The characteristic scales of the urban landscape in the Nanjing metropolitan region
南京市城市景观的特征尺度

The purpose of this scale-related study is to find the underlying mechanisms of ecological phenomena by using appropriate spatial-temporal scales.Scale holds the key to understand pattern-process interactions,but due to limitations of existing theories and methods,in-depth studies on scales in landscape ecology have been limited,especially in urban ecosystems with complex structures.Recently,many new methods were developed and applied in landscape ecology dealing with spatial-temporal processes,e.g.,semivariogram analysis,point pattern analysis,and wavelet analysis.Wavelet analysis can associate a spatial or temporal pattern with different scales and locations and is effective in characteristic scale detection.With one scene of SPOT-5 imagery,one dimensional wavelet analysis was conducted to study the structural features of the urban landscape in the Nanjing metropolitan region.One transect traversed the metropolitan in the main diagonal direction of the city,and three sections named A,B and C were distinguished which corresponded to the northern suburban,urban and southeast suburban area,respectively.To check the validity of main diagonal sampling,three short transects from each of the three areas were randomly placed near the main diagonal line.The results were then compared with the corresponding results along the main diagonal direction,and no remarkable differences were found,suggesting that we could use the results derived from the main diagonal line to represent features of the landscape.Three common mother wavelets,Haar,Morlet and Mexican Hat,were used.The results indicated that the Morlet wavelet was most suitable for detecting characteristic scales of the urban landscape.The Mexican Hat wavelet could be used for the same purpose but it lacks the ability to reveal some details of scales,while Haar wavelet was not capable of detecting characteristic scales with continuous imagery data.The one-dimensional Morlet wavelet transform showed that the landscape of both northern and southeastern suburban areas had one characteristic scale corresponding to the average grain size of cropland patches,362 m and 446 m,respectively.While for the urban area,multiple characteristic scales(about 292,835 m and 2200 m) were distinguished,which corresponded to architectural clumps,small blocks and large blocks,respectively.The results revealed that a grain-nesting structure existed in the urban area,in the order of architectural clumps,small blocks,and large blocks.This nesting structure is significantly more complicated than that in suburban areas.