Flood Frequency Analysis


Flood Frequency Analysis, the determination of flood flows at different recurrence intervals, is a common problem in hydrology. The standard procedure to determine probabilities of flood flows consists of fitting the observed stream flow record to specific probability distributions. However, this procedure only works for basins;

  • that have stream flow records, where stream flow records are 'long enough' to warrant statistical analysis;
  • where flood flows are not appreciably altered by reservoir regulation, channel improvements (levees) or land use change.




Continuous hydrologic simulation is a valuable tool to determine flood frequencies in gaged watersheds that have short streamflow records or are heavily regulated. Meteorologic data for most watersheds in the United States extend 40 to 80 years. Stream flow records, if available at all, are often much shorter. Continuous hydrologic simulation can use the observed meteorologic data available to extend the existing stream flow record from a few years to 40 or 80 years. The extended record can then be fitted to a probability distribution.

Hydrologic simulation can also be used to determine flood frequencies in ungaged streams. In this case the model is calibrated to a near-by, hydrologically similar, gaged stream. The model parameters are then adjusted to reflect the physical changes between the calibration watershed and the ungaged watershed. Finally, all the available observed meteorologic data are used to create a long stream flow record for the ungaged stream, which can then be fitted to a statistical distribution. This method produces much better results than alternative simplified approaches such as comparison to similar watersheds, or indirect approaches that equate runoff frequency to precipitation frequency (such as unit hydrographs or the rational formula).

Land use changes can have a significant effect on flood flow frequencies. Historic stream flow records may be non-stationary for basins in which widespread urbanization is taking place. Hydrologic simulation uses historic flow records to calibrate to the historic conditions and it then incorporates the effects of future urbanization.

Similarly, in basins where reservoir regulation significantly affects flood flows, continuous hydrologic simulation can isolate the effect of the reservoir. The model makes it possible to compare flood levels with and without the reservoir and for various reservoir operations. When the possibility of a dam failure is considered, the results from hydrologic simulation can be used together with a full equations routing model. (1)

In addition to generating or extending stream flow records, hydrologic simulation can be used to study the validity of an assumed probabilistic distribution for peak flows. The U.S. Water Resources Council recommends the use of the Log-Pearson Type III frequency distribution. (2) However, it has been Hydrocomp's experience that this distribution can produce unreasonable results in semi-arid areas: The calculated runoff intensity above certain return periods greatly exceeds the precipitation for the same return period. Since continuous hydrologic simulation maintains a continuous accounting of soil moistures, it provides a unique tool to analyze the complex relationship between frequencies of precipitation, soil moisture and runoff.


(1) For 'Dam Break' analysis, Hydrocomp uses output from the HFAM as input to FEQ, a full equations routing model.
(2) "Guidelines for Determining Flood Flow Frequency", Bulletin #17 of the Hydrology Committee, March 1976, U.S. Water Resources Council, Washington , D.C.

Return to Index
Hydrocomp, Inc.