Floods Glossary

frequency analysis

Frequency analysis helps to draw conclusions about the frequency of flood events at a certain gauge. In the simplest case the highest annual values of a daily discharge series are taken as input. These values are arranged according to size (descending) and attributed with ranks. These ranks are divided by the number of observation years in the time series (plus 1 - for methodical reasons). The longer the time series, the better the resulting exceedance probability P as a measure for the true (unknown) flood probability. The empirical probabilities (resp. plotting positions) and statistical moments of a time series are taken to fit different distribution functions. The annuality is calculated from the distribution functions. To enhance the significance of a statement like: "the exceedance probability is 0.01 (1%)" the reciprocal value 1/P = T is more commonly used. This value is called recurrence interval. In our case (P = 0.01) that would mean a 1%-probability to encounter one (on average) flood event exceeding a recurrence interval of T = 100 for an observation point in any year. By no means does such an event happens every 100 years or at least once in a hundred years.