Laserfiche WebLink
<br />. difference for larger retum periods is due to different methods (rainfall-runoff modeling and regional analysis <br />of streamflow-gaging station data). The paleoflood data is helpful for extrapolating frequency relations, but <br />there is an uncertainty of abut 25 percent with those discharge estimates. If there had been a flood of <br />about 10,000 ft3/s or larger in Elkhead Creek Basin, flow depths would be 3 fl or more and velocities of 7 <br />fVs or more on the floodplain. Such flows would be more than sufficient to erode the silt-sized material and <br />deposit coarse-grained flood bars. <br /> <br />The belief that large rainstorms produce large floods in the Yampa River Basin has had implications for <br />flood-plain management as well as for dam safety (discussed in the next section). A number of streamflow- <br />gaging stations in the Yampa River Basin have nearly 100 years of record and essentially all peak flows <br />resulted from snowmelt These long-term gaged data were assumed not to be representative of extreme <br />flood potential. Thus, the flood hydrology used in flood-plain studies was based on enhancing the upper <br />end of the flood-frequency relation using rainfall-runoff modeling or rainfall-runoff modeling alone for ungaged <br />basins such as for the town of Steamboat Springs (Federal Emergency Management Agency, 1976). The <br />FEMA flood-frequency relation for the Yampa River at Steamboat Springs, where data collection began in <br />1904, is shown on figure 17a, There is good agreement in magnitude-frequency relations to about the 50- <br />year event For larger recurrence intervals, the results based on rainfall-runoff modeling increases <br />substantially and do not fall within the 95 percent confidence limits for the gage flood-frequency relation, In <br />addition, the 500-year FEMA estimate is double the maximum paleoflood of 10,000 fl3/S at this site, <br />. Similarly, the FEMA"rainfall-runoff magnitude-frequency relation for Walton Creek (figure 17b) also does not <br />to agree with the gaged-flood data. Uncertainties in rainfall-runoff modeling to estimate flood characteristics <br />result from a lack of rainfall, infiltration data, and streamflow data in northwestem Colorado, particularly for <br />short-duration events. <br /> <br />Implications to Hydrologic Aspects of Dam Safety <br /> <br />Extensions of flood-frequency relations to rare floods (eg., 1 O,OOo-yr recurrence interval or 0.0001 <br />exceedence probability) are tenuous for short-gaged records, but can be done cautiously when paleoflood <br />data are included in flood-frequency analysis (Costa, 1978; Kochel and Baker, 1982; Jarrett, 1987; Jarrett <br />and Costa, 1988; Jarrett and Waythomas, in press), The National Research Council (1985) indicated that <br />reasonable and realistic risk investigations can be conducted by linear extensions of the frequency curve <br />to the PMF estimate. These extensions provide a simple approach to place PMF estimates in perspective <br />with gaged, regional, and paleoflood estimates of flood potential (Jarrett and Costa, 1988). The recurrence <br />interval for the maximum paleofloods in Elkhead Creek ranges from about 1 ,000 to more than 10,000 years <br />(figure 17d), The 1 O,ooo-yr flood is about 6,500 fl3/S at Elkhead Reservoir. For comparative purposes, <br />the PMF estimate for Elkhead Creek at the dam (Ayres Associates, written commun., 1996) is 36,052 ft3/s, <br />which also is shown on figure 17d. <br /> <br />. Efforts have been made to assign a frequency to the PMF (National Research Council, 1986, 1988), but <br />they are subjective. There is disagreement as to the range of frequency for the PMF as well as if a <br /> <br />28 <br />