Measuring changes in landscape pattern from satellite images: short-term effects of fire on spatial diversity

Satellite images can be considered a very convenient tool to measure landscape patterns since they provide a digital mosaic of the spatial arrangement of landcovers. This paper describes a practical application of several indices frequently used in landscape ecology to measure changes in the spatial mosaic as a result of large fires. These indices were computed from Normalized Difference Vegetation Index images of Landsat Thematic Mapper and Advanced Very High Resolution Radiometer data acquired before and after a large fire affecting the Mediterranean coast of Spain in 1991. The following landscape pattern indices were computed: standard deviation of spatial profiles, spatial auto-correlation, texture, number of patches, diversity, compactness and fractal dimension. To avoid potential bias in the results, these indices were calculated for several scenarios. The modification of the study area limits and spatial resolutions of the images, the measurement scale and the number of categories were tested in this study. All the indices computed showed a short-term increase in spatial homogeneity after the fire. Spatial auto-correlation increased, while texture, number of patches and mean diversity decreased. Compactness was reduced and fractal dimension increased, which also suggests that the fire created more elongated and complex forms. These effects were observed in both high- and low-resolution images.