FLOOD09620 - Laserfiche WebLink

Home Browse Search

2976 JARREIT AND TOMLINSON: REGIONAL INTERDISCIPLINARY PALEOFLOOD METHOD ~!t . I ~ ~iV 40 Long Lake tnlet / [-- g.g. ungaged paleoflood N ~ 30 .!! M E ,; a . ~ ~ c 20 Plc:eance Cr Trib. Yellow Cr \ ~. NW Colorado ,/ c => 10 . . o 1000 2000 2500 3000 Elevatlon, m 3500 1500 Figure 10. Relation between maximum unit discharge and elevation with envelope curve for northwestern Colorado and eastern Colorado [Jarrett, 1990b]. S-1 occurred in Yellow Creek near Rangley (streamftow- gaging station 09306255, drainage area of 679 km'). This flood was a hyperconcentrated flow that resulted from a localized rainstorm storm over less than about 50 km~' of the steep, sparsely vegetated basin (U.S. Geological Survey, unpublished data, 1978). For rivers draining higher mountain areas in the study area, peak flows are dominated by snowmelt runoff. For comparison, the envelope curves for streams b(:low 2300 m in eastern Colorado [Jarrett, 1990bJ and for the United States [Costa, 1987a] also shown on Figure 9 help demonstrate the lower-magnitude flooding in northwestern Colorado. Maxi- mum flooding in eastern Colorado is about 3 times larger than for similarly sized streams (>-3 km2) in northwestern Colo- rado. Maximum flooding in eastern Colorado streams is slightly smaller than maximum flooding in the United States (Figure 9). The envelope curve (Figure 9) of maximum 1l00ding can be used to estimate the hypothetical maximum finJd for Elkhead Creek at Elkhead Reservoir. For a drainage-b,:.tsin size at the reservoir (531 km2) the corresponding maximurJ. flood is about 240 m3 S-1. The maximum paleoflood estimate, of 135 m3 S-1 (Table 2, site 48) for Elkhead Creek downstroam from Elk- head Reservoir is 56% of the envelope curve value. The maximum unit discharge for streams in northwestern Colorado (Figure 10) is 5.2 m' S'I km" for Piceance Creek tributary (2.8 km2) near Rio Blanco (streamflow-gaging station 09306042) resulting from a locaiized rainstorm [Jarrett, 1987J. Four other small streams have had unit discharges greater than 3 m3 S-l km-2 resulting from intense, localized rainfall (Figure 10). The largest unit discharge in the highest mountains in the Park Range (Long Lake Inlet, Figure 10) is located in the area of maximum snowfall and represents maximum snowmelt run- off in northwestern Colorado. For comparison, maximum unit discharge is about 38 m' S'I km'z for small I;treams (<-10 km2) below 2300 m in eastern Colorado; the envelope curve for eastern Colorado is provided for comparison [Jaf7ett, 1990b]. Such a small maximum unit discharge in northwest Colorado is significant in that the storm occurre d in the YeHow Creek and Piceance Creek basin where steep hillslopes with sparse vegetation exacerbate runoff. Although maximum unit discharge gradually decreases with elevation in northwestern Colorado (Figure 10), the decrease is much more pronounced in eastern Colorado, where lower elevations are subject to 4000 extreme rainstorms (intensity, amount, and size) and thus se- vere flooding. Above about 2300 m in northwestern Colorado, unit discharges are slightly higher than east of the Continental Divide, which reflects the maximum snowmelt runoff from the Park Range. S}. Flood.Frequency Analysis Flood-frequency relations with EMA for selected streams in northwestern Colorado (Table 3) were developed using the recorded annual peak-flow data and paleoflood data (Table 2). Flood-frequency analyses were done using the paleoflood dis. charge, which was varied by the estimated uncertainty (e.g., site 61 in Table 2, 95 m' S'I :t 25% for the Elkhead Creek gage). The paleoflood record length of time (age) and the age reli. ability (range) for the maximum paieoflood (Table 2) was used to define the paleoflood record length in the analysis (e.g., 5000 == 1000 years for the Elkhead Creek gage). Because regional skew estimates [Interagency Advisory Com- mittee on Water Data, 1981] were developed over 30 years ago and do not incorporate paleoflood data, two EMA runs were made. The first EMA runs used station skew (Table 3). Then, to assess if regional skew may affect results, station skews (Table 2) were reviewed to assess if regional skew relations with station drainage area, period of record (station and pa- leoflood record length), and gage elevation. There were no statistically significant relations, perhaps because of using only eight stations, homogeneity of the study area, or the narrow range of skew about zero (-0.15 to +0.17) for these sites, which suggests paleoflood data may provide a stable at.site skew. Therefore a second set of EMA runs was made using the arithmetic average of the at-site skew values as a regional skew (-0.03) (Table 3), which is essentially a lognormal distribution, and these are considered the preferred curves. The 95% confidence limits were approximated using the B17B approach. Confidence limits only reflect parameter and peak discharge uncertainties; uncertainties such as best model, representative data, proper identification of censoring thresh- olds, and selection of proper skew are more difficult to quan- tify and were not included. In addition, the effects of climate change (natural or anthropogenic) may be the greatest source of uncertainty, and they are difficult to quantify [NRC, 1999]. Thus confidence limits do not reflect the total uncertainty in the frequency analysis (limits are too narrow), but no method is currently available to make such an assessment. Estimated flood quantiles listed in Table 3 for the eight streamflow-gaging stations reflect the average age and average discharge for the maximum paleoflood (Table 2). Flood- frequency relations for EMA for Elkhead Creek near Elkhead incorporating paleoflood data (the rectangle brackets the likely range of discharge and age for the maximum paleoflood) and station skew are shown on Figure 11. EMA results using the regional (average) skew of -0.03 also are listed in Table 3 and shown on Figure 11. It is not surprising that EMA frequency relations using the regional skew are not that different from the station relations (-4% difference for the 10,000.year flood in Table 3) because the at-site skew values have small varia- tion. A flood-frequency relation is needed at Elkhead Reservoir; however, there is no streamflow-gaging station close enough to the reservoir to transfer (scale) the gaged frequency estimates with the commonly used drainage-area ratio approach [Hosk- ing and Wallis. 1997]. Regional flood-frequency relations avail- able for western Colorado were developed by Kircher et at. J t ~