Abstract:

Urban spatial analysis should be based on reliable measurements, and the most basic measurement of a city is its size. Defining urban boundaries objectively is fundamental for determining effective city size. In recent years, a number of Chinese and international scholars have developed improved methods of urban boundary identification. Among these, the majority apply vector data that can reflect the spatial organization relationships of entities internal of cities. However, access to these vector data is often limited. In this study, based on existing research a new method of urban boundary identification with remote sensing data as input and using neighborhood dilation and quantification is put forward. Our method takes a spatial neighboring merging approach. By changing the neighboring range of pixels, different numbers of spatial clusters are obtained. An optimal radius can be determined according to the scaling relationships between the neighboring range of pixels and the numbers of spatial clusters. GIS technology is then adopted to define urban boundaries. By applying this method to analyze remote sensing images of the Beijing area, we found the effective range of pixels. Remote sensing data used by this method are characterized by real-time acquisition and easy access. Also, the calculation procedure is straightforward. Thus, in future efforts of urban boundary identification, our new method may provide a complement to existing methods.

Key words: urban boundary identification, urban patterns, city clustering, neighborhood dilation and quantification, fractals, scaling, Beijing