Combining FIA plot data with topographic variables: Are precise locations needed?This article is part of a larger document. View the larger document here.
Plot data from the USFS FIA program could be combined with terrain variables to attempt to explain how terrain characteristics influence forest growth, species composition, productivity, fire behavior, wildlife habitat, and other phenomena. While some types of analyses using FIA data have been shown to be insensitive to precision of plot locations, it has been suggested that terrain-based models may require the use of precise plot coordinates. This study compares results obtained from a variety of terrain-based analyses conducted in the Blue Ridge of North Carolina using both precise and perturbed (fuzzed and swapped) FIA plot locations, and documents differences between field-estimated slope and aspect and GIS-derived slope and aspect. Digital elevation model (DEM) data were used to derive simple topographic parameters such as elevation, slope percent, azimuth of aspect, terrain curvature, flow accumulation, slope position, and compound topographic index. These values were then compared in a pairwise fashion for plots using precise and perturbed coordinates. Correlations between precise and perturbed plot locations ranged from r = -0.006 to r = 0.383, except for precise versus perturbed plot elevations where r = 0.929. Second, a simple, terrainbased forest site quality index (FSQI) was calculated for the each plot. This index defines site quality classes for forest productivity based on azimuth of aspect, slope percent, and slope position. FSQI classifications were compared for precise and perturbed plot coordinates; at best only 40% of plots resulted in the same productivity class (out of 5). Finally, field-obtained estimates of slope and aspect were compared with GIS-derived estimates from precisely-located plots to assess their level of agreement. Correlations between field-measured and GIS-derived values were r = 0.6 for slope and r = 0.4 for aspect. Results of these experiments indicate that perturbed plot locations may not be suitable for such fine-scale applications.