| China is a typical hilly country, hilly cities account for over50%of the total number of cities. Since Chinese cities have developed rapidly in decades, hilly cites also experience rapid urban expansion. Since China is lack of land resources, in order to achieve the sustainable development, it is urgent to allocate urban land scientifically and control the urban scale. Urban expansion has the features of open, nonlinear and uncertainty, so the traditional methods can rarely describe the land use changes precisely. This study chose Agent-based Model (ABM) to define the decision-makers’will, combined with cellular automata model (CA), applied its "bottom-up" dynamic evolution capabilities to simulate urban land changes, so as to give some useful reference for urban planning.Chongqing is a typical hilly city, since being the unique municipal city in western China, its economic strengthen heavily especially after the reform and opening up. Additionally, Chongqing has being the core of western China, it owns rich resources for urban expansion, which make me choose urban Chongqing as the research area. This study concluded the urban expansion feature of Chongqing, based on that, the simulation was done by ABM. In order to reach the goal, the global research about urban expansion was summarized, according to which, the technology method was listed. Since there are lots of expansion metrics, it is necessary to consider the special topography of research area, get the urban expansion degree and its morphology. After doing the qualitative analysis of driving forces, I calculate the parameters of ABM, which consisted of the model basis. The ABM was run for several times to get the relatively accurate result, finally the function of ABM in hilly cities was concluded.In this paper, the urban expansion metrics was expounded firstly, and then the appliance of CA and ABM in urban expansion was introduced. In the description of urban expansion features in1988,2001and2010, the following indexes were used, namely land use change, urban expansion speed, population and elastic coefficient of urban land-use, water and traffic buffer zone, landscape indexes. The driving forces of urban expansion were composed by two parts, the first was macro forces, which consider economic, social, natural and policy factors in Chongqing, while the second focused on micro forces, which according to research area and sample points features, get the concrete parameters for ABM. The year of2001was chosen to be the base period, ABM for urban expansion was processed to simulate the land use in2010, after getting the simulated results, the land use map of2010was used to compare and explained the differences.The result shows:(1) urban expansion in2001-2010is much faster than1998-2001, and such expansion is mostly from farmland; the research area experiences the process of linear development along the valley to traffic.(2) The concept of "groups" is unclear in1988, while in2010the polycentric pattern has been obvious, the core of each main urban area is almost close to the group center.(3) ABM appliance in hilly cities requires to consider natural factors firstly, define the restraint level, so as to escape from the steep slope and keep the mountainous landscape; The fitting accuracy of ABM is88.72%, general kappa coefficient is0.77, so the simulation conforms to the land use changes in2010.The existed deviation between modeling map and land use map was caused by several factors. At first, Chinese government polices play a key role in urban planning, which guides the urban expansion, however, such policies are hardly to quantize in ABM. Also, the threshold and weight in models were able to be more reasonable after doing more analysis. Finally, for further research, Amish level data with higher precision can be used to further explore the ABM, at the same time, optimize the code to reflect the Macro policy is important to improve the simulation precision of ABM. |
