Simulation of land use change using multi-agent model

To put forward feedback information of the scenario simulation model for the references of land use planning decision, we reviewed and analyzed the land use characteristics (1996-2009) in Xinbei district of Changzhou, and extracted three important types of land use: cultivated land, construction land and rural residential land. Typical relation analysis method and spatial gridding Logistic regression analysis method were used to made quantitative analysis of these three types of driving forces and obtain the main influential driving forces for each land use. Based on the analysis of land use characteristics and driving forces, the study built land use scenario characteristic system from three aspects which included political characteristics, global characteristics and spatial characteristics, and expressed land use scenario characteristics quantitatively. The MAS technology was introduced to construct an agent-based land use scenario simulation model, and using the development platform of Eclipse+Swarm, realize the land use scenario simulation model. The results showed that the changes of structure in Xinbei district were increasingly complicated; reduction of agricultural land contributed most to the increase of urban land. In term of spatial arrangement, the changes of land use types were the greatest in the study areas of eastern or southeastern parts along the administrative region boundary, or the Yangtze River catchment. 1) The results showed that fitting degree of the spatial global simulation was high, that of construction land, rural residential land and cultivated land was 78.45%, 99.85%, 98.35% separately, and fitting simulation of spatial variation for construction land, rural residential land and cultivated land was 55.18%, 0.02%, 69.11% separately. 2) The study set three scenarios of socio-economic development (with high, medium and low developing speed) and simulated the land use changes from 2006 to 2020. The results showed that in the three scenarios, the construction land would increase 3788.44, 2922.83 and 2114.64 hm2; the cultivated land would increase 1106.60, 858.61 and 592.35 hm2; the rural residential land would increase 5378.72, 4817.03 and 4238.97 hm2. In the scenario I, the urbanization would go fast, the urban-rural land convertion would be obvious, and the renovation and reconstruction of the rural residential land would enhanced; in the scenario II, the urban expansion and population growth would slow down, and the cultivated land loss would be decrease; in the scenario III, the urban-rural land convertion would decrease obviously, and the cultivated land loss would be prevented effectively. 3)The compactness analysis under the three scenarios showed that the compact degree would be low in the scenario I (construction land, rural residential land and cultivated land was 0.0059, 0.0036, 0.0024, separately)and in the scenario III (construction land, rural residential land and cultivated land was 0.0057, 0.0031, 0.0024, separately), but high in the scenario II(construction land, rural residential land and cultivated land was 0.0061, 0.0034, 0.0022 separately); the rural residential land would be dispersive in the scenario III and the renovation of the rural villages would be the least in the scenario II.