Get reliable science

Grazing lands in the United States consist of approximately 312 million hectares of federal, state, and private land. A
recent review by the National Research Council indicates a need to collect comprehensive, consistent, and uniform data
to evaluate rangeland health. Pacific Northwest National Laboratory is assisting the U.S. Bureau of Land Management,
and working cooperatively with the U.S. Forest Service and the U.S. Natural Resource Conservation Service to address
these data needs using new remote sensing methods designed to characterize large areas quickly and at minimal cost.
Anticipated application for new remote sensing protocols range from conducting national assessments of the ecological
condition of rangelands to monitoring change in the condition of individual grazing allotments. Our work focuses on
the spatial characterization of soil stability and the biotic (plant) community using an array of remote sensing data (fineresolution hyperspectral and multispectral data and coarse-resolution multispectral data), spatial modeling, and
statistical analysis. The general strategy is to fuse low-cost Landsat-7 imagery with geographically explicit physical
data to derive a first-order spatial characterization of soil and vegetation for large areas of rangelands. A key
component of this strategy is the use of hyperspectral data to model/understand the spectral, spatial, and temporal
information contained in both the Landsat and new high-resolution systems. Some preliminary results from summer
and fall 1999 hyperspectral images (3- to 5-meter resolution) taken over test areas in southwestern Colorado are
presented to illustrate the development of a new endmember selection algorithm to unmix hyperspectral data.

source:abstract