Flood Frequency Analysis

Overview

Flood frequency has often been used in the design of engineering structures such as culverts, bridges and dams. Flood frequency analysis is used to provide a predictive capacity to the stochastic nature of stream flow. We use the existing, preferably long-term stream flow data to predict the likely magnitude of events of a given return period.

The Flood Frequency Analysis tool selects the maximum discharge for each year of the record (Figure 1) and ranks the selected floods in order of magnitude. Once the peak flows have been determined, the occurrence of outliers (upper or lower) in the flood series is considered and these outliers are discarded (see Institution of Engineers Australia (1987) for the procedure). Then a Log-Pearson Type 3 (LP3) curve is fitted to the remaining flood series to predict average recurrence interval (ARI) values (Institution of Engineers Australia, 1987).

The Flood Frequency Analysis builds a predictive model based on existing data and, as with all such empirical models, the accuracy is greatly enhanced with a large data set. A general rule for flood frequency analysis is that the longer the flow record then the more accurate the predicted ARI.

The data used to determine the flood frequency commences at the start of a user-defined “water year” and ends on the date that describes the maximum number of entire years of record. For example, if 15.5 years of record have been entered, and the start of the water year is 1 January, then the flood frequency will be calculated based on the first 15 years of the record and the final 6 months will be discarded.

The annual series flood frequency is calculated using the largest flood in any given year of the record. No measure of independence of flood events is considered. The user should commence the analysis period (i.e the start of the water year) during low-flow months of record.

The Flood Frequency Analysis tool is suitable for long return period flows (i.e. greater than 10 years). For predicting floods with return periods of less than 10 years a partial series analysis is more suitable (Institution of Engineers Australia, 1987). Partial-series analysis is not currently available in Source.

Definitions

Term

Definition

Annual exceedance probability (AEP)

The probability of occurrence of a flood of given size or larger in any one year (e.g. 1% flood)

Average recurrence interval (ARI)

The average period in years between the occurrence of a flood of a given size or larger (e.g. 100 year flood).

A one year ARI flood will on average occur every year, a 2 year ARI every 2 years an so on:

ARI = 1/AEP

Creation

Flood Frequency Analyses can only be created for flow time series (i.e. time series with units of volume/time).

Flood Frequency Analyses can be created for a single time series or one or more time series in a custom chart. To a flood frequency analysis for a single time series:

right-click on the time series name in the Results Manager tree menu
select Statistics » Flood Frequency Analysis

For a custom chart:

right-click on the custom chart name in the Results Manager custom charts and statistics pane
select Statistics » Flood Frequency Analysis

Once created, the Flood Frequency Analysis will be visible in the Results Manager custom charts and statistics pane.

Figure 2. Flood Frequency Analysis input screen, illustrating results for two time series.

Configuration

The Flood Frequency Analysis is configured by setting the water Year Start day and month (Figure 2).

Outputs

Overview Tab

The Overview tab (see Figure 2) provides a table recording the values listed below:

Recorded Output	Description
General
Time Series	The name of the time series used in the analysis
Start	The start date of the analysis
End	The end date of the analysis
Number of Years	The number of years used in the analysis
Number of Floods	The number of floods identified
Log-Pearson Type 3 Parameters
LP3 Mean	The mean of the fitted LP3 distribution
LP3 Std	The standard deviation of the fitted LP3 distribution
LP3 Skew	The skew of the fitted LP3 distribution

Chart Tab

The Chart tab (see Figure 2) plots the annual flood peaks and the fitted LP3 curve for each time series.

Table Tab

The Table tab (see Figure 2) displays a table of the flood magnitudes predicted by the LP3 distribution for pre-defined return periods (ARIs) ranging from 1.01 years to 500 years.

The accuracy of the result for a given ARI depends on the length of record. The length of record should generally be much larger than the ARI to avoid errors in the extrapolation of the data.

Acknowledgements

This material has been adapted from:

Marsh, N. (2004) RAP river analysis package: user guide, version 1.1. CRC for Catchment Hydrology, Australia, Jan 2004. www.toolkit.net.au/rap

References

Institution of Engineers Australia (1987) Australian Rainfall and Runoff: a Guide to Flood Estimation, Revised Edition. Institution of Engineers Australia.