基于约束性CA模型的城市开发边界划定方法

为了解决土地利用总体规划中划定城市开发边界问题,该文研究了基于约束性元胞自动机(cellular automata,CA)模型的城市开发边界划定方法,以辅助规划方案的编制。将CA模型与城市发展适宜性、规划指标约束、发展空间约束等相结合,在Arc GIS系统平台上,利用Python语言实现了城市开发边界划定系统工具。基于河南省巩义市中心城区2005和2010年2期数据,模拟了2015年的城市边界,并与2015年真实数据对比验证。结果表明,城市发展总规模和新增建设占用耕地规模均不超设定的规模指标2 865.92和282.80 hm~2;新增用地在空间上避让了基本农田等限制区,达到了空间限制性要求;以城市土地利用适宜性为转换规则,体现了规划的"上下结合"原则;城市用地模拟精度总体达到93.09%,Kappa系数90.41%。该方法可用于土地利用总体规划中划定城市开发边界,为规划方案的制定提供辅助决策支持。

Partition method of urban development boundary based on constrained cellular automata model

Abstract: In order to solve the problem of urban development boundary delimitation in general land use planning, this study has designed the cellular automata (CA) model for identifying urban development boundaries. To maximize public interest, the general land use plan is made to consist of a system of planning restriction indicators at different spatial levels, embodying the top-down planning pattern. However, existing research of urban evolution CA model is mainly focused on its transition rules. The states of one cell are determined by the threshold value, which is hard to realize the constraint for urban land size. That is to say, the previous studies could not satisfy the needs of the urban development boundaries'' delimitation. The CA model, designed in this study, makes a greater contribution to satisfying the planning constraints for the quotas of incremental urban land and incremental urban encroachment on cultivated land. The rationale is enabling CA available for land use planning through adding relevant planning constraints. The planning constraints for urban development boundaries comprise the quotas of land use and incremental urban encroachment on cultivated land. Transition probability of urban land depends on the factors including urban development suitability, neighborhood principles and urban expansion restrictions. Urban development suitability is the determinant of inherent urban development. This suitability is normally represented in the form of probability matrix in the CA model. The urban land transition probability matrix, based on the cellular scale, is calculated by the logistic model. The dependent variable is whether the type of cellular land use is changed into urban land. The changing cellular ones are assigned with the value 1, otherwise the value 0. The independent variables include factors concerning the aspects of terrain, location, fundamental infrastructure, and so on. In this paper, Moore 3×3 window is selected as the CA neighborhood model. Cellular status value is binary. That is, cellular type of urban land is assigned with the value 1, otherwise the value 0. This research uses the land use maps of 2 phases for calculating and making statistics of the probability of cellular land type changing into urban land. Spatial constraints of urban development boundaries mainly concern the eco-environmental aspect, basic farmland restriction and competition of various land use types. The CA model applies the selection strategy based on the maximized probability. The CA model of urban development boundary delimitation is established based on the Python language, referring to the Arcpy and Numpy class databases. The software interface is a user-defined tool of ArcGIS, which is easy to be invoked in Geographic Information System (GIS). The Gongyi City in Henan Province is chosen as the case study area to validate the designed CA model, based on the remote sensing images of 3 phases (i.e., the year 2005, 2010 and 2015). The validity of the simulation results has been verified and manifested as the 3 following aspects: 1) The quota requirements of urban land use size (2 865.92 hm2) and incremental construction land through occupying cultivated land (282.80 hm2) have been satisfied; 2) The requirements of space expansion restriction have been met, such as protecting basic farmland and ecological land; and 3) The CA model applies the urban development suitability as the cellular transformation rule, embodying the top-down planning principle. The results of simulation have a higher accuracy. The accuracy is 93.09% and the Kappa coefficient is 90.41%. In a nutshell, the CA model designed in this study meets the requirements of defining the boundaries of urban development in general land use planning and can provide decision support for the plan-making system of urban development boundary.