干旱区生态环境知识资源中心

Questions Do the spectral patterns recorded by Landsat Thematic Mapper (TM) for various sagebrush species differ from each other? Could ancillary data derived from topographic variables (elevation, slope and aspect) strengthen the relationship between sagebrush and spectral characteristics? Location Rawlins, Wyoming, USA. Methods Field data on sagebrush species and structural characteristics collected in 2005 were regressed against spectral values and transformed indices derived from Landsat TM data. Step-wise regression methods were used to build (1) a combined model that ignored species and physical differences in sagebrush, and (2) four different models that took those differences in account. Parsimonious models were selected with significant independent variables that accounted for the largest amount of the ground-measured variation in cover. Results In the combined model, a subset of Landsat bands and topographic values accounted for 65% (P < 0.001) of variance in sagebrush canopy cover. However, when separate regression models were fitted based on species type (big and low sagebrush) and height (for big sagebrush only), Landsat data accounted for 7185% (P < 0.001) of the ground measured variance. Conclusions Spectral patterns of sagebrush species and their physical characteristics were distinguishable with Landsat TM data. Transformed indices derived from Landsat data accounted for more variance than the original spectral bands. Future research should focus on testing the relationship between Landsat TM data and sites where sagebrush is present with other vegetation to facilitate landscape and regional mapping of changing sagebrush ecosystems.