Multi-scale Study Of Urban Growth And Landscape Change In Hangzhou Metropolitan Area

Since the reform and opening up, rapid urbanization has become a prominent phenomenon influencing social economic development in China. Research on spatial-temporal pattern, dynamics, driving forces and modelling of urban growth has long been a focus in urban studies. In practice, it is urgent to study how to guide urban growth to reduce the negative impact of excessive growth in rapid urbanization areas at present.According to hierarchy theory, scale is now increasingly recognized as a central concept in the description of the hierarchical organization of our world. In urban studies, scale can be essential to understanding urban complex systems. In general, scale can be threefold: temporal, spatial and decision making, where spatial scale can be further divided into grain and extent. The variations of scale have been shown to affect the analysis of urban landscape pattern and simulation of landscape change. Therefore, Multi-scale framework is integrated into the modelling to deal with scale-dependence in recent studies. The grain and extent effect of landscape pattern have been focused for a long time, but the sensitivity of urban growth model such as Cellular Automata to scale relatively lack by now. Especially, rare attention has been paid to the scale issue from the perspective of master plan or land use planning and management. Thus this dissertation specifically explored scale-dependence of spatial determinants of urban growth, and a multi-scale method of urban landscape pattern analysis as well as urban growth model's sensitivity. This dissertation took Hangzhou metropolitan area as a case study based on remote sensing and social economic data.The flowchart of the dissertation included three parts, namely scale, methods and results. This dissertation started from the definition of scale concept and put forward the way to study spatial extent. Then spatial determinants, dynamics and simulation were explored using quantitative methods. At last, findings and results were concluded. Accordingly, seven chapters were divided in this dissertation as follows:Chapter one was the introduction of research background and objectives. Chapter two reviewed the research of scale issue in urban growth and landscape change, and then pointed out the necessity of scale study in landscape analysis and simulation. Chapter three introduced the overall schedule of the research, including data collection and processing and hierarchy theory and diffusion-coalescence hypothesis employed in the study.Chapter four analyzed the spatial determinants of urban growth from a multi-scale perspective. According to relative studies, research questions of urban growth pattern can be divided into three subquestions: change probability, change density and change intensity. The dissertation put forward another subquestion of change sequence in the sense of temporal scale. Then exploratory spatial data analysis and logistic regression model were employed to uncover the relationship between spatial determinants and urban growth pattern in multiple scales. In the above model, social economic factors, biophysical factors and planning policies were selected as exploratory variables while change probability, density and sequence were dependent ones. The results showed that minor roads, neighbouring developed areas and spatial policies were influential across all scales, but major roads, neighbouring available areas and population change varied with scale in Hangzhou. Proximity to major roads and industrial center, neighbouring developable areas and population density increase were the major concerns for large-scale urban development in Hangzhou. Instead, proximity to minor roads and neighbouring developed areas were becoming the major concern at the scale of change probability and sequence.Chapter five analyzed the landscape pattern change in multiple scales. In urban area level and land use region level, analysis of landscape gradient and infill process could be used to infer change probability and change density respectively. In landscape pattern analysis, landscape fragmentation metrics should be used in urban area level while landscape patch shape and connectivity metrics in urban land use level instead. Five urban-rural gradients along the corridors of urban expansion provided useful information about urban patches. It was seen from these gradients of Mean Patch Size in urban class decreased from near the urban core to countryside and multiple peaks of MPS in new developed area. Results of Landscape Shape Index didn't represent the same dramatic decrease in suburban areas. Fraction dimension of 1.76 reflects perfected infill capacity of urban growth when comparing the shape of largest urban patches across time. Furthermore, the potential cause might be low proportion of infilling growth or spontaneous growth and high proportion of edge growth. Based on the diffusion and coalescence hypothesis, it could be inferred that coalescence followed by the diffusion of 1996 towards a saturated urban landscape.Chapter six presented scale sensitivity of SLEUTH model (i.e. Slope, Landuse, Excluded, Urban extent, Transportation, Hillshade), simulated and evaluated different scenarios produced by SLEUTH. The exploration results of scale sensitivity of SLEUTH showed that spontaneous growth was overestimated due to a lot of undesirable factors and low-density development affected the calibration results when scaling up in spatial extent. So it was reasonable to reduce spread coefficient and define excluded layer in order to prohibit excessive growth. Then SLEUTH model was used to predict four future growth scenarios in Hangzhou, including Maintain Status Quo, Expanded Roads, Moderate Farmland Protection and Compact City. The alternative scenarios were assessed by the proposed aspects such as amount and spatial distribution of urban growth, and consistence with master plan. The results showed that urban growth in MSQ scenario would be faster than allowable amount of master plan or land use plan and farmland would be more fragmented. In particular, urban form would be compact through guiding urban development to hot-spot areas in ER. Furthermore, edge growth could be promoted by MFP and CC. Finally, abrupt policy change should be further considered in SLEUTH model in order to promote its accuracy.Above all, Chapter seven drew a conclusion and briefly discusseed policy implication, and also pointed out the shortcomings.