Comparison of Landsat TM and ERS-2 SAR data for discriminating among grassland types and treatments in eastern Kansas

Abstract

The tallgrass prairies of North America are among the most biologically diverse grasslands in the world. The way these lands are managed can have significant impacts on the biophysical and compositional structure of plant and animal communities. Soil stability and other hydrologic factors are also affected by grassland utilization practices. To better understand how changing grassland management practices are impacting their respective ecosystems, we must be able to map and monitor changing land use practices over large geographic areas. We examined the potential of multitemporal Landsat Thematic Mapper (TM) and ERS-2 Synthetic Aperture Radar (SAR) imagery, and the combination of these two data sources for discriminating among three commonly used tallgrass land management practices in areas dominated by cool- and warm-season grass species in eastern Kansas. Our results showed that cool- and warm-season grasses could be discriminated with 90.1% accuracy using the TM data and 73.2% using the SAR data. The three grassland management practices were correctly classified 70.4% of the time using TM data and 39.4% using SAR data. When TM and SAR data were combined, the information contribution by SAR data to the discrimination of grasslands was statistically insignificant. From our findings, we believe Landsat TM data can be used to discriminate among various grassland types at a level of accuracy suitable for land use change monitoring and assessing the impacts of changing government land use policies such as the US Department of Agriculture's Conservation Reserve Program (CRP).