A statistical study on spatial scaling effects of urban landscape pattern: A case study of the central area of the external circle highway in Shanghai

Based on SPOT remote sensing images and GIS, choosing the central area of the external circle highway in Shanghai as a case study area, the paper studied the spatial scaling effect of the urban landscape pattern with different grains and extents. The conclusions are drawn as follows: (1) The spatial autocorrelation of urban landscape pattern depends on different scales within a certain range of scales, and Moran I and Geary c related to the grains, which characterized the spatial autocorrelation of the urban landscape structure, have the same sensitive points to the scaling at the level of 50 m. (2) The patches of all kinds of landscapes have the fractal character. The fractal dimensions of landscapes respond to scaling differently, and the present nonlinear change trends with grains. The fractal dimensions of landscapes are obviously different at a small grain, but the differences become not obvious with the increasing grain. (3) The landscape diversity are closely linked to the location and the pattern of human activities, especially to economic and social activities. Due to the high land cost in the urban center, the dominant landscapes are mainly for business and culture, and their patches have the characters of high congregation and high fragmentation. While agricultural landscapes with low economic benefit can only be located at the fringe or outskirt of the urban area, and they have the characters of simplex, larger patches and less fragmentation. (4) The landscape diversity depends on spatial scale. With the increasing extent, the Shannon diversity index (SHDI) increases and the spatial pattern of landscape varies dramatically. At 0.5 km extent, the maximum of diversity is in the center of the urban area. The landscape diversity is distributed with a ring mode up and down from the center to the outer. With the increasing extent, the maximum of diversity moves to the urban-rural transition zone where landscape types change dramatically. (5) The semivariogram discloses the spatial variance character and internal mechanism of landscape diversity. At a small scale, the spatial variance of diversity is more complicated. The spatial heterogeneity, which is caused by spatial autocorrelation, contributes a lot to the total spatial heterogeneity of terrestrial ecosystem, while the spatial heterogeneity caused by random factors (scale and measure error) contributes less. The increasing scale washes off the detailed variances in a fine scale. The coarse scale may result in more nugget effect and less contribution, which is caused by spatial autocorrelation.