This is a land-cover classification in-progress for the Teton County area, our pilot study area. A combination of techniques and images are being used to develop land-cover data. The TM scene shown here was segmented into 100 spectral classes using an unsupervised classification, and then labeled based on interpretations of various ancillary data including digital orthophoto quadrangles. Decision-making rules and models are being developed using data such as elevation (shown below), slope and aspect, image texture, and block group census data in order to separate confused spectral classes.

Click on the image to see a larger version.

This ancillary data layer is being used to resolve confusion in spectral classes that can be separated based on elevation. For instance, most crops occur below 2125 meters therefore eliminating confusion with high elevation meadows.