The analysis of spatial-temporal dynamics of urban landscape structure: A comparison of two petroleum-oriented cities.

摘要

This dissertation integrates remote sensing, spatial metrics, and urban modeling to map, compare, and model the urban process in two petroleum-oriented cities, Houston in the United States and Daqing in China. The primary objective of my research is to understand the relationships between the human behavior and natural environments under different socio-political contexts at a regional scale. Accordingly, there are two major research foci in this dissertation: (1) improving the accuracy of remote sensing classification in the highly fragmented urban landscapes, especially the human land use classes such as the residential area and the industrial/commercial area; and (2) testing the utility of sub-pixel information for a self-organized model towards a framework to support and improve urban modeling. Although a variety of studies have focused on the urban remote sensing and urban modeling, this research is the first investigation on the relationship between sophisticated remote sensing techniques and urban process models within socioeconomic dimensions. The results of this study provide evidences on the threatened natural landscape and environment deterioration during the urbanization processes in two petroleum-oriented cities and suggest that the utility of sub-pixel techniques improves the accuracy of both mapping and modeling in urban research through the cost effective satellite imagery.