FLOOD06431 - Laserfiche WebLink

Home Browse Search

underestimation of 40 percent for the 10 drainage basins tested (Eash, 1993, p. 180-181). For this reason, the basin-slope characteristic was deleted from the drainage- basin characteristics data set during the initial multiple-regression analyses. Basin-slope comparisons appear to indieate that the 1:250,000-scale DEM data used to create the elevation-contour digital maps are not capable of reproducing all the sinuosity of the elevation contours depicted on the 1:250,000-scale DMA topographic maps. The elevation contours generated using the GIS procedure are much more generalized than the topographic-map contours; thus, the total length of the elevation contours are undermeasured when using the "contour-band" method of calculating basin slope (BS) (Appendix A). A cor>parison of the elevation contours shown in figure 4C for the Black Hawk Creek at Grundy Center (station number 05463090; map number 73, fig. 1) drainage basin to those depicted on the DMA 1:250,000-scale Waterloo topographic map showed a significant difference in the sinuosity of the elevation contours depicted. Drainage-Basin Characteristic Equations The 26 drainage-basin charaeteristics listed in Appendix A were quantified for 164 streamflow-gaging stations (fig. 1) and investigated as potential explanatory variables in the development of multiple-regression equations for the estimation of design-flood discharges. Because of the previously described problems concerning measurement verification of basin slope and because of the difficulty associated with manual measurements of total stream length, six basin characteristics were deleted from the regression data set. The excluded characteristics were basin slope (BS), total stream length (TSL), stream density (Sm, constant of channel maintE'nance (CCM), ruggedness number (RN), and s'.ope ratio (SR). Several other drainage-basin characteristics also were deleted from the data set because of multicollinearity. Multicollin"arity is the condition where at least one explanatory variable is closely related tc. (that is, not independent oD one or more other explanatory variables. Regression models that include variables with multicollinearity may be unreliable because coefficients in the models may be unstable. Output from the ALLREG analysis and a correlation matrix of Pearson product-moment correlation coefficients were used as guides in identifying the variables with multicollinearity. The hydrologic validity of variables with multicollinearity in the context of flood runoff was the principal criterion used in determining which drainage-basin character- istics were deleted from the data set. Upon completion of the ALLREG analyses, any remaining multicollinearity problems were identified with the SREGRES procedure by checking each explanatory variable for variance inflation faetors greater than 10. Statewide flood-estimation equations were developed from analyses of the drainage-basin characteristics using the ordinary least-squares and weighted least-squares multiple-regression techniques previously described. The best equations developed in terms of PRESS statistics, coefficients of determination, and standard errors of estimate are listed in table 2. The charaeteristics identified as most significant in the drainage-basin equations are contributing drainage area (CDA), relative relief (RR), drainage frequency (DF), and 2-year, 24-hour precipitation intensity (TTF). Table 9 (at end of this report) lists these significant drainage-basin characteristics, as quantified by the GIS procedure, for 164 streamflow-gaging stations in Iowa. Three of the four charaeteristics listed in the drainage-basin equations (table 2) are calculated from primary drainage-basin characteristics. The drainage-basin equations are comprised of six primary drainage-basin charaeteristics. Contributing drainage area (CDAl is a measure of the total area that contributes to surface-water runoff at the basin outlet. The primary drainage-basin characteristics used to calculate contributing drainage area are total drainage area (TDA) and noncontributing drainage area (NCDA). Relative relief (RR) is a ratio of two primary drainage-basin characteristics, basin relief (BR) and basin perimeter (BP). Drainage frequency (DF) is a measure of the average number of first-order streams per unit area and is an indication of the spacing of the drainage network. The primary drainage-basin characteristics used to calculate drainage 16 ESTIMATING DESIGN-FLOOD DISCHARGES FOR STREAMS IN IOWA ~