PUBLICATIONS--Transmittal of National Research Council Report on Estimating Probabilities of Extreme Floods In Reply Refer To: May 16, 1988 WGS-Mail Stop 415 OFFICE OF SURFACE WATER TECHNICAL MEMORANDUM NO. 88.08 Subject: PUBLICATIONS--Transmittal of National Research Council Report on Estimating Probabilities of Extreme Floods Attached for your information is a National Research Council (NRC) report on methods and recommended research for estimating probabilities of extreme floods. Although the report does not recommend specific procedures for opera- tional use, its reviews of statistical and modeling methodologies and its recommendations for research and development should be of interest to all those concerned with evaluation of risks due to extreme floods. The extreme floods that are the subject of this report are those of such low probability as to be unlikely to be included at all in hydrologic records of reasonable length. Such events are of interest in certain engineering situa- tions such as dam spillways and nuclear powerplants where very high relia- bility over a long design lifetime is required. The events of interest generally would be considered outliers in those records in which they did occur. Traditional methods of single-site flood-frequency analysis are inadequate for reliable estimation of the magnitude or probability of such events. For example, only one 50-year record out of 20, on average, would contain an event of 1,000-year recurrence interval (0.001 annual probability). Such one-in-a- thousand events are substantially underrepresented in the 19 50-year records out of 20 where they are absent and substantially overrepresented in the one record in 20 where they are present. Thus, no single-site record can adequately represent the magnitude and frequency of such rare events. The committee therefore considered procedures by which multisite data sets and regional rainfall-runoff data might be used. The committee's report includes generalized reviews of multisite statistical estimation, statistical modeling of rainfall, rainfall-runoff transformations, and characteristics of basic rainfall and streamflow data obtained from sys- tematic, historical, and paleohydrologic data-collection procedures. Although the report does not provide or recommend specific procedures for operational use, it does make recommendations for research and development. The main recommendations include: further development of index-flood-type regional (multisite) estimation procedures; development of multisite (regionalized) nonparametric extrapolation techniques; development of stochastic rainfall models with emphasis on upper-tail behavior, spatial structure, and storm transposition; evaluation and testing of the performance of rainfall-runoff models in extreme floods; research and operational development of paleohydro- logic methods; and research on statistical models of historical and paleohy- drologic data generation and utilization. Of particular interest to the U.S. Geological Survey is the committee's recommendation to compile available data in suitable format for completion of nationwide catalogs of extreme storms and floods. OFFICE OF SURFACE WATER TECHNICAL MEMORANDUM NO. 88.08 2 The present NRC study complements an interagency work group study of the feasibility of assigning a probability to the probable maximum flood (PMF). That study report was distributed by Surface Water Branch Technical Memorandum 87.01 dated December 2, 1986. The PMF report focused on the unavailability of procedures for estimating probabilities of PMF-scale floods, on the inability of available statistical procedures and models to provide credible estimation of probabilities of PMF-scale floods, and on fundamental theoretical and prac- tical obstacles that had to be circumvented before credible probability esti- mates could be made. The NRC report, on the other hand, emphasizes the poten- tial of concerted research and development efforts to achieve progress in this area. Please bring the attached report to the attention of those in your office who have an interest in the frequency analysis of extreme floods. Ernest F. Hubbard Acting Chief, Office of Surface Water Attachment WRD Distribution: FO-LS, SL