A simple and accurate method for selecting forested sites having similar attributes would be useful for selecting areas for experimental replication. This thesis research tested the potential of using satellite spectral data to locate similar sites on the Priest River Experimental Forest in northern Idaho. Landsat Thematic Mapper data were used in an attempt to locate similar sites in terms of tree crown density, forest type, and timber size class.

Ten target sites were located representing four crown density classes (<20 percent, 20-40 percent, 40-60 percent, and > 60 percent), three forest types (western red cedar/western hemlock, grand fir/Douglas fir, western white pine/western larch), and three timber size classes (saw, pole, sapling). For each target site, three spectral indices were then used to locate ten potentially similar sites. The three spectral indices included a euclidian distance using TM bands 1 through 5, the TM4/TM3 ratio which is related to the Normalized Difference Vegetation Index (NDVI), and the TM5/TM4 ratio which is related to structual parameters such as shadowing and non-photosynthetic material.

Three hundred sites were located as potential sites similar to the target sites. These potentially similar sites were then evaluated by stereoscopic visual interpretation of color aerial photographs or by measurements at the site.

The methods used were useful for locating potential replication sites if the similarity attribute of interest was crown density or forest type. None of the spectral indices were useful for locating similar sites based on timber size class. The Euclidian distance index predicted the best for sites with 36 of 40 (90 percent) crown density sites correctly predicted and 20 of 30 (67 percent) of the forest type sites correctly predicted.
 

 email: leeh@pcprus.net