HR: 0800h
AN: B51C-0967    [Abstracts]
TI: Monitoring Rangeland Health With MODIS Vegetation Index Data
AU: * Brown, J F
EM: jfbrown@usgs.gov
AF: SAIC, EROS Data Center, 47914 252nd Street, Sioux Falls, SD 57198 United States
AB: Rangelands cover approximately one third of the land area of the conterminous U.S. These lands supply much of the forage for the U.S. cattle industry. Large area monitoring of these vast expanses of range has proved challenging since most of these lands are in the western U.S., are relatively sparsely populated, and are not well covered by meteorological weather stations. Improvements in the spatial and temporal precision of rangeland health information would be useful both for the cattle industry and for scientific studies of soil erosion, water runoff, ecosystem health, and carbon cycling. Optical multispectral remote sensing data from satellites are an objective source of synoptic, timely information for monitoring rangeland health. The objective of this study is to develop and evaluate a method for measuring and monitoring rangeland health over large areas. In the past, data collected by the Advanced Very High Resolution Radiometer has proved useful for this purpose, however the basic 1 km spatial resolution is not ideal when scaling up from ground observations. This study assesses MODIS 250 meter resolution vegetation index data for this purpose. MODIS data not only have finer spatial resolution and improved geolocation, but they also exhibit enhanced vegetation sensitivity and minimized variations associated with external atmospheric and non-atmospheric effects. Ground data collected over 51 sites in western South Dakota over four years are used as training for regression tree models of range health. Range health maps for the growing season derived from the models are presented and evaluated.
DE: 9350 North America
DE: 0400 Biogeosciences
DE: 0933 Remote sensing
SC: Biogeosciences [B]
MN: 2004 AGU Fall Meeting