Using Color Infrared Imagery and Remote Sensing Software to Classify Vegetation at Agassiz National Wildlife Refuge

The ability of remote sensing applications to accurately differentiate priority vegetation types was evaluated on a 664-hectare habitat management unit on Agassiz National Wildlife Refuge, located in Marshall County in northwest Minnesota. The Refuge is a diverse complex of wetland and upland habitats, largely inaccessible by foot. Its relative inaccessibility, coupled with the known occurrence of various non-native and invasive plant species, presents a critical need for inventory and monitoring of Refuge flora. Aggressive species such as narrow-leaved cattail (Typha angustifolia), common reed (Phragmites australis), and willow (Salix spp.), all prevalent on the Refuge, are of special management interest. The ability to determine change in percent cover of priority vegetation types over time is important in evaluating the success or failure of habitat management practices and the Refuge‟s progress in meeting habitat objectives. This study was designed to measure the capabilities of Definiens eCognition and ERDAS TM software in delineating and classifying these vegetation types across both upland and wetland Refuge habitats.