Urban Microsimulation For Spatial Plan: Data, Modelling, And Evaluation

This dissertation aims to introduce microsimulation into spatial plans to supporturban planning compilation and evaluation. The spatial plan as an effective measure formanaging urban growth attracts extensive attentions from aspects of geographicalinformation system (GIS), remote sensing (RS), as well as land use&transportationintegrated models. Urban systems as a type of complex adaptive system, however, arecomposed by numerous parcels in the physical space and urban residents in the socialspace. The bottom-up microsimulation approaches, such as cellular automata (CA) andmulti-agent system (MAS), have their opportunities in analyzing and simulating spatialplans. This dissertation will apply GIS, CA, and MAS based microsimulationtechniques to develop microsimulation models for supporting urban spatial plancompilation and evaluation as follows.Firstly, we proposed a data synthesis approach for urban microsimulation models.We disaggregate individual micro data using aggregate data, small-scale surveys andempirical researches to feed microsimulation models to tackle the current data sparsecondition in China.Secondly, we developed two microsimulation models for supporting spatial plancompilation. The first model, BUDEM, is developed based on CA incorporating fourtypes of constraints to simulate future urban growth. The simulation results can beadopted as spatial plan alternatives as well as urban growth boundaries (UGB). Thesecond model, FEE-MAS, is for calculating commuting energy consumption andenvironment impact for urban form in the inner city level. The quantitative relationshipsamong them can then be identified accordingly using a global sensitivity analysisapproach, thus supporting the compilation and evaluation of spatial plan.Thirdly, we conducted two researches for evaluating spatial plan alternatives usingthe BUDEM model. One, spatial plan can be evaluated as possible or impossible interms of the availability of urban policies, which is the reversed process of conventionalurban growth scenario analysis. Two, spatial plan implementation effectiveness isspatiotemporally evaluated for five master plans in Beijing. In sum, several key solutions are proposed in this thesis for introducingmicrosimulation into spatial plan with empirical researches in the hypothetical spaceand Beijing, respectively. The approaches included in this dissertation range from GIS,RS, CA, MAS, spatial analysis, and artificial intelligence, and the spatial plans coverboth master plans and detail plans. Therefore, this dissertation is promising forpromoting planning support techniques for spatial plans in China.