Laserfiche WebLink
<br />12 <br /> <br />TECHNIQUES OF WATER-RESOURCES INVESTIGATIONS <br /> <br />structures on or near a stream. The usual ap- <br />proach is to estimate the discharge from a <br />regional relation and to compute the stage <br />from this discharge and a channel survey. A <br />simpler though less accurate method relates <br />stream depth to discharge or to basin charac- <br />teristics (see Thomas, 1964, and Gann, 1968). <br />A more comprehensive study was made by <br />Stall and Yang (1970) in which stream depth I <br />(and other measures of channel geometry) <br />were related to flow frequency and drainage <br />area. All three of the above-referenced studies <br />are based on the pioneering work on channel <br />geometry by Leopold and Maddock (1953). <br /> <br />Procedures for Other Flow <br />Characteristics <br /> <br />Multiple regression has been used to re- <br />gionalize mean annual flows, mean monthly <br />flows, annual minimum flows, annual flood <br />volumes, and some other characteristics. <br />Thomas and Benson (1970) described a study <br />of relations for estimating streamflow charac- <br />teristics from drainage-basin characteristics <br />in four hydrologically differing regions of the <br />United States. An even more comprehensive <br />use of the multiple-regression method for <br />regionalization of flow characteristics was <br />performed in each State of the conterminous <br />United States during 1970. Results of this <br />study are given in a series of reports, gen- <br />erally one for each State; a typical one is by <br />Collings (1971). <br />In most humid regions mean flow is closely <br />related to drainage area and mean annual <br />precipitation. Thomas and Benson (1970) <br />found a standard error of regression of 14.4 <br />percent using those two variables in Potomac <br />River; they reduced it further by including <br />channel length and mean annual snowfall. <br />Standard errors of 10 to 15 percent have been <br />attained in other humid regions. <br />In semiarid regions of large relief the rela- <br />tion of mean flow to drainage area and pre- <br />cipitation may not be usable because of (1) <br />the great range in precipitation with eleva- <br />tion, (2) the lack of good information on <br />precipitation, and (3) the strong influence of <br />geology on mean flow. For example, the <br /> <br />standard error of the regression equation for <br />mean flow in New Mexico is 53 percent (Bor- <br />land, 1970). <br />In certain humid regions a satisfactory <br />regionalization of mean flows is not attain- <br />able because of the movement of ground <br />water across topographic divides. Some re- <br />gions exhibiting this condition are the Ump-. <br />qua River basin in Oregon, the Red Rock <br />River basin in Montana, and the Balcones <br />Fault region in Texas. <br />Although the principles of regional analy- <br />sis apply to all flow characteristics, the appli- <br />cation to low flows is least successful because <br />of the greater dependence of low flows on <br />basin characteristics that are imperfectly <br />known and that cannot be described by sim- <br />ple indexes. Geology is the chief basin charac- <br />teristic, other than drainage area, controlling <br />the size of low flows in a region of homoge- <br />neous climate. Evapotranspiration, especially <br />from the channels and flood plains, also has a <br />substantial effect on low flows in many basins. <br />Most reported attempts at regionalization <br />of low flows on a statewide basis have been <br />unsuccessful. Forty-seven Geological Survey <br />districts participated in and reported on their <br />comprehensive regionalization studies in <br />1970. Most districts reported either standard <br />errors of low flows in excess of 100 percent <br />(average of plus and minus percentages) or <br />that no meaningful relation was derived. A <br />notable. exception was Connecticut; there the <br />7 -day 10-year low flow was related to drain- <br />age area, channel slope, mean basin elevation, <br />and percentage of basin covered by stratified <br />drift, with a standard error of 68 percent <br />(Thomas and Cervione, 1970). This small <br />standard error (relative to those found in <br />most regions) resulted from the inclusion of <br />the fourth parameter. In a previous paper, <br />Thomas (1966) reported large unit base flows <br />from stratified drift and very small ones from <br />till, the predominant surficial glacial deposit. <br />Regionalization of low flows in a few geo- <br />logically homogeneous regions of limited ex- <br />tent has produced useful results. A "Tech- <br />niques of Water-Resources Investigations" on <br />low flow investigations now (1972) being <br />prepared, will consider regionalization of low <br />flows in more detail. <br /> <br />., <br /> <br />. <br /> <br />. <br />