|
<br />8
<br />
<br />JIM E, O'CONNOR ET AL,
<br />
<br />Table 2. Long-Term Flood Frequency on rhe Colorado River
<br />Annual exceedance 50% 10% 5% 1% 0,5% 0,1%
<br />probability
<br />Recurrence interval (yrl 2 10 20 100 200 1000
<br />Analyzed flow IecOIds Flood quantiles ldisclwge :!: standatd enol) Im'sec-I)
<br />Gaged Iecoro (Lees Ferry) 2133 :!: 147 3778 :!: 351 4443 :!: 469 6022 :!: 798 6732 :!: 962 8468 :!: 1401
<br />Gaged + stratigraphic 2131 :!: 147 3997 :!: 181 4777 :!: 205 6674 :!: 305 8749 :!: 471 9708 :!: 562
<br />records
<br />ease 3 2001 :!: 116 3384 :!: 127 3927 :!: 138 5192 :!: 193 5751 :!: 228 7102 :!: 335
<br />Case 4 22W :!: 168 4509 :!: 233 5512 :!: 273 8033 :!: 427 9221 :!: 523 12254 :!: 819
<br />
<br />Note.' All analyses were perlormed with the computer progtsm developed by Stedinger et sl, (19881 usuminS that the flood
<br />observstions came &om s log-nonnal population distribution, Cases 3 snd 4 result &om ..suming that all minimum dischaqes
<br />snd observation thresholds ..lOciated with the stratigraphic =ro of floods are ove1e8timated lease 31 or unduestimated lease 4)
<br />by 25%, For ease 3, we ..sumed that deposit GI w..left by the flood of 19211this would be consistent with the revised diacIwxe
<br /><atimale8\, Thoae estimates 101 flow IeCUItel1ce should only be considered .. long-term norages, becaU5e the stratigtapbic and
<br />Psed rerords possibly include time periods of distincdy different flood frequency resulting &om c:bsnges in' climate or watershed
<br />conditions, Therefore, the flood irequency estimates given in this table are not necessatily indicative 01 the present probability of
<br />flooding,
<br />
<br />the recurrence of large floods compared to using
<br />the gaged record alone (figure 5, table 2). Adding
<br />the stratigraphic record to the gaged Iecord results
<br />in slightly increased (<15%) discharge estimates
<br />for floods of 10 to 1000 yr recurrence intervals,
<br />More important, however, the stal1Mrd errors for
<br />the estimates are much smaller: for flows with re-
<br />currence intervals 2:20 yr, the standard errors of
<br />the quantile estimates are less than half of those
<br />resulting &om an equivalent analysis of the gaged
<br />record alone ltable 2), This result indicates that'in-
<br />corporation of the stratigraphic record is equiva-
<br />lent to quadrupling the effective record length,
<br />Cases 3 and 4 of table 2 show the grellt sensitivity
<br />of the quantile estimates to the accuracy of the
<br />minimum discharge estimates, emphasizing the
<br />need for careful and accurate discharge calculation
<br />procedures when using historical or prehistorical
<br />records of flooding in this type of analysis,
<br />
<br />Conclusions
<br />
<br />A stratigraphic record of flooding, going back at
<br />least 4500 yr, provides evidence of at least 15
<br />floods on the Colorado River in the Grand Canyon
<br />with discharges similar to the large historic flows
<br />of 1884 and 1921. Ten floods during the last 2000-
<br />2300 yr had discharges greater than 6800 m3sec -I,
<br />One flow, 1600-1200 cal yr B,P., had a discharge
<br />larger than 14,000 masec-I, The evidence of this
<br />large flood supports Kieffer's (19851 suggestion, on
<br />the basis of channel morphology, that many of the
<br />rapids in the Grand Canyon have been affected by
<br />a flow greater than 11,000 m3sec -I,
<br />The stratigraphic record of the largest Colorado
<br />River floods, combined with a long record of an.
<br />
<br />nual peak discharges &om a nearby streamflow
<br />gage, provides quantile estimates of 10-1000 yr
<br />floods that are somewhat larger, but similar, to
<br />those obtained by evaluating the gaged record
<br />alone, For example, the loo-yr flood determined
<br />from a maximum likelihood analysis incorporating
<br />the stratigraphic record is 4000 :!: 180 masec-I,
<br />compared to 3780 :!: 350 masec -I obtained from a
<br />similar analysis of the gaged record alone, Further-
<br />more, the stratigraphic record provides informa-
<br />tion regarding the past sequence of extremely large
<br />floods, This geologic information, unobtainable by
<br />extrapolation of modem streamflow records, is im-
<br />portant for understanding the sequence of events
<br />that have acted cumulatively to shape the present
<br />Colorado River geomorphology, where timescales
<br />of adjustment for some processes are longer than
<br />modem records of observation,
<br />
<br />ACKNOWLEDGMENT
<br />
<br />J, R, Stedinger advised us on the flood frequency
<br />analysis and application of the MAX program,
<br />R, D, Jarrett, R, H, Webb, J. E, Costa, S, W, Kieffer,
<br />K. M, Nolan, and Journal of Geology referees pro-
<br />vided reviews, This research was partially sup-
<br />ported by the Engineering directorate, Natural and
<br />Man made Hazards Mitigation Program, National
<br />Science Foundation, Grant BCS-8901430; the Bu-
<br />reau of Reclamation (Glen Canyon Environmental
<br />Studies); and a National Research Council Post-
<br />doctoral fellowship to O'Connor, This article is
<br />contribution no, 14 of the Arizona Laboratory of
<br />Paleohydrologica1 and Hydroclimatological Analy-
<br />sis (ALPHAl, University of Arizona,
<br />
|