|
<br />The flood-frequency relation for Box Elder Creek near 1-70 (FEMA, 1993) also is shown for
<br />comparative purposes (fig, 9), This relation plots above the largest paleofloods and historical
<br />floods, Will Thomas (Michael Baker, Jr" written commun., 1997) developed preliminary
<br />regional-regression equations for the 100-year flood using streamflow-gaging station data to
<br />assist FEMA in improving flood-frequency relations for eastern Colorado, Thomas found that
<br />incorporating a basin-shape factor, which helps account for narrow (elongated) basins,
<br />reduces the standard error of estimate to about 25 percent in his regression model. The 100-
<br />year flood estimates of Thomas are shown for Cherry and Box Elder Creeks on figures 6 to 9,
<br />Thomas' 100-year flood estimates are very similar to results from this study for Cherry
<br />Creek (figs, 6 and 7) and Box Elder Creek at Elizabeth (fig, 8), However, Thomas' 100-year
<br />flood estimate of about 16,800 ft3/s for Box Elder Creek near 1-70 (fig. 9) plots more than
<br />twice the 100-year flood of the U,S, Army Corps of Engineers (1990) and CH2MHILL, 1995).
<br />Thomas' 100-year flood is about three times the flood of record of about 6,000 ft3/s in about
<br />100 years, about 1.9 times larger than a paleoflood of 8,700 ft3/s in 100 to 500 years, and
<br />about equal to the largest paleoflood in several hundred to several thousand years. The
<br />paleoflood data support the flood-frequency relations developed from rainfall-runoff modeling
<br />for Box Elder Creek (figs, 8 and 9), The good agreement (+1-10 percent) between rainfall-
<br />runoff modeling flood-frequency relations (U,S, Army Corps of Engineers, 1990; CH2MHILL,
<br />1995) and historical and paleoflood data probably reflects the ability to better assess unique
<br />hydrologic characteristics (e,g" narrow basin, high infiltration in soils and into streambeds,
<br />and availability of rainfall-runoff data for Denver area streams to calibrate the model) than
<br />presently available estimates from existing regression equations for Box Elder Creek,
<br />
<br />For comparative purposes, PMF estimates for Cherry Creek are shown on figures 6 and 7 and
<br />for Box Elder Creek are shown on figures 8 and 9. Although extrapolations of flood-frequency
<br />relations have uncertainties, they can be used to estimate the probability of extremely large
<br />floods when paleoflood data are available (Jarrett and Costa, 1988; Jarrett, in review a, Levish
<br />and others, 1994; Ostenaa and Levish, 1995), Floods the magnitude of PMF values for Cherry
<br />and Box Elder Creeks have recurrence intervals far in excess of a 10,000 years based on the
<br />flood-frequency analyses incorporating the paleoflood data,
<br />
<br />DISCUSSION AND SUMMARY
<br />
<br />Although paleoflood estimates involve uncertainties, they are based on interpretations of
<br />physical data preserved in channels and on floodplains for at least several thousand years.
<br />These uncertainties primarily are related to post-flood changes in channel geometry and flood
<br />heights interpreted from PSis. Where possible, paleoflood sites in this study were located in
<br />bedrock controlled channels that minimize potential changes in channel geometry that may
<br />occur in alluvial channels; there is little evidence that major changes in channel geometry have
<br />occurred in alluvial channels in the study area in many thousands of years, HWM-PSI relations
<br />developed from recent, extreme floods in the western United States clearly demonstrate that the
<br />top of flood-deposited sediments (PSis) are equal to the elevation of flood HWMs, thus, helps to
<br />reduce the uncertainty of paleodischarge estimates (Jarrett, in review b). Paleoflood estimates
<br />for Colorado rivers are believed to have uncertainties of about 25 percent (Jarrett and
<br />Way thomas, in press; Jarrett, in review a and b; Jarrett and others, in review), Paleoflood
<br />data for 134 sites on the Palmer Divide provide additional data for large floods in the past
<br />several thousand years not available by other sources, provide a higher level of confidence in
<br />
<br />10
<br />
|