Laserfiche WebLink
The accuracy of photogrammetric measurements is affected to varying degrees by the clarity of the <br />photographs, differences in flow level, and distortion. Differences in clarity lead to problems in <br />the interpretation of features and the accuracy with which they can be digitized; differences in flow <br />level affect the planform area of the river and associated features; and distortion near the edges of <br />photographs can make objects appear larger or smaller than they really are. For practical reasons, <br />we did not rectify the photographs to correct for distortion. We did, however, evaluate the <br />potential error from these various sources. Errors due to interpretation and tracing of objects on a <br />set of photographs were evaluated by re-digitizing reaches of the river and comparing the results to <br />the original measurements. Errors due to differences in flow level were evaluated from field <br />measurements of channel cross sections at different flows. Finally, errors due to distortion were <br />estimated by measuring the area of 20 islands near the center of the photographs, and comparing <br />this to the area of the same islands when they were near the edges of the adjacent photographs. <br />The results of these tests indicate that the error associated with interpreting and tracing the main <br />channel boundary is negligible (2%). The error associated with tracing side channels and <br />backwaters is more sizable (10%), because these features are harder to interpret. Differences in <br />discharge have a negligible (-3%) effect on measurements of planform area, as long as the <br />difference in discharge is less than about 30%. Thus, for the Colorado River, we feel confident <br />comparing the photographs from 1954 with those from 1968, and those from 1937 with those <br />from 1993, but not in comparing them all together. For the Gunnison River, the difference in <br />discharges between 1937 and 1995 is relatively large (-80%), but the higher discharge occurs in <br />the more recent set of photographs, which would tend to make the channel appear larger, even if it <br />had not changed. With respect to other sources of error, the average error due to distortion at the <br />edge of photographs is approximately 3%, but since we tried to avoid measuring features near the <br />edge, the error introduced by distortion is certainly much less. Even so, if we assume a worst case <br />scenario where the individual errors are additive, then it is possible that the photogrammetric <br />measurements of main channel area are off by as much as 8%, and that the measurements of side <br />channel and backwater area are off by as much as 16%. If we further assume that every feature <br />was overestimated in one set of photographs, and underestimated in another set, then the maximum <br />potential error could be twice as large. Although it is highly unlikely that the errors are all additive <br />and always in the same direction, we use these values as a basis for saying whether or not the <br />observed changes in channel morphology are significant. <br />Field Studies <br />Field studies were conducted from 1993 through 1996 to (1) monitor geomorphic changes; (2) <br />determine the average characteristics of the main channel (width, depth, slope and grain size); and <br />(3) develop flow and sediment transport models. Geomorphic changes in side channels and <br />backwaters were monitored by repeated surveys. Prior to the start of the 1993 snowmelt runoff <br />period, three side channel-backwater sites along the Colorado River were selected for detailed <br />study. USFWS biologists recommended these sites to us because they were typical of habitats <br />used by adult Colorado squawfish. Figure 3 (presented earlier) shows a site that we monitored in <br />the 15-mile reach; the mouth of this side channel is a backwater at most flows. Another site, <br />located near river mile (RM) 162 in the 18-mile reach, is formed by an alternate bar and chute <br />channel. The chute channel conveys water at moderate to high flow but becomes a backwater at <br />low flow. The third site, located near RM 160, is formed by a permanent island and side channel. <br />This side channel conveys water at moderate to high flow but likewise becomes a backwater at low <br />flow. At each site, a series of cross sections were surveyed around the head and mouth of the side <br />channel. These areas were of interest to us because they control the amount of flow into and out of <br />the side channel, which determines whether or not fish can access the site. <br />10