Image of PNWRC Logo
Improved Feature Extraction Using Multivariate Stacking

Synopsis Idaho State University (ISU) is one of six institutions (Pacific Northwest National Laboratory (PNNL), Idaho National Engineering and Environmental Laboratory (INEEL), ISU, Oregon State University, University of Idaho, and University of Washington) participating in the Pacific Northwest Regional Collaboratory (PNWRC). The mission of the PNWRC is to facilitate the integration of geospatial technologies and data to serve the needs of resource managers and decision-makers across the Pacific Northwest. Of the several PNWRC projects, ISU has participated in 2003-2004 in developing a multitemporal stacking tool for the detection of cheatgrass in southern Idaho. The project goals were to research and experiment with the concept of temporal stacking of Landsat imagery for better detection of cheatgrass; and to apply linear unmixing (or equivalent classification algorithms) to help identify cheatgrass cover. Once these goals were accomplished, we began preliminary technology transfer from a remote sensing environment (e.g. ENVI) to a geographic information system (GIS) environment (e.g. ArcGIS dlls) for ease-of-use by a broad audience. The majority of our work concentrated on a field area in southeastern Idaho, Bureau of Land Management’s Big Desert Allotment. A secondary analysis was performed in southwestern Idaho in the Birds of Prey Area. ISU partnered with PNNL and INEEL on this project; PNNL developed a Bayesian linear unmixing model (BLUM); while INEEL developed a model with the use of MODIS satellite imagery to identify ideal times to acquire the multitemporal imagery based on cheatgrass phenology.

--enter this site--