Simulation of land use change of Erhai Lake Basin based on ant colony optimization

Abstract: Studies of Erhai Basin indicate that Land use change by human activities in the watershed is the leading cause of regional climate, hydrology, water quality and ecological changes. Therefore, it is necessary to study the relationships between human activities and land use/cover change (LUCC), which is beneficial to offer the scientific decision support for reasonable land planning and land use. Combined with GIS technologies of spatial analysis and using the artificial intelligence algorithm Ant Colony Optimization(ACO) for optimizing, in this paper, we applied the method of Agent-based modeling to establish the spatiotemporal process model of LUCC in order to simulating the dynamic change of land use in whole watershed. Firstly, we made a choice and evaluation for impact factors of land use changes, as well as constructions of the cost of land use change equations in order to construct more reasonable decision rules of land use choice. Then, we have extracted three agents composed by microcosmic and macrocosmic systems which were farm agent, resident agent and government agent. Also, microcosmic rules of decision and behavior were created according to ACO. On the other hand, we have established macrocosmic decision rules according to a resistance coefficient system from the land use planning, as well as a comprehensive decision rule. And then, based on Java language and Repast platform of modeling, the program design, implementation and simulation of model were given in detail. Finally, the validation, calibration and verification of model and analysis of the simulated results were also conducted. Our conclusions from the experiment were three: 1) Ant colony algorithm was more effective in promoting the significant moving and decision of agents, and the simulated results gained better accuracies in both mathematics (up 5.6%) and geometry (up 3.4%) than using a random algorithm. However, the merit of ACO was not suitable for its use in all of land-use types. For an instance, there were no any improvements and sometimes even reduction in accuracy for those land-use types which were less affected by human activities, such as forest, grassland and wetland uses. Thereby, we suggested that ACO was more sensitive to interaction between human and land-use changes, and it was suitable for optimizing human behaviors and decisions of land-use transfer. 2) If the policy on land use was kept unchanged, the major contradiction between human and land in the future ten years should be the persistent reduction of agricultural land (127.64 hm2 cultivated lands and 11.20 hm2 garden lands) and the continuous increase of urbanized land (95.80 hm2). This indicated a big cost of urbanization in Erhai Lake Basin, which also gave a warning of increasing impervious surfaces (IS) produced in future rapid urbanization, and the IS may raise risks of urban non-point source pollution in the future. 3) The fast increasing of wetlands (growth rate of 50.07% was the fastest in the change of all land use types) indicated that the governmental land use policies to ecosystem protection have played a better role in macro control of land resources allocation. From this research, we suggested that the local government should maintain the existing strategies of ecological environment protection to reduce the risk of water pollution in Erhai Lake Basin. The competition in the market economy model of land-resources-commercial should be encouraged to balance the next major conflicts between human activities and land resources.