ZHAO Ping, FENG Xue-Zhi
It is useful for the sustainable development of city to analyse spatial characteristics and evolvement of urban system. In this paper, spatial structure and pattern of urban system in Shaoxing based on remote sensing and GIS are analyzed by fractal geometry. The rule of evolvement is discussed by comparing the data in 1997 with that in 1984. Remote sensing techniques can provide the necessary data rapidly and periodically for the spatial characteristics and evolvement study. GIS is an effective tool to analyze the spatial characteristics and evolvement with the strong spatial analysis functions. Firstly,the study area including Shaoxing City and towns is cut from TM images in 1984 and 1997 after geometric correction based on the 1:50,000 topographic map. Secondly, the spectral characteristics of the urban system and background on the TM images are analyzed by samples statistics. Thirdly, a model is established based on the difference of spectral values between them to extract the urban information. Fourthly, the result of extraction is transformed to vector map in order to correct some errors and imported to GIS. Then the aggregation dimension, spatial correlation dimension and fractal dimension of pattern in different times are calculated one by one based on the attribution table of GIS. The results of analysis suggest that both spatial structure and pattern of urban system in Shaoxing are obviously fractured. The aggregation dimension D denotes the aggregation or dispersion state around the centre of the urban. If D<2, the density of towns is attenuating from the centre to the surrounding. If D=2, the towns are distributed evenly around the centre of the city. If D>2, the density of towns is increasing from the centre to the surrounding, which is not natural. The correlation dimension D denotes the uniform extend of the towns aggregation. If D→0, the distribution of towns is centralized. If D→1, the towns are aggregating to a geographic line. If D→2, the towns are distributed evenly around any town. The fractal dimension D of spatial pattern denotes its complexity and stability, which can discover the trend of the urban evolvement. It varies from 1 to 2,and the greater the D, the more complex the distribution of towns. If D<1.5, the shape of the urban system is becoming more simple. If D=1.5, the stability is worse,and the urban system is in the Brown movement. If D > 1.5, the shape is more complex and instable. If D=2, one place on the plane is filled by the graph of the town. According to the results of the calculation, the aggregation dimension of Shaoxing urban system is smaller than 2, which suggests that the density of towns decreases gradually around the centre. The spatial structure is optimized. The correlation dimension is nearly 1, so the distribution of towns is not proportional, which tend to aggregate to the traffic routes at the same time. As far as the evolution of urban system is concerned, from 1984 to 1997, the towns tend to be distributed evenly around the city, and the fractal dimension of the spatial structure is decreased. Every town expands around its center gradually, and the spatial correlation is unchanged on the whole. But the spatial pattern of urban system becomes more complex, unstable and random. Especially, with the increasing of economy in Shaoxing, the city develops more quickly, and the spatial pattern is more complex and instable than other towns. Lastly,some suggestions about the development of Shaoxing are brought forward based on the conclusions.