HR: 0800h
AN: IN21C-1197    [Abstracts]
TI: Using Current and Historic Climate Data and Bayesian Belief Networks to Predict Optimum Satellite Image Acquisition Periods for Detecting Cheatgrass on the Snake River Plain, Idaho
AU: * Rope, R C
EM: Ronald.Rope@inl.gov
AF: Idaho National Laboratory, PO Box 1625-2213, Idaho Falls, ID 83415 United States
AU: Ames, D P
EM: amesdani@isu.edu
AF: Idaho State University, Dept. of Geosciences Idaho State University, Pocatello, ID 83209 United States
AU: Jerry, T D
EM: Jerry.Tagestad@pnl.gov
AF: Pacific Northwest National Laboratory, 3110 Port of Benton Blvd., Richland, WA 99354 United States
AU: Cherry, S J
EM: Shane.Cherry@inl.gov
AF: Idaho National Laboratory, PO Box 1625-2213, Idaho Falls, ID 83415 United States
AB: Invasive plant species, such as Bromus tectorum (cheatgrass), cost the United States over $36 billion per year and have encroached upon over 100 million acres while impacting range site productivity, disturbing wildlife habitat, altering the wildland fire regime and frequencies, and reducing biodiversity. Because of these adverse impacts, federal, tribal, state, and county land managers are faced with the challenge of prevention, early detection, management, and monitoring of invasive plants. Often these managers rely on the analysis of remotely sensed imagery as part of their management plan. However, it's difficult to predict specific phenological events that allow for the spectral discrimination of invasive species using only remotely sensed imagery. To address this issue tools are being developed to model and view optimal periods to collect high spatial and/or spectral resolution remotely sensed data for refined detection and mapping of invasive species and for use as a decision support tool for land managers. These tools involve the integration of historic and current climate data (cumulative growing days and precipitation) satellite imagery (MODIS) and Bayesian Belief Networks, and a web ArcIMS application to distribute the information. The general approach is to issue an initial forecast early in the year based on the previous years' data. As the year progresses, air temperature, precipitation and newly acquired low resolution MODIS satellite imagery will be used to update the prediction. Updating will be accomplished using a Bayesian Belief Network model that indicates the probabilistic relationships between prior years' conditions and those of the current year. These tools have specific application in providing a means for which land managers can efficiently and effectively detect, map, and monitor invasive plant species, specifically cheatgrass, in western rangelands. This information can then be integrated into management studies and plans to help land managers more accurately and completely determine areas infested with cheatgrass to aid in their eradication practices and future management plans.
DE: 0430 Computational methods and data processing
DE: 0439 Ecosystems, structure and dynamics (4815)
DE: 0480 Remote sensing
DE: 1632 Land cover change
DE: 9350 North America
SC: Earth and Space Science Informatics [IN]
MN: Fall Meeting 2005