FLOOD01775 - Laserfiche WebLink

Home Browse Search

" many hours performing a detailed analysis of the Buffalo Creek flash flood and five other rainstorms in Buffalo Creek in 1996. He works closely with Urban Drainage and Flood Control District in Denver, which has included analyzing radar-estimated rainfall with a dense network of rainfall stations in the District. Henz has developed a procedure for estimating maximum rainfall for storms on a case-by-case basis using an algorithm that is conceptually different from the NWS. The NWS uses Z-R relationships that are not necessarily calibrated to local storm events on a case-by-case basis because of resource constraints. The 100-year, 1-hr rainfall is about 55 mm for the Buffalo Creek area as shown in figure 5 (Miller et aI., 1973). Diller (1 997) analyzed 24-hr, annual maximum rainfall values for 50 stations near Buffalo Creek having a similar climate regime. He conducted a regional rainfall-frequency analysis using the method of L-moments (Hosking and Wallis, 1997). The regional1-hr rainfall frequency curve developed by Diller (fig. 5) is within about 5 percent of the rainfall-frequency relation of Miller et al. (1973). The recurrence interval far exceeded 500 years for the July 12, 1996 1-hr and 24-hr rainfall amounts in Buffalo Creek. Maximum water depths as much as 4 m occurred within 30 minutes of the storm's onset in Buffalo Creek (fig. 6), Spring Creek, and the NF and South Platte Rivers. High-water marks (HWMs) generally were good-to-excellent and comprised of charcoal, leaf, and needle litter, silt, bent grass, and wash lines. HWMs were used to estimate the water slope and flood depth for each cross section. Peak discharge estimated with the slope-conveyance (SIC) method usually is less accurate than estimates using multiple cross sections. However, estimates provided here reflect an average of several SIC estimates along a reach of channel. These estimates probably are more accurate than a single SIC estimate because of a good agreement of estimates along a channel. The SIC uncertainty in discharge estimates is caused primarily by n values, bulking of flow with sediment and debris, and channel changes. Estimates of uncertainties are shown in parentheses. The peak discharge estimate was 450 m3/s (+1-20%) for Buffalo Creek near the N F South Platte River (fig. 1, site 1; fig. 6). This estimate reflects runoff from the burned area in Buffalo Creek and its tributaries (notably Sand Draw, Spring Gulch, Shinglemill Creek, and Morrison Creek). Sand Draw, with a drainage area of about 3.6 km2 (figs. 1 and 7, site 2), had an estimated peak discharge of 200 m3/s (+1-25%). The estimated peak discharge was 510 m3/s (+1- 25%) for Spring Creek upstream from its confluence with the South Platte River (fig. 1, site 3). Maximum flooding occurred in Spring Creek because the basin has slopes greater than about 30 percent, extensive bedrock exposure, hydrophobic soils, maximum rainfall occurred in its headwaters, and the storm moved from west to east down the basin. The South Platte River streamflow-gaging station (06707500) measures the cumulative runoff from the 1996 burn area (fig. 1, site 4), but principally from Buffalo and Spring Creeks. The SIC estimated peak discharge was 325 m3/s (+1- 20 'Yo), reflecting attenuation of flood peaks from Buffalo and Spring Creeks. The July 12, 1996 peak stage in the South Platte River at South Platte gage stilling well was 1.52 m lower than excellent high-water marks (HWMs), probably due to an insufficient size intake to fill the large-volume, stilling well. The peak discharge using the incorrect gage height (estimated during the storm) from the rating curve was only 81 m3/s; it is likely the entire flood hydrograph is suspect. Incorrect gage recordings are a serious concern for flash-flood detection and issuing warnings to the public. Large quantities of sediments were mobilized on hillslopes and in channels in the burned area during the July 12th flood (eg., figs. 3c and 7). A distinct black, burn boundary on rocks in pre- flood surfaces was used as a reference to estimate the general surface erosion from sheet wash. Care was taken to estimate the general erosion rather than the local erosion around the rock. In addition, pillars of soil were preserved under some surface rocks and metal objects on the burned areas. About a hundred hillslope measurements throughout the study area suggest an average of about 10 mm of erosion. The area of maximum sheet wash was limited to the headwaters of Shinglemill Creek, Spring Gulch, Sand Draw, and the upper third of Spring Creek (e.g., figs. 3c and 3d) and was used to help define the area of maximum rainfall (fig. 4). Locally, small streams produced up to 6 m of scour, primarily by headcutting (fig. 3c). Hundreds of trees, some as large as 0.75 m in diameter, toppled into the floodwaters, which exacerbated Draft 3/30/98 5