Resource information
There is an increasing interest in applying fractal analysis to measure the spatial complexity of remotely sensed images. This study provides an evaluation of the effectiveness of three fractal algorithms (isarithm, triangular prism, and variogram) for characterizing urban landscapes in Indianapolis, USA, based on eight satellite images taken by five sensors (Landsat Multispectral Scanner (MSS), Landsat Thematic Mapper (TM)/Enhanced Thematic Mapper Plus (ETM+), Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), and IKONOS). Fractal dimensions (FDs) were computed for both individual spectral bands and classified images to understand the fractal characteristics of land-use and land-cover (LULC) patterns. Results indicated that FD values were sensitive to the computational algorithms used. It was found that the performance of different FD algorithms varied by application. The red and the near-infrared bands were the most spatially complicated bands among all spectral bands of the Landsat and ASTER images as well as the IKONOS images. The most fragmented LULC type was grassland. FD analysis of Landsat-derived LULC maps revealed that the urban landscape in Indianapolis had experienced great change between 1975 and 1991 as well as between 1991 and 2000. It is concluded that the fractals were useful in discriminating the image spatial complexity of various types and show potential in characterizing the temporal changes of urban landscape. However, fractal analysis alone may not be sufficiently effective to characterize urban landscape changes. Detailed quantitative change information is needed to assist in the interpretation of resultant FD values.
