Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: The Kress Modeler for Multi-Criteria Evaluation for Livestock Distribution

Authors
item Ganskopp, David
item Johnson, Michael - OREGON STATE UNIV
item Johnson, Douglas - OREGON STATE UNIV
item Harris, Norm - UNIV. ALASKA FAIRBANKS
item Louhaichi, M - OREGON STATE UNIV
item George, Mel - UNIV. CALIF/DAVIS

Submitted to: Society for Range Management Meeting Abstracts
Publication Type: Other
Publication Acceptance Date: July 29, 2004
Publication Date: N/A

Technical Abstract: The Kress modeler was developed to rapidly predict grazing distribution maps for livestock in pastures. Predicted animal use at a specific location is derived from relationships between the animal and its environment. For example one would expect to see grazing animals in thermally neutral areas with greater standing forage, high forage quality, gentler slopes and close to water. Each factor in the relationship is weighted as to importance and applied via multi-criteria mathematical models. Once a model is built, it can be saved and applied to a new pasture or landscape. We have programmed a model evaluation routine that uses data gathered from animal GPS collars. The GPS information is read and, if desired, a random selection is taken from it. The frequency of animal presence in each cell on the landscape is determined and compared to a random distribution model using a ROC analysis. Our program was built to be flexible and broadly useful to natural resource applications. Since we work primarily with lands used for domestic animals, water, and wildlife, natural resource applications are the primary focus. We made it as close to an open "blackboard" format as we could. Thus a scientist or manager should be able to conceptualize linear, non-linear, or mixed models and, if spatial data exists for the parameters chosen, rapidly apply them to a landscape. The KRESS modeler is designed to be used in conjunction with a full-fledged GIS program which is required to develop the data layers that the modeler uses.

Last Modified: 7/23/2014
Footer Content Back to Top of Page