupies an absolute advantage.On the other hand,the degree of aggregation is higher than other landscape types,but the shape of landscape is simple and seriously disturbed by human beings.At the level of landscape,the landscape heterogeneity of the study area is low,and the degree of aggregation is high.Besides,the connectivity of the study area is good,and the fragmentation is low.(4)K-Nearest Neighbor(KNN)regression,random forest and Gradient Boosted Decision Tree(GBDT)regression were used to simulate the impact of natural environment on forest landscape pattern in the study area.The results show that the GBDT method has the best accuracy among the three methods,whose coefficient of determination(R2)is 0.773 and Root Mean Square Error(RMSE)is 0.051.The second is random forest regression method,whose R2 is 0.752 and RMSE is 0.054.KNN regression method is the worst,and R2 is 0.623,RMSE is 0.065.(5)Principal component analysis,Kriging interpolation and natural breakpoint classification were used to evaluate the stability of forest landscape pattern in the study area.The results show that the proportion of stable area is 71.91%,and mainly distributed in the middle and southeast of the study area,where mainly are broad-leaved forest landscape with rich vegetation and better protection.The proportion of unstable area is 28.09%,and mainly distributed in the southwest and the boundary of the study area.The tree species of unstable area are single and seriously disturbed by human activities.The study shows that the overall forest landscape stability of Guangyun Forest Farm in Guangxi is good,but the protection of forest resources needs to be strengthened.