Evolution And Prediction Of Landscape Pattern In Coal Mining Area Based On Multi-agent Simulation

China is the largest producer and consumer of coal in the world,and Shanxi Province is the most important energy and heavy chemical industrial base in China with a long history of coal mining.The landscape pattern of mining areas is subject to dynamic spatial and temporal evolution due to various factors such as mining scale,mining method and stakeholders’ behavior disturbance,etc.Simulation and prediction of the landscape pattern of mining areas can provide an important scientific basis for effective land planning and reclamation.The traditional landscape pattern simulation model focuses more on the study of "land" in the human-land coupling system than on the study of "human-land",and ignores the response of the ecosystem to the change of land use behavior of stakeholders caused by macro policies.Based on high-resolution remote sensing images and participatory community development assessment research,this paper analyzed the increase and decrease of different landscape cover types and the transfer trend of different landscape cover types in Liuwan mining area,a typical mining area in Shanxi Province,the dynamic evolution of the spatial distribution of the landscape pattern and the fluctuation of the landscape pattern index from a quantitative and spatial perspective.The trajectory of land use decision-making behavior of government,industrial and mining enterprises,and farmers’ behavior with the life cycle of the coal industry was explored.The feedback relationship between government,enterprise and farmers in the evolution of the landscape pattern of the mining area was analyzed,and the key influencing factors affecting the land use decision of multi-agents in the mining area were screened and the rules for land use decision were constructed.Based on the theory of multi-agent system,a coupled government-enterprise-farmers model of coal mine area landscape pattern evolution(MAS-LUCC)was constructed in Liuwan mining area to simulate how governments’ macro-regulation,enterprises’ interest pursuit and farmers’ livelihood development shape the change of mine landscape pattern and predicte the land use decision-making behavior of multi-agent and the ecosystem response under different development scenarios.The interaction mechanism between individual decisionmaking behavior and landscape pattern evolution was revealed.The main results of this paper are as follows:(1)Under the intervention of mining activities,a complex dynamic transformation process among landscape cover types has been occurring in the Liuwan mining area over the past 15 years,with a large amount of arable land transformed into industrial and mining construction land,and grassland,as a transition type of transformation among other landscape cover types,basically presenting a decreasing evolutionary trend.The industrial and mining land distributed in the north and west-central part of the research area area widely in the form of strips,and expanded to the surrounding areas,and the shape of the newly expanded areas tended to be regular,with low landscape heterogeneity in the area.(2)The development of mining economy in Liuwan mining area has experienced a period of rapid development,stable development and transformation.The decision-making behavior of various land use agents is closely related to the landscape pattern response of the mining area ecosystem.Mining area governments,industrial and mining enterprises,and farmers made land use decisions based on the characteristics of agents and external influences,and then obtained the decision-making preferences of different types of agents for different types of landscape cover.Based on the principle of maximum spatial efficiency,government agents restricted and guided the land use decision-making behavior of other types of agents.The industrial and mining enterprise agent took the maximization of investment income as the decision-making rule guidance,and produced the development behavior on the plot unit after the decision.Farmers made bounded rational decisions based on their own development needs.The three types of agents complemented each other.Through the mutual feedback mechanism with the environmental system,the landscape pattern of the mining area was transformed,and based on this,the land use decision-making rules of various types of agents were quantified.(3)We proposed the MAS-LUCC model for the evolution of landscape pattern in Liuwan mining area,simulated the landscape pattern in 2020 based on the 2005 landscape pattern image of the research area,overlay the simulation results with the actual distribution image,verified the fitting degree of the landscape type of each plot unit through point-bypoint compariso,and tested the accuracy in combination with Kappa coefficient.The validation showed that the model simulation results were reliable and could be used to simulate and predict the land use behavior and ecosystem response of agents in the coal mining area.(4)The mining agent system actively adapted to the surrounding environment and adjusted its own behavior rules.This continuous interactive and nonlinear process was presented as the result of the ecological response of the environmental system.Based on the characteristics of economic and social development in the research area and the different decision-making tendencies of various agents,we set up three mining area development scenarios by correcting the driving factors and key parameters to reveal the dynamic change trend of mining area landscape pattern in 2030 under different policy guidance directions and agent decision-making.Under the natural development scenario,the landscape pattern basically maintained the current situation.Under the resource development scenario,the urban,industrial and mining construction land continued to expand,and a large number of other landscape cover types were expropriated,resulting in the deterioration of the ecological environment.Under the ecological protection scenario,the government implemented strict environmental protection constraints,industrial and mining enterprises carried out ecological restoration and environmental governance on the damaged surface,the speed of land acquisition slowed significantly,the area of forest land and grassland increased,and the landscape pattern of the mining area was optimized.This research introduced the idea of multi-agent into the study of landscape pattern in mining areas,which could help to clarify the relationship between land acquisition compensation,macro policies,economic and social development and livelihoods of landless farmers,and then guide the rational allocation of land resources,alleviate land competition conflicts among stakeholders in coal mining areas,improve the science and effectiveness of spatial planning policies,and explore a harmonious development model for economic construction in mining areas.