HR: 1330h
AN: H12D-1009 [PDF]
TI: Regional Flood Frequency Equations: What Level of Complexity is Rational?
AU: * Perica, S
EM: perica@eng.utah.edu
AF: University of Utah, Department of Civil and Environmental Engineering, 122 South Central Campus Drive,
Salt Lake City, UT 84112 United States
AU: Mizukami, N
EM: N.Mizukami@m.cc.utah.edu
AF: University of Utah, Department of Civil and Environmental Engineering, 122 South Central Campus Drive,
Salt Lake City, UT 84112 United States
AU: Merrill, A
EM: annmerrill@utah.gov
AF: University of Utah, Department of Civil and Environmental Engineering, 122 South Central Campus Drive,
Salt Lake City, UT 84112 United States
AB:
The "single-return-period" prediction equation of the log-log multivariate regression form in which a peak discharge of a
specific return period is related to one or more watershed and meteorologic characteristics is the most frequently used
regional flood frequency procedure in the US for ungaged, uregulated rural streams. Equations are developed for a region that
is identified as homogeneous based on underlying hydrologic/meteorologic and geologic/soil properties. The number and type
of explanatory variables used in regression equations vary. For example, most regional regression equations developed by the
US Geological Survey and compiled into "The National Flood Frequency Program," are based on watershed characteristics such
as: drainage area, mean basin elevation, and channel slope. Other explanatory watershed variables used include parameters
such as: storage area of lakes/ponds, forest cover, channel length, basin shape, high elevation index. Climatic
characteristics present in some of the equations consist of mean annual rainfall, rainfall amount for a specified duration,
mean annual snowfall and/or minimum mean January temperature.
The development of GIS based models, such as ArcHydro or Watershed Modeling System, has created an opportunity to easily
produce extensive sets of hydrologic parameters that could be investigated as possible predictors of T-year discharges. As a
result, prediction equations tend to be even more complex than they used to be. However, based on our results, it appears
that such equations may actually generate less accurate flood estimates than very simple equations that include only one or
two predictors. Equations are highly sensitive to uncertainties (errors) in explanatory variables, both in calibration and in
prediction mode. Based on our study for several watersheds in Utah we'll try to answer the question on how much complexity
in prediction equations is really rational.
DE: 1821 Floods
DE: 1836 Hydrologic budget (1655)
DE: 1860 Runoff and streamflow
DE: 1869 Stochastic processes
DE: 1884 Water supply
SC: Hydrology [H]
MN: 2003 Fall Meeting