Nationwide Estimation of Extreme Floods for Bridge-Scour Analysis

Size: px

Start display at page:

Download "Nationwide Estimation of Extreme Floods for Bridge-Scour Analysis"

Adrian Bradley
5 years ago
Views:

218 TRANSPORTATION RESEARCH RECORD 1290 Nationide Estimation of Extreme Floods for Bridge-Scor Analysis W. 0. THOMAS, JR.,. H. KIRBY, J. B. ATKINS, AND M. R. }RNNINGS Procedres are described for estimating floodpeak discharges and flood hydrographs for extreme floods, sch as the 500-year flood, to evalate scor at bridges.

1 218 TRANSPORTATION RESEARCH RECORD 1290 Nationide Estimation of Extreme Floods for Bridge-Scor Analysis W. 0. THOMAS, JR.,. H. KIRBY, J. B. ATKINS, AND M. R. }RNNINGS Procedres are described for estimating floodpeak discharges and flood hydrographs for extreme floods, sch as the 500-year flood, to evalate scor at bridges. These procedres are being incorporated (Jne 1990) in a microcompter program called "National Flood Freqency" developed by the U.S. Geological Srvey in cooperation ith the Federal Highay Administration and the Federal Emergency Management Agency. U.S. Geological Srvey flood reports typically describe procedres for estimating flood-peak discharges for retrn periods of 2 to 100 years for rral and rban ngaged atersheds. This paper describes procedres in the National Flood Freqency program for extrapolating design floods ith retrn periods of 2 to 100 years to the 500-year flood. INTRODUCTION The planning and design of bridges reqires the estimation of extreme flood discharges to facilitate an adeqate and safe design. Recent bridge failres, reslting from excessive scor, have prompted the Federal Highay Administration (FHWA) to develop procedres for evalating scor at bridges. As part of this program, the FHWA advised the State Departments of Transportation (DOT's) nationide to evalate the risk of their bridges being sbjected to scor damage dring floods on the order of a 100- to 500-year or greater average retrn period. In response, State DOT's are developing and implementing improved procedres for evalating the magnitde of scor dring extreme floods and for designing bridges to minimize scor damage. Essential to these improved procedres is a consistent, nbiased, and easy to apply techniqe for estimating discharges of floods having retrn periods on the order of 100 to 500 years. 1 Hydrologist, U.S. Geological Srvey (USGS), Reston, VA., Hydrologist, USGS, Rest on, VA., Hydrologist, USGS, Tscaloosa, AL. and Hydrologist, USGS, Astin, TX., respectively. The prpose of this paper is to describe a procedre, applicable nationide, for estimating extreme floods that ill meet the design needs of FHWA and State DOT'S. The U.S. Geological Srvey (USGS), in cooperation ith FHWA and the Federal Emergency Management Agency (FEMA), is developing (Jne 1990) a National Flood Freqency (NFF) microcompter program for estimating floodpeak magnitdes of T year events (2 to 500 years) and for estimating a typical hydrograph associated ith sch events for ngaged rral and rban sites in most areas of the United States. Jennings and Cookmeyer (1) provide an overvie of the plans for NFF. The flood-peak estimation techniqes are a compilation of regional regression eqations developed by USGS (primarily in cooperation ith State DOT's)) over the last 15 years or so. The flood hydrograph estimation techniqe is based on nit-hydrograph theory and tilizes a dimensionless hydrograph that is defined by the flood-peak discharge and the atershed lagtime. This paper primarily is concerned ith describing an extrapolation procedre for estimating the 500-year flood for bridge-scor evalation. ESTIMATING FLOOD DISCHARGES The USGS has developed regional regression eqations for estimating floods ith selected recrrence intervals for rral atersheds for every State and Perto Rico. In some areas of the Nation, hoever, data are inadeqate to define flood-freqency characteristics. Jennings and Cookmeyer ( 1) report that the NFF microcompter program ill inclde abot 1500 eqations for 214 flood regions in the United States. USGS reports typically give regression eqations for estimating the 2-, 5-, 10-, 25-, 50- and 100-year flood-peak discharges for an average of abot for flood regions in each State. Only in a fe States has USGS pblished

2 Thomas et al. 219 regression eqations for the 500-year flood; hence the need for an extrapolation procedre. The regional regression eqations are developed by relating the T-year flood-peak discharges at gaging stations to atershed and climatic characteristics by means of mltiple regression eqations. (The T-year discharges at gaging stations are determined by folloing gidelines adopted by the Interagency Advisory Committee on Water Data, (2)). Estimates at ngaged sites then can be compted by sing the appropriate atershed and climatic characteristics hich sally are available from topographic maps and climatic reports. Regression eqations for estimating flood-freqency characteristics for rban atersheds have been pblished in 18 States. In some instances, these eqations are applicable stateide, in other instances, the eqations are applicable only for one metropolitan area. These eqations are based on atershed and climatic characteristics as ell as on rban characteristics, sch as the percentage of the atershed that is impervios, or a measre of the state of development of the drainage system, sch as the basin development factor described by Saer and others (3). For those States ithot rban regression eqations, a nationide techniqe, developed by Saer and others ( 3), is sed to adjst the rral freqency crve to rban conditions. Thomas (4) provides a listing by State of all USGS pblished reports (as of December 1986) for estimating rral and rban flood-freqency characteristics. EXTRAPOLATION FOR EXTREME FLOODS To date, the USGS has pblished regression eqations for estimating the 500-year flood in only seven States. A procedre ill be given in the NFF program for extrapolating the regional regression eqations in any State to the 500-year flood. Basically, the extrapolation procedre consists of fitting a log-pearson Type III crve to the 2- to 100-year flood discharges given by NFF and extrapolating this crve to the 500-year flood discharge. The procedre consists of the folloing steps for a given atershed: 1. Determine the flood-peak discharges for selected retrn periods from the appropriate regional regression eqations given in NFF. At least three points are needed to define the ske coefficient reqired in a sbseqent step. Use of additional points improves the definition of the freqency crve that is defined by the regional eqations and helps to average ot any minor irreglarities that may exist in the relations among the regional eqations. The NFF program ill se all available regional eqations to define the freqency crve. 2. Fit a qadratic crve to the selected points on log-probability paper sing mltiple regression analysis. The variables sed in the regression analysis are the logarithms of the selected discharges and the standard normal deviates associated ith the corresponding probabilities. The prpose of this qadratic crve is to obtain a smooth crve throgh the selected flood-peak discharges from step 1 above. The qadratic crve is an approximation of the log-pearson Type III crve that ill be compted. 3. Determine the ske coefficient of the log-pearson Type III freqency crve that passes throgh the 2-, 10-, and 100-year floods defined by the qadratic crve. The ske coefficient is defined approximately by the formla (IACWD, 2) G = log (Q100/Q1ol I log (Q10/Q2l. 4. Replot (conceptally) the selected discharges and retrn periods sing a Pearson Type III probability scale defined sch that a freqency crve ith the compted ske plots as a straight line. This scale is defined by plotting probability vales p at positions x on the probability axis, here x is defined by the standardized Pearson Type III deviate (K vales) for the given ske and probability. A Wilson-Hilferty approximation (Kirby, 5) is sed to compte the K vale. 5. Fit a straight line by least sqares regression to the points plotted in step 4 and extrapolate this line to the 500-year flood-peak discharge. The variables sed in the regression analysis are the logarithms of the selected discharges and the Pearson Type III K vales associated ith the corresponding probabilities.

3 220 TRANSPORTATION RESEARCH RECORD 1290 Figre 1 is an example of a floodfreqency crve provided by this procedre for the Fenholloay River near Foley, Florida. The solid triangles shon in figre 1 are the regional flood-freqency vales as estimated by the eqations given in Bridges (6), hich ill be incorporated in the NFF program. The 500-year vale shon as a solid circle in figre 1 (12, 800 ft3/s) is estimated sing the extrapolation procedre described above. Note that the extrapolated 500-year vale is a reasonable extension (see dotted line) of the regional freqency crve. 20, ,----,---,---,-----,--, Cl z 8 10,000 (/) a.. tij ~ co ::::> ~ ' (.!) <{ i (/) Ci A Regional vales of the 2- to 5CJO-year flood discharges e Extrapolated vale of the 500-year flood discharge / / ~ () 100 5CJO Figre 1. freqency Fenholloay Florida RECURRENCE INTERVAL, IN YEARS Regional crve for River near floodthe Foley, The solid triangle shon in figre 1 (11,500 ft3/s) for the 500-year vale is the regional vale as obtained directly from the 500-year eqation given in Bridges (6). The 500-year flood for the Fenholloay River can be estimated ithot extrapolation since Florida is one of the seven States for hich 500- year regression eqations have been pblished. The difference beteen the to 500-year vales is percent. This is typical of several comparisons of extrapolated 500-year floods to pblished regional eqations that have indicated most reslts agree ithin pls or mins 15 percent. Figre 2 is a smmary of the inpt data, qestions, and responses that created the freqency crve in figre 1. The information reqired to compte the regional flood-freqency crve in figre 1 is that the atershed is in Region B ith a drainage area of 120 sqare miles and that 0.37 percent of the atershed area is covered by lakes. Figre 2 shos the regional flood freqency vales and their associated standard errors. For comparison and evalation, the NFF program ill compare each extrapolated 500-year flood-peak discharge ith the maximm floodenvelope crves given by Crippen (7). These flood-envelope crves ere developed by plotting maximm knon flood discharges against drainage areas for 17 flood regions of the United States. Therefore, these flood-envelope crves approximate the maximm floodpeak discharge that has been regionally experienced for a given size atershed. Since there is no freqency of occrrence associated ith the envelopecrve estimates, the comparison of these vales to the extrapolated 500-year flood is merely a qalitative evalation. In general, one old expect the extrapolated 500-year floodpeak discharge to be less than the envelope-crve vales, assming that several atersheds in a given region have experienced at least one flood exceeding the 500-year vale dring the period of data collection. For the Fenholloay River' near Foley, Florida, the 500-year flood estimates range from 11,500 to 12,800 ft3/s. The envelopecrve vale from Crippen (7) is 101,000 ft3/s given that the atershed is in region 3 (this information mst be provided by the analyst) as defined by Crippen (7).

4 Nationa1 F1ood Freqency Log Session NFF Log session started 07/03/ :04 Enter state id code : FL Enter name of basin nder stdy: FENHOLLOWAY RIVER NEAR FOLEY, FL. List of Hydrologic Regions in Florida Region Nmber Region Name 1 Region A 2 Region B 3 Region C Is basin contained in more than one hydrologic region? (Y/N) N Hydrologic region? (1-3) : 2 Region B parameters: Drainage Area (sq mi), DA ( ) : Lake Area(%), LK ( ) : 0.37 Enter maximm flood region ithin hich the basin is contained (See Report). Enter 0 if not applicable (e.g. otside of conterminos United States) : 3 Table of rral flood event vales FENHOLLOWAY RIVER NEAR FOLEY, FL. Recrrence Peak, Eq. Yrs. I Inter., yrs cfs % Std. Err. Record 1 ~;;:=-=====~.e:=:n.::c RQ RQ RQlO RQ RQ RQlOO RQ MAXIMUM FLOOD ENVELOPE = cfs List of Entered Parameters Name Vale DA : LK : Do yo ant to calclate a eighted average of observed and regression estimates? (Y/N) N Do yo ant to perform rban calclations? (Y/N) N Do yo ant to rite a flood freqency plot inpt file for TELAGRAF? (Y/N) N Do yo ant to compte a hydrograph for the rral peak calclated? (Y/N) N Do yo ant to do more flood freqency calclations in Florida? (Y/N) N Do yo ant to do flood freqency calclations in another state? (Y/N) N Program terminated. NFF Log session ended 07/03/ :05 Figre Smmary of inpt data, qestions and responses dring an interactive session ith the Nationa1 Flood Freqency Program

5 222 TRANSPORTATION RESEARCH RECORD 1290 FLOOD-HYDROGRAPH ESTIMATION Procedres also ill be provided in NFF for compting a flood hydrograph associated ith the 500-year flood-peak discharge and floods of other retrn periods. The flood hydrographs prodced by these procedres may be applicable only in certain parts of the contry. These comptations are based on procedres described by Inman (8) and are only briefly smmarized here. Using these procedres, a dimensionless hydrograph can be converted to a typical T-year hydrograph by estimating the T year flood-peak discharge and atershed lagtime (time from centroid of rainfall e"x"cess to centroid of rnoff). The flood hydrograph that reslts from this procedre is an average or typical hydrograph that corresponds to a given T-year flood-peak discharge. The volme nder the hydrograph may, or may not, have the same recrrence interval as the peak discharge. Briefly, the procedres for estimating flood hydrographs, as described by Inman ( 8), involves the folloing steps. Rainfall-rnoff data are sed to determine nit hydrographs and atershed lagtimes for selected atersheds in a region. These nit hydrographs are averaged to obtain an average nit hydrograph for the region or State. This average nit hydrograph is converted to an average dimensionless hydrograph by dividing the ordinates by the flood-peak discharge and the abscissa by the atershed lagtime. A typical T-year hydrograph is then obtained by mltiplying the ordinates of the dimensionless hydrograph by the T year peak discharge and mltiplying the abscissa by the atershed lagtime. The flood-peak discharge ill be estimated by regression eqations in the NFF program. The atershed lagtime for rban atersheds ill be estimated by regression eqations in the NFF program (Saer and others (3)). For rral atersheds, the vales for atershed lagtime ill be inpt by the ser. The compted T-year flood hydrograph can be sed along ith the peak discharge to determine bridge scor, flood-hazard risk, and other environmental impacts. Figre 3 is an example of a typical flood hydrograph for a 50 0-year peak discharge of 12,800 ft3/s for the Fenholloay River near Foley, Florida. Flood hydrographs, sch as the one shon in figre 3, can be sed to determine the dration that flo is above a given discharge. This information shold be sefl in evalating scor at bridges. 13,000 0 z 12, ,000 CJ) 10,000 Cl. f iii 7000 ::i &XXl ii: (.!) <I: 4000 I :n:xl CJ) TIME, IN HOURS Figre 3. Typical hydrograph associated ith year flood-peak discharge Fenholloay River near Florida. SUMMARY flood a 500- for the Foley, The procedres to be provided in the NFF microcompter program give the highay engineer relatively easy and straightforard methods of estimating the flood-peak discharge and flood hydrograph of extreme floods, sch as the 500-year flood. The rral flood discharge estimates are based on regression eqations that have been developed by the USGS over the last 15 years or so. The eqations to be given in the NFF program ill be pdated periodically as ne eqations are developed. Procedres ill also be available for estimating flood-peak discharges for rban atersheds. If Stateide rban flood-freqency procedres are not available, then nationide techniqes, described by Saer and others (3), can be sed. Finally, a flood hydrograph, associated ith the T-year rral or rban flood discharge, can also be compted to aid in the determination of scor dring extreme floods.

6 Thomas el al. 223 REFERENCES 1. Jennings, M.E., and Cookmeyer, E.N., II. Plans For National Flood Freqency by Microcompter: Proceedings of the American Society of Civil Engineers National Conference on Hydralic Engineering, p , Interagency Advisory Committee on Water Data, Gidelines For Determining Flood Flo Freqency: Blletin 17B of the Hydrology Sbcommittee, 183 p., Saer, V.B., Thomas, W.O., Jr., Stricker, V.A., and Wilson, K.V. Flood Characteristics of Urban Watersheds in the United States: U.S. Geological Srvey Water-Spply Paper 2207, 63 p., Thomas, W.O., Jr. Techniqes Used By The U.S. Geological Srvey In Estimating The Magnitde and Freqency Of Floods: Catastrophic Flooding, edited by L. Mayer and D. Nash, The Binghamton Symposia in Geomorphology, International Series, No. 18, p , Kirby, W.H. Compter-oriented Wilson-Hilferty transformation that preserves the first three moments and the loer bond of the Pearson Type 3 distribtion: Water Resorces Research, Vol. ~, No. 5, p , Bridges, W. C. Techniqes For Estimating Magnitde and Freqency Of Floods In Florid U.S. Geological Srvey Water-Resorces Investigations , 45 p., Crippen, J.R. Envelope Crves for Extreme Flood Events: American Society of Civil Engineers, Vol. 108, No. HYlO, p , Inman, E.J. Simlation of Hydrographs For Georgia Streams: Geological Srvey Water-Spply 2317, 26 p., Flood U.S. Paper

Network Sampling to Estimate Distribution of Pavement Condition and Costs

Network Sampling to Estimate Distribution of Pavement Condition and Costs 8 Transportation Research Record 148 Netork Sampling to Estimate Distribtion of Pavement Condition and Costs J. TEMPLETON, R. L. LYTTON, and A. GARCIA-DIAZ ABSTRACT Srveying the pavement condition of a