9493 - Laserfiche WebLink

Home Browse Search

FINAL REPORT, November 2003 High-jlow Requirements for the Duchesne River Error in polygon area caused by photograph distortion has been empirically determined to average about 3 percent when working near the edges of photographs where distortion is most exaggerated (Van Steeter and Pitlick 1998). As we mapped near the centers of photographs as much as possible, error in polygon area from photograph distortion is probably well below this value. Large-scale registration errors like those described by reported RMS errors have a muted effect on errors in polygon area because these deviations from true coordinates tend to displace all portions of a polygon in a similar direction and magnitude. The result, especially for small polygons, is a simple translation of polygon location with little effect on polygon shape or area. Taken together, we assumed that all sources of error produce no more than a 10 percent error in the average channel width reported in this study. An exception was made for subreach 2 from the 1948 coverage, where the reported channel width is unreliable. Areas of Erosion and Deposition Areas of erosion and deposition between sequential coverages were obtained for subreaches 5-19 by performing a spatial union in ArclInfo and classifying the resulting change polygons as areas of erosion, deposition, or no change (Table 3). Important potential sources of error in the reported erosion/deposition areas included mapping errors and positional errors due to systematic shifts in the coordinate system. Mapping error in this context consisted of incorrect or inconsistent placement of unit boundaries due to the gradational nature of many boundaries. An example would be a floodplain surface that gradually grades into a terrace surface over an extended slope. Variation in placement of the boundary between the two units on successive photo series could result in an apparent change from floodplain to terrace where no real change occurred. Errors of this type were located and eliminated by visual inspection of the relevant change polygons. It should be pointed out that transitions between high bars and floodplains were considered to represent no change with respect to erosion or deposition because these surfaces are found at similar elevations in this study area. Their differences lay primarily in the extent of vegetation colonization and in lateral distance from the active channel. Positional errors associated with the overlay of polygons from coverages for different years are more problematic than are errors within a single coverage. This is because feature shifts due to photograph distortion or inadequate control points in one coverage are independent of feature shifts associated with the other coverage. Consider two successive coverages, each 18