are limited to record lengths <strong>of</strong> less than about 30 years; basic rainfall and temperature data areavailable <strong>for</strong> some stations <strong>for</strong> up to 150 years, but in most cases are limited to less than 100 years.Historical DataHistorical data can provide a means <strong>for</strong> extending the length <strong>of</strong> record <strong>for</strong> many types <strong>of</strong>data, in particular <strong>for</strong> observations <strong>of</strong> the most extreme events. These data are most commonly usedto extend streamflow records <strong>of</strong> peak discharge prior to organized stream gaging. Historicalobservations can provide in<strong>for</strong>mation <strong>for</strong> other types <strong>of</strong> data such as weather patterns and thefrequency <strong>of</strong> extreme storm events, or changes in land use or vegetation that may be significant torun<strong>of</strong>f modeling calculations. However, as with any type <strong>of</strong> historical data, the accuracy and validity<strong>of</strong> the observations must be carefully assessed and compared to the other types <strong>of</strong> data used in theanalysis.Pale<strong>of</strong>lood DataPale<strong>of</strong>lood hydrology is the study <strong>of</strong> past or ancient flood events which occurred prior to thetime <strong>of</strong> human observation or direct measurement by modern hydrological procedures (Baker, 1987).Unlike historical data, pale<strong>of</strong>lood data do not involve direct human observation <strong>of</strong> the flood events.Instead, the pale<strong>of</strong>lood investigator studies geomorphic and stratigraphic records (various indicators)<strong>of</strong> past floods, as well as the evidence <strong>of</strong> past floods and streamflow derived from historical,archeological, dendrochronologic, or other sources. The advantage <strong>of</strong> pale<strong>of</strong>lood data is that it is<strong>of</strong>ten possible to develop records that are 10 to 100 times longer than conventional or historicalrecords from other data sources in the western United <strong>State</strong>s. In addition, the pale<strong>of</strong>lood record is along-term measure <strong>of</strong> the tendency <strong>of</strong> a river to produce large floods. In many cases, pale<strong>of</strong>loodstudies can provide a long-term perspective, which can put exceptional annual peak dischargeestimates in context and assist in reconciliation <strong>of</strong> conflicting historical records.Pale<strong>of</strong>lood data generally include records <strong>of</strong> the largest floods, or commonly the limits on thestages <strong>of</strong> the largest floods over long time periods. This in<strong>for</strong>mation can be converted to peakdischarges using a hydraulic flow model. Generally, pale<strong>of</strong>lood data consist <strong>of</strong> two independentcomponents. One component is a peak discharge estimate; the second is a time period or age overwhich the peak discharge estimate applies. Pale<strong>of</strong>lood studies can provide estimates <strong>of</strong> peakdischarge <strong>for</strong> specific floods in the past, or they can provide exceedance and non-exceedance bounds<strong>for</strong> extended time periods. Each <strong>of</strong> these differing types <strong>of</strong> pale<strong>of</strong>lood data must be appropriatelytreated in flood frequency analyses.Extrapolation Limits <strong>for</strong> Different Data TypesThe primary basis <strong>for</strong> a limit on credible extrapolation <strong>of</strong> extreme flood estimates derivesfrom the characteristics <strong>of</strong> the data and the record length used in the analysis. The data used in theanalysis provide the only basis <strong>for</strong> verification <strong>of</strong> the analysis or modeling results, and as such,extensions beyond the data cannot be verified. Different risk assessments require flood estimates <strong>for</strong>different ranges <strong>of</strong> annual exceedance probability (AEP), and there<strong>for</strong>e analysis procedures and datasourcesshould be selected to meet project requirements. The greatest gains to be made in providingPaper 18 – Swain, Bowles, Ostenaa 128
credible estimates <strong>of</strong> extreme floods can be achieved by combining regional data from multiplesources. Thus, analysis approaches that pool data and in<strong>for</strong>mation from regional precipitation,regional streamflow, and regional pale<strong>of</strong>lood sources should provide the highest assurance <strong>of</strong>credible characterization <strong>of</strong> low AEP floods.For many Reclamation dam safety risk assessments, flood estimates are needed <strong>for</strong> AEPs <strong>of</strong>1 in 10,000 and ranging down to 1 in 100,000, or even lower. Developing credible estimates at theselow AEPs generally require combining data from multiple sources and a regional approach. Table 1lists the different types <strong>of</strong> data which can be used as a basis <strong>for</strong> flood frequency estimates, and thetypical and optimal limits <strong>of</strong> credible extrapolation <strong>for</strong> AEP, based on workshop discussions orsubsequent communications. The limits presented in the table represent a general group consensus;however, opinions differed amongst workshop participants. In general, the optimal limits are basedon the best combination(s) <strong>of</strong> data envisioned in the western U.S. in the <strong>for</strong>eseeable future. Typicallimits are based on the combination(s) <strong>of</strong> data which would be commonly available and analyzed <strong>for</strong>most sites.Many factors can affect the equivalent independent record length <strong>for</strong> the optimal case. Forexample, gaged streamflow records in the western United <strong>State</strong>s only rarely exceed 100 years inlength, and extrapolation beyond twice the length <strong>of</strong> record, or to about 1 in 200 AEP, is generallynot recommended (IACWD, 1982). Likewise, <strong>for</strong> regional streamflow data the optimal limit <strong>of</strong>credible extrapolation is established at 1 in 1,000 AEP by considering the number <strong>of</strong> stations in theregion, lengths <strong>of</strong> record, and degree <strong>of</strong> independence <strong>of</strong> these data (Hosking and Wallis, 1997). Forpale<strong>of</strong>lood data, only in the Holocene epoch, or the past 10,000 years, is climate judged to besufficiently like that <strong>of</strong> the present climate, <strong>for</strong> these types <strong>of</strong> records to have meaning in estimates <strong>of</strong>extreme floods <strong>for</strong> dam safety risk assessment. This climatic constraint indicates that an optimallimit <strong>for</strong> extrapolation from pale<strong>of</strong>lood data, when combined with at-site gaged data, <strong>for</strong> a singlestream should be about 1 in 10,000 AEP. For regional precipitation data, a similar limit is imposedbecause <strong>of</strong> the difficulty in collecting sufficient station-years <strong>of</strong> clearly independent precipitationrecords in the orographically complex regions <strong>of</strong> the western United <strong>State</strong>s. Combined data sets <strong>of</strong>regional gaged and regional pale<strong>of</strong>lood data can be extended to smaller AEPs, perhaps to about 1 in40,000, in regions with abundant pale<strong>of</strong>lood data. Analysis approaches that combine all types <strong>of</strong>data are judged to be capable <strong>of</strong> providing credible estimates to an AEP limit <strong>of</strong> about 1 in 100,000under optimal conditions.In many situations, credible extrapolation limits may be less than optimal. Typical limitswould need to reflect the practical constraints on the equivalent independent record length that apply<strong>for</strong> a particular location. For example, many at-site streamflow record lengths are shorter than 100years. If in a typical situation the record length is only 50 years, then the limit <strong>of</strong> credibleextrapolation might be an AEP <strong>of</strong> about 1 in 100. Similarly, many pale<strong>of</strong>lood records do not extendto 10,000 years, and extensive regional pale<strong>of</strong>lood data sets do not currently exist. Using a recordlength <strong>of</strong> about 4,000 years, a typical limit <strong>of</strong> credible extrapolation might be an AEP <strong>of</strong> 1 in 15,000based on regional streamflow and regional pale<strong>of</strong>lood data.The in<strong>for</strong>mation presented in Table 1 is intended as a guide; each situation is different andshould be assessed individually. The limits <strong>of</strong> extrapolation should be determined by evaluating thelength <strong>of</strong> record, number <strong>of</strong> stations in a hydrologically homogeneous region, degree <strong>of</strong> correlationbetween stations, and other data characteristics which may affect the accuracy <strong>of</strong> the data.129 Paper 18 – Swain, Bowles, Ostenaa
- Page 1 and 2:
The National DamSafety ProgramResea
- Page 3 and 4:
collaboration and expertise, and th
- Page 5 and 6:
TABLE OF CONTENTSPage #FOREWORD....
- Page 7 and 8:
NEW DEVELOPMENTS AND NEEDS IN SITE-
- Page 9 and 10:
FOREWORDThe Federal Emergency Manag
- Page 11 and 12:
The PMF represents an estimated upp
- Page 13 and 14:
RESEARCH AREASResearch needs were b
- Page 15 and 16:
Lack of Historical Data for Extreme
- Page 17 and 18:
should be developed that will allow
- Page 19 and 20:
• Storms used in HMR 55A and in t
- Page 21 and 22:
last 20,000 year s in submitting th
- Page 23 and 24:
d. USBR Guidelines Publication, Dam
- Page 25 and 26:
was severe. There was a potential,
- Page 27 and 28:
Thirdly, we must develop better way
- Page 29 and 30:
For most projects, the first step f
- Page 31 and 32:
Instead of lumping the loss rates t
- Page 33 and 34:
adjust the rating curve for when th
- Page 35 and 36:
In response to this situation, TVA
- Page 37 and 38:
TVA established a Regional Resource
- Page 39 and 40:
The Utah Hydrological ExperienceByM
- Page 41 and 42:
The Utah Hydrological ExperienceByM
- Page 43 and 44:
Late SpringRain on snow eventModel
- Page 45 and 46:
State of GeorgiaFEMA Workshop on Hy
- Page 47 and 48:
a. HEC1• DOS based program - glit
- Page 49 and 50:
Long te rm hydrologic needs include
- Page 51 and 52:
Current State and County PracticesA
- Page 53 and 54:
present standards and, as such, are
- Page 55 and 56:
Research Needs in Dam Safety Analys
- Page 57 and 58:
Looking at the four test cases give
- Page 59 and 60:
Hydrologic Analyses Related to Dam
- Page 61 and 62:
(iii) Modified Expected Cost Approa
- Page 63 and 64:
Parameter Estimation Based on Joint
- Page 65 and 66:
estimation of lag times and loss ra
- Page 67 and 68:
Hydrology for Dam Safety - Private
- Page 69 and 70:
Probable Maximum PrecipitationAll o
- Page 71 and 72:
The most common synthetic methods a
- Page 73 and 74:
the standard would be very conserva
- Page 75 and 76:
65Paper 12 - Cecilio
- Page 77 and 78:
67Paper 12 - Cecilio
- Page 79 and 80:
69Paper 12 - Cecilio
- Page 81 and 82:
REFERENCES1. U. S. Army Corps of En
- Page 83 and 84: Overview of Flood Estimation Proced
- Page 85 and 86: Extreme FloodsExtreme floods, the t
- Page 87 and 88: Temporal patterns used to distribut
- Page 89 and 90: Preliminary Estimates of Rainfall a
- Page 91 and 92: 24 to 72 hours). CRC Research Repor
- Page 93 and 94: CURRENT AND FUTURE HYDROLOGIC RESEA
- Page 95 and 96: ecause existing procedures are inad
- Page 97 and 98: REGULATORY INVOLVEMENT AND COMMUNIC
- Page 99 and 100: Decision BasisThe third feature of
- Page 101 and 102: However, should it be revealed that
- Page 103 and 104: supporting sub-processes. Therefore
- Page 105 and 106: Probability ofFailure P f1Probabili
- Page 107 and 108: Since any rigorous analysis is an i
- Page 109 and 110: well, and usually do include subjec
- Page 111 and 112: Against this background, there is n
- Page 113 and 114: Howson, Colin., and Peter. Urbach.
- Page 115 and 116: Procedures and Analysis Technologie
- Page 117 and 118: Currently transposition limits for
- Page 119 and 120: pro cedures are used to allow the h
- Page 121 and 122: Use of Atmospheric Models in Rainfa
- Page 123 and 124: mountain watersheds in response to
- Page 125 and 126: 18. Schaefer MG, Stochastic Modelin
- Page 127 and 128: dams which could be used to save li
- Page 129 and 130: Hydrodynamic modelMIKE 21 BACKGROUN
- Page 131 and 132: A FRAMEWORK FOR CHARACTERIZATION OF
- Page 133: c) Decision Level Risk Assessment:
- Page 137 and 138: Fitting a distribution to data sets
- Page 139 and 140: unoff volumes. Examples of this typ
- Page 141 and 142: used to estimate frequency distribu
- Page 143 and 144: Hosking, J.R.M., and J.R. Wallis, 1
- Page 145 and 146: in west central Arizona. The draina
- Page 147 and 148: Trial and error reservoir routings
- Page 149 and 150: Figure 1 Alamo DamAlamo Dam and Spi
- Page 151 and 152: Paper 19 - Chieh and Evelyn 146
- Page 153 and 154: esources are made to mitigate flood
- Page 155 and 156: implementing a risk-assessment meth
- Page 157 and 158: Typically, the elevation of the top
- Page 159 and 160: similar to that of the high-gradien
- Page 161 and 162: and provides a cost-effective frame
- Page 163 and 164: 10000E. Colorado
- Page 165 and 166: 10000x Peak Discharge95% confidence
- Page 167 and 168: An integrated science approach will
- Page 169 and 170: ReferencesBaker, V.R., Kochel, R.C.
- Page 171 and 172: Partnerships, Proceeding of the 200
- Page 173 and 174: Flood HydrographThe flood hydrograp
- Page 175 and 176: A variety of data including informa
- Page 177 and 178: Improved Flood Frequency Extrapolat
- Page 179 and 180: RESEARCH NEEDS SUMMARYJerry Webb, H
- Page 181 and 182: o Unsteady flow computer models for
- Page 183 and 184: • Upd ate “Generalized Snowmelt
- Page 185 and 186:
DISCUSSION1. Genera l - After the p
- Page 187 and 188:
participant was asked to pick what
- Page 189 and 190:
RISK ANALYSIS100Relative Comparison
- Page 191 and 192:
WORKSHOPONHYDROLGIC RESEARCH NEEDSF
- Page 193:
JOE SKUPIENPrincipal Hydraulic Engi