An evaluation of selected extraordinary floods in the United States reported by the U.S. Geological Survey and implications for future advancement of flood science

Thirty flood peak discharges determine the envelope curve of maximum floods documented in the United States by the U.S. Geological Survey. These floods occurred from 1927 to 1978 and are extraordinary not just in their magnitude, but in their hydraulic and geomorphic characteristics. The reliability of the computed discharge of these extraordinary floods was reviewed and evaluated using current (2007) best practices. Of the 30 flood peak discharges investigated, only 7 were measured at daily streamflow-gaging stations that existed when the flood occurred, and 23 were measured at miscellaneous (ungaged) sites. Methods used to measure these 30 extraordinary flood peak discharges consisted of 21 slope-area measurements, 2 direct current-meter measurements, 1 culvert measurement, 1 rating-curve extension, and 1 interpolation and rating-curve extension. The remaining four peak discharges were measured using combinations of culvert, slope-area, flow-over-road, and contracted-opening measurements. The method of peak discharge determination for one flood is unknown.

Changes to peak discharge or rating are recommended for 20 of the 30 flood peak discharges that were evaluated. Nine floods retained published peak discharges, but their ratings were downgraded. For two floods, both peak discharge and rating were corrected and revised. Peak discharges for five floods that are subject to significant uncertainty due to complex field and hydraulic conditions, were re-rated as estimates. This study resulted in 5 of the 30 peak discharges having revised values greater than about 10 percent different from the original published values. Peak discharges were smaller for three floods (North Fork Hubbard Creek, Texas; El Rancho Arroyo, New Mexico; South Fork Wailua River, Hawaii), and two peak discharges were revised upward (Lahontan Reservoir tributary, Nevada; Bronco Creek, Arizona). Two peak discharges were indeterminate because they were concluded to have been debris flows with peak discharges that were estimated by an inappropriate method (slope-area) (Big Creek near Waynesville, North Carolina; Day Creek near Etiwanda, California). Original field notes and records could not be found for three of the floods, however, some data (copies of original materials, records of reviews) were available for two of these floods. A rating was assigned to each of seven peak discharges that had no rating.

Errors identified in the reviews include misidentified flow processes, incorrect drainage areas for very small basins, incorrect latitude and longitude, improper field methods, arithmetic mistakes in hand calculations, omission of measured high flows when developing rating curves, and typographical errors. Common problems include use of two-section slope-area measurements, poor site selection, uncertainties in Manning’s n-values, inadequate review, lost data files, and insufficient and inadequately described high-water marks. These floods also highlight the extreme difficulty in making indirect discharge measurements following extraordinary floods. Significantly, none of the indirect measurements are rated better than fair, which indicates the need to improve methodology to estimate peak discharge. Highly unsteady flow and resulting transient hydraulic phenomena, two-dimensional flow patterns, debris flows at streamflow-gaging stations, and the possibility of disconnected flow surfaces are examples of unresolved problems not well handled by current indirect discharge methodology. On the basis of a comprehensive review of 50,000 annual peak discharges and miscellaneous floods in California, problems with individual flood peak discharges would be expected to require a revision of discharge or rating curves at a rate no greater than about 0.10 percent of all floods.

Many extraordinary floods create complex flow patterns and processes that cannot be adequately documented with quasi-steady, uniform one-dimensional analyses. These floods are most accurately described by multidimensional flow analysis.

Within the U.S. Geological Survey, new approaches are needed to collect more accurate data for floods, particularly extraordinary floods. In recent years, significant progress has been made in instrumentation for making direct discharge measurements. During this same period, very little has been accomplished in advancing methods to improve indirect discharge measurements. Greater use of paleoflood hydrology could fill many shortcomings of U.S. Geological Survey flood science today, such as enhanced knowledge of flood frequency. Additional links among flood runoff, storm structure, and storm motion would provide more insight to flood hazards. Significant improvement in understanding flood processes and characteristics could be gained from linking radar rainfall estimation and hydrologic modeling. Additionally, more could be done to provide real-time flood-hazard warnings with linked rainfall/runoff and flow models.

Several important recommendations are made to improve the flood-documentation capability of the U.S. Geological Survey. When very large discharges are measured by current meter or hydroacoustics, water-surface slope should be measured as well. This measurement would allow validation of roughness values that can significantly extend the discharge range of verified Manning’s n for 1-dimensional and 2-dimensional flow analyses. At least two of the floods investigated may have had flow so unstable that large waves affected the interpretation of high-water marks. Instability criteria should be considered for hydraulic analysis of large flows in high-gradient, smooth channels.

The U.S. Geological Survey needs to modernize its toolbox of field and office practices for making future indirect discharge measurements. These practices could include, first and foremost, a new peak-flow file database that allows greater description and interpretation of flow events, such as stability criteria in high-gradient, smooth channels, debris flow documentation, and details of flood genesis (hurricane, snowmelt, rain-on-snow, dam failure, and the like). Other modernized practices could include (a) establishment of calibrated stream reaches in chronic flash flood basins to expedite indirect computation of flow; (b) development of process-based theoretical rating curves for streamflow-gaging stations; (c) adoption of step-backwater models as the standard surface-water modeling tool for U.S. Geological Survey field offices; (d) development and support for multidimensional flow models capable of describing flood characteristics in complex terrain and high-gradient channels; (e) greater use of the critical-depth method in appropriate locations; (f) deployment of non-contact instruments to directly measure large floods, rather than attempting to reconstruct them; (g) increased use of paleoflood hydrology; and (h) assurance that future collection of hydro-climatic data meets the needs of more robust watershed models.