Flood Frequency Analysis
Also known as: At-Site Flood Frequency Analysis, Annual Maximum Flood Frequency, Extreme Value Flood Analysis, Design Flood Estimation
Flood frequency analysis estimates how often floods of a given magnitude occur at a river site by fitting an extreme-value probability distribution to the record of annual maximum discharges and then inverting it to read off design floods for specified return periods. The classical approach uses the Gumbel distribution, the limiting form for maxima of light-tailed variables; the more general Generalized Extreme Value (GEV) distribution adds a shape parameter that lets the tail be lighter or heavier, while the log-Pearson Type III distribution is the U.S. federal standard codified in USGS Bulletin 17C. Hosking, Wallis, and Wood's 1985 work on probability-weighted moment estimation of the GEV made robust at-site fitting practical, and Bulletin 17C (England et al., 2018) sets out the modern operational procedure. The output — the 100-year flood, the 500-year flood — underpins dam design, floodplain mapping, and infrastructure standards worldwide.
Key highlights
- Grounded in extreme-value theory, which gives a principled justification for the GEV/Gumbel family as the limiting distribution of block maxima.
- Produces directly usable engineering outputs — design floods for any return period — from a single station's record.
- L-moment / probability-weighted-moment estimation is robust and reliable even for the short records common in hydrology.
- Standardized and codified (e.g., USGS Bulletin 17C), giving a defensible, reproducible procedure for regulatory and design use.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use flood frequency analysis when you have a reasonably long, homogeneous record of annual maximum discharges at a site and need design-flood magnitudes for specified return periods — for dam spillway sizing, levee and bridge design, floodplain delineation, or insurance and zoning standards. It is appropriate when the flood-generating process is approximately stationary, when peaks are independent between years, and when the record length is adequate relative to the return periods of interest (extrapolating a 1000-year flood from 25 years of data is hazardous). It is less suitable at ungauged sites, where regional flood frequency analysis is needed instead; where strong non-stationarity from climate change, urbanization, or reservoir regulation invalidates the identically-distributed assumption; or where the annual maximum series mixes distinct flood populations that should be modeled separately. When only a few large events matter and much information is discarded by taking one peak per year, a peaks-over-threshold formulation can be more efficient.
Strengths & limitations
- Grounded in extreme-value theory, which gives a principled justification for the GEV/Gumbel family as the limiting distribution of block maxima.
- Produces directly usable engineering outputs — design floods for any return period — from a single station's record.
- L-moment / probability-weighted-moment estimation is robust and reliable even for the short records common in hydrology.
- Standardized and codified (e.g., USGS Bulletin 17C), giving a defensible, reproducible procedure for regulatory and design use.
- Extrapolating to return periods far beyond the record length produces estimates dominated by the assumed tail shape and large sampling uncertainty.
- The method assumes a stationary, identically-distributed series, which climate change, land-use change, and reservoir operation increasingly violate.
- Taking only one peak per year discards information about other large floods within a year, reducing statistical efficiency.
- Results are sensitive to the choice of distribution family and to outliers and historical events, which can shift the fitted tail substantially.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
What does a '100-year flood' actually mean?
It is the flood magnitude with a 1% probability of being equalled or exceeded in any single year — that is, an annual exceedance probability of 1/100, the inverse of the return period. It is not a flood that occurs once every hundred years on a schedule. Because each year is an independent trial, two '100-year floods' can occur in consecutive years, and over a 30-year mortgage the chance of seeing at least one is about 26%. Flood frequency analysis produces this quantity by fitting an extreme-value distribution to annual peaks and reading off the value at the 99th percentile.
How should I choose between the Gumbel and GEV distributions?
The GEV is the more general model; its shape parameter ξ determines whether the tail is light (ξ = 0, the Gumbel case), heavy (ξ > 0), or bounded (ξ < 0). Fix ξ = 0 only if you have a strong physical or statistical reason to believe the tail is light, since wrongly imposing Gumbel on a heavy-tailed series severely underestimates rare floods. With adequate record length, estimate ξ from the data using L-moments and test whether it differs from zero; with very short records the extra parameter can be poorly determined, which is one reason regional pooling is valuable.
How long a record do I need, and how far can I extrapolate?
A rough rule is that you should not extrapolate much beyond two to four times the record length without strong caution: a 30-year record can support a defensible 100-year estimate but only a highly uncertain 1000-year one. Bulletin 17C recommends at least about 10 years as a bare minimum and far more for reliable design floods, and it provides methods to incorporate historical and paleoflood information to effectively lengthen the record. Always accompany extrapolated quantiles with confidence intervals, which widen sharply at long return periods.
Sources
- 1.England, J. F., Jr., Cohn, T. A., Faber, B. A., Stedinger, J. R., Thomas, W. O., Jr., Veilleux, A. G., Kiang, J. E., & Mason, R. R., Jr. (2018). Guidelines for Determining Flood Flow Frequency — Bulletin 17C. U.S. Geological Survey Techniques and Methods, book 4, chap. B5, 148 p.
- 2.Hosking, J. R. M., Wallis, J. R., & Wood, E. F. (1985). Estimation of the Generalized Extreme-Value Distribution by the Method of Probability-Weighted Moments. Technometrics, 27(3), 251-261.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). Flood Frequency Analysis. ScholarGate. https://scholargate.app/disaster-studies/flood-frequency-analysis