The Changes Of Characteristic Scales Of Landscape Pattern And Spatial Differentiation During Urbanization

As a complex ecosystem, landscape structure and function of a city will generate sequence of adaptive changes with the regional population growth and the development of social economy, to meet the needs of the regional social and economic development. With its powerful ability to penetrate and radiate, urban landscape not only determine the structure and pattern characteristics of its own, but also significantly restrict the structure of the surrounding landscape of other types, causing landscape evolution and pattern changes in peri-urban especially suburban. And the changes in the pattern and landscape constitute the urbanization process. Studying me temporal and spatial dynamics ot landscape pattern of urbanization is an important way to understand urbanization process and its driving mechanism. Investigating characteristic scale of the landscape will help to select the appropriate measurement scale to reveal the regularity of landscape pattern, and also help to reveal patterns and dynamics mechanism of landscape change.The Yangtze Delta has the highest speed in urbanization process in China. Since the reform and opening up, with the gradual progress of urbanization, human-induced land use and land cover change (LUCC) changed dramatically. As one of the core cities in the Yangtze Delta, the urbanization process of Nanjing has important reference value for the development of other cities in the Yangtze River Delta and even the the country. In order to explore the change of the characteristic scale structure of the landscape in urbanization process, in this paper, with 3 landsat TM remote sensing images of Nanjing in 1988,1998 and 2006, using wavelet analysis and local quadrat variances, we studied the landscape structure of four suburb area:Pukou, Hexi, Jiangning, and Xianlin from 1988 to 2006, which was also compared with centre city. The results are:By wavelet analysis, we discovered that from 1988 to 2006, downtown and all suburban districts had 4 characteristic scale domains, and there was no obvious relationship between the level of nesting. During the study period, every scale domain of the 4 districts had consistent trend, there was just some differences in the rate and magnitude of change. Besides, variation coefficient of the smaller scale domain(1,2) was larger than the large-scale ones(3,4), revealing that differences between the scale within the larger scale was less than the small-scale, and the larger scale also had a higher degree of development. Comparing coefficient of variation of the same scale domain in different times of Pukou, Hexi, Jiangning, and Xianlin, we find that heterogeneity between medium and large-scale of different locations showed rhythmic changes with the urbanization process.Atter comparing cnaracier(?) time, we found that scale domain 4 of each location in three periods had no significant difference, however, in three other small scale, there were differences between the downtown and suburban districts. The small and medium-sized characteristic scale of the city center was smaller than the other location, showing that small and medium-sized structure of the suburban was thicker than the city center. And the greatest characteristic scale in the districts did not show significant differences.After comparing characteristic scale field of same location in different time, we found that under certain conditions characteristic scale might show the phenomenon of inheritance. But it sometimes only occured in the scale of individual domains, and because of the differences in some locations, it would appear in different scale domains. Meanwhile we discovered mutation and stability in the process of landscape evolution, and the stable surface w regarded as metastable phase of the landscape.Local quadrat variance results showed that the level of landscape heterogeneity of other districts was generally increasing or first decreased and then increased. Downtown showed variation different from suburban, characteristic scale detected was also relatively significant than suburban districts. This might be related to the degree of urbanization downtown. At the same time, comparing with the number of characteristic scale from wavelet analysis, the resulting number from local quadrat variances analysis was significantly less than the former. Although local quadrat variances analysis could detect the landscape structures overall average patterns, but the landscape patterns distributed on the multi-scale did not find.