LUCC Modeling In The Mountainous Region Of Tropical Northern Margin Based On CLUE-S Model

The quantization of policy factors is the difficult to simulate land use sudden changes.Due to the scale effect of land use patterns,exploring the optimal grid size can serve the optimization of LUCC simulation studies.Land use/land cover change(LUCC)in the tropics seriously affects global biodiversity and its ecological services.Clarifying the spatial-temporal process of land use change and simulating the future land use change trend will provide a more reliable basis for rational planning in this region.In the present study,the Mengla County of Xishuangbanna,located at the northern edge of tropical Asia is taken as a study case.Spatio-temporal patterns and processes of LUCC in Mengla County of Xishuangbanna,located at the northern edge of tropical Asia,from 1994 to 2014 were examined using a post-classification change detection technique based on Landsat TM/OLI images acquired at five time points(e.g.,1994,1999,2004,2009,and 2014).The Maximum likelihood method,Support vector machine,Object-oriented method,and random forest classifier was used to classify land use/land cover types in 1994 through integrating spectral,textural,and topographic information.The optimal classification method is obtained and the images of the remaining epoches are classified by this method.CLUE-S model is used to simulate the land use in Mengla County based on the classification results in 1994,and the grid scales of 60m×60m,90m×90m,120m×120m,150m×150m,180m×180m,210m×210m,240m×240m,and 270m×270m are used to simulate the land use in 2014 for obtaining the optimal grid scale for simulating and adjusting the model parameter.At the same time,by adding limiting factors of the built-up land and removing the built-up land to quantify the influence of sudden changes in simulation.Finally,the LUCC in Mengla County in 2030 was predicted under the optimal grid scale and the parameter setting.The results show that(1)The spectral,texture and topographic information were classified by random forest method.,and the overall accuracies of land use/land cover classification reached 90.13%–97.90% with corresponding Kappa coefficients of 0.84–0.96.By analyzing the classification results,the LUCC in Mengla county was characterized by the decrease of forest and the rapid expansion of plantation land;(2)The forests around protected areas are gradually transformed into plantations and the ecological environment is destroyed;(3)The optimal simulation accuracy is under grid size of 150m×150m grid scale;(4)Quantifying policy factors makes the simulated Kappa of built-up land improve about 0.15;(5)The land use change of Mengla County until 2030 shows a trend of increase in forests and decrease in plantations,while the restoration of forests around protected areas was poor.