9493 - Laserfiche WebLink

Home Browse Search

FINAL REPORT, November 2003 High-jlow Requirements for the Duchesne River in the perceived locations of these boundaries reflects the changes in vegetation density and fluvial activity being evaluated by these metrics. Digitizing error is the result of the inability of a GIS technician to exactly trace mapped lines during the digitizing process. The effect of this error on polygon area depends on polygon shape, polygon size, and the magnitude ofthe digitizing error. Sondossi (2001) empirically determined the effect of random linear errors of 2 m or less for polygons with an average length- to-width ratio of 5: I and a range of areas ranging across five orders of magnitude. He reported that the percent error in polygon area for a given linear error of 2 m declined with polygon size according to the relationship: EA = 27.827A-0.464 (2) where EA is the percent error in polygon area and A is polygon area. This relationship should hold for all polygons of similar shape. As Sondossi' s polygons also represented fluvial surfaces along a river corridor and have a similar relationship between polygon perimeter and area [Sondossi reported P = 3.36Ao.59 while a test of this data set using polygons representing channel, high bar, and floodplain areas from the 1980 coverage yielded P = 3.364A0.565], it was assumed that the Sondossi relationship between percent error in area due to a given linear error and polygon area hold for this data set as well. Sondossi also determined that for a given polygon size and shape, percent error in area scales linearly with the random linear error. In other words, doubling the magnitude of the random linear error results in a doubling of the percent error of a given polygon. Following Sondossi, random linear errors associated with this study were evaluated in terms of photograph scale. The ability of a GIS technician to accurately trace and digitize unit boundaries is limited by the width of the lines used to map unit boundaries on air photos A 0.5- mm pencil was used for mapping in the present study, and the smallest-scale photo enlargements were 1: 14,500. A 0.5-mm accuracy limit for drawing and digitizing unit boundaries on our enlargements therefore incorporates maximum positional errors about 7.2 m on the ground. Low-flow channel polygons used in this study average approximately 41,000 m2 in area while floodway channel polygons are considerably larger. Application of the Sondossi relationships to the average areas of polygons used to calculate channel width in this study using a random linear error of 7.2 m results in an insignificant expected error ofless than 1 percent. 17