Evolution And Simulation Of Land Use Pattern In Mining Area Based On Multi-agents Decision-making

China is the world’s largest coal producer and consumer,and coal resources occupies a special position in China’s energy structure.Shanxi Province is a major coal-producing province in China.Under different temporal and spatial scales,coal mining methods,and disturbance levels,the landscape pattern of mining areas has undergone tremendous changes.Traditional landscape simulation and prediction focused more on the development of natural cover types and the driving effects of socio-economic factors,while ignored the decision-making process and adaptive learning of the subject of land use.This paper taked farmers in the mining area as the first point of view,based on high-resolution remote sensing images to explore the temporal and spatial evolution characteristics of the village land use pattern before and after coal mining.Used the sustainable livelihoods approach framework and participatory rural appraisal to understand the specific environmental characteristics,mining history,land acquisition compensation of Yangquan typical mining area in Shanxi Province and analyzed the decision-making trajectory of farmers and coal mining enterprises in the mining development process;classified the types of farmers,research the livelihood capital structure and sustainable development capacity of different types of farmers.Logistic regression model was used to study the internal mechanism that affects farmers’ decision-making willingness and the important factors that promote the transformation of farmers’ land acquisition willingness at the micro level.Constructed a land use pattern evolution model based on multi-agent simulation to explore in depth how the mining area environment and corporate measures affected the changes in farm households’ land use behavior,simulated the selection of multi-agents under different policy measures and the consequences of behavioral decision-making,and revealsed the interaction mechanism between micro individual decision-making behavior and macro landscape pattern change.The main results were as follows:（1）From the perspective of quantity and space,it presented the increase and decrease range,transfer direction and distribution pattern of land use types in different mining stages.In the process of mining development,the cultivated land has been continuously reduced and converted into industrial and mining land and forest land,which was currently only distributed around residential areas.Under the enterprises’ s investment management decision,industrial and mining land and forest land continued to expand to the north of the village.The grassland showed a trend of increasing first and then decreasing,the area of residential land increased slowly,and water areas appeared in some areas of the village.Complicated mutual conversion processes continue to occur between different land types in the region.（2）The mining economy of Yuanji village has experienced a period of construction and rapid development,stability and transformation.In the largescale mining process of enterprises,the capital combination and livelihood behavior of farmers in mining area showed corresponding changes with the mining development,and farmers gradually showed the characteristics of parttime employment.There were significant differences in the livelihood capital of different types of farmers.The difference in livelihood capital structure resulted in unequal sustainable development of rural households’ livelihoods.The greater the accumulated value of livelihood capital,the stronger the sustainability of rural households’ livelihoods.（3）Livelihood capital has different effects on farmers’ willingness to land acquisition.The willingness decision-making of various types of farmers were quite different,but on the whole,Yuanji village farmers had a positive attitude to the land acquisition of coal mining enterprises.The choice of pure agricultural households’ willingness to acquire land mainly considered the influence of the family’s per capita annual income,cultivated land area and quality;labor ability and education level had a significant positive impact on the choice of part-time farmers;family per capita annual income had a negative impact on the choice of land acquisition intention of non-agricultural farmers.The land acquisition willingness of neutral farmers was transformed under certain capital conditions.If the same capital increases or decreases a unit change,farmers’ willingness might tend to different directions,and then affected land use decision-making behavior.（4）This paper selected farmers and enterprises as the model agents,and reclassified the farmers agents,constructed three land-use decision modules of farmers,enterprises and scenarios.Quantified the agents behavior rules,and established a multi-agent model.Based on the 2012 land use data,simulated the land use pattern of the study area in 2019,superimposed the simulation results with the actual data,and compared them point by point to test the fit of each pixel land type.The accuracy of the model was further verified by kappa coefficient.The results showed that the overall simulation effect of the model was good,and it can be used to predict the main decision-making process and land use changes in Yuanji village.（5）The background of mining development and the decision-making of farmers agents had a great impact on the village land use structure.Combining the characteristics of regional development,constructed three policy scenarios,predicted land use distribution patterns under different scenarios in 2026,and analyzed them in terms of quantity and spatial changes.Under the natural endowment scenario,the landscape pattern is consistent with that in 2019,and the ecological environment maintains the status quo;under the resource development scenario,the village land pattern is characterized by the continuous increase of industrial and mining land,large-scale occupation of ecological space of other land types,and further deterioration of the ecological environment;under the ecological protection scenario,enterprise agents paid more attention to land reclamation and ecological restoration,and the land pattern is characterized by cultivated land,forest land and grassland increased,industrial and mining land decreased,and the ecological environment continued to improve.This research has important theoretical significance for broadening the landscape pattern-process model research and understanding the complex dynamic evolution mechanism of landscape ecosystems.It also has important practical significance for exploring effective environmental management models in mining areas,coordinating enterprise-farmer contradictions,and implementing ecological control strategies.