A multi-scale analysis of landscape statistics

It is now feasible to monitor some aspects of landscapeecological condition nationwide using remotely-sensed imagery andindicators of land cover pattern. Previous research showedredundancies among many reported pattern indicators andidentified six unique dimensions of land cover pattern. Thisstudy tested the stability of those dimensions and representativepattern indicators across different types of land cover maps. Themaps were derived from Landsat Thematic Mapper images of theTennessee River and Chesapeake Bay watersheds, and they differedin resolution, number of attributes, and method of delineatinglandscape unit boundaries. A multivariate analysis of patternmetrics was conducted separately for each map, and the resultswere then compared among types of maps. Measures of land coverdiversity, texture, and fractal dimension were more consistentthan measures of average patch shape or compaction among the landcover maps.