GRS used Landsat Thematic Mapper data to classify vegetation types on the 700,000 acre Applegate Watershed located in Southwestern Oregon.
USGS DEM data were processed to produce slope and aspect grids subsequently used to remove illumination variation in the satellite imagery. GRS also incorporated soils data to improve classification accuracy. Vegetation and land cover polygons were developed with attributes of tree size, canopy cover, and species composition using Discrete Classification, a methodology that uses a combination of supervised and unsupervised image classification techniques. Approximately 100 image training sites were visited on the ground (located with GPS) to determine the specific vegetation characteristics for each site using the line point transect method and the GRS canopy-densitometer. Ecological rules based polygon formation processes were used to develop the detailed vector GIS database. In addition to class values, all polygons are represented by discrete estimates of the descriptive attributes of tree size, canopy closure, and percent conifer composition. Since each polygon is represented by discrete estimates the user can then set or change the bounds of the general classes using simple GIS queries such as “select all polygons where density >0 and <15.” The vector database, developed in the MGE GIS environment, was delivered in ARC/INFO format. Pixel maps and topographically normalized imagery were delivered in grid ASCII format (ARCGRID,ERDAS, FARSITE).
