In practice, the assumption about independent data is usually taken into account by selecting annual maximum values of the variable of interest (e.g. annual maximum discharge). In this way, an *annual maximum series* is established, which are the most common type of series used in flood frequency analysis.

The annual maximum series contains information only on the largest peak flows from every year, which are not necessarily all floods. It may happen that the second (or third, etc.) largest value from a wet year represents severe flood and is larger than annual maximum in a dry year; that kind of information on floods is lost in annual maximum series. An alternative is to create *partial duration series*, a series of data which are greater than some predefined base value. In this manner, there could be more than one peak flow value per year, but care should be taken that the selected flood peaks are independent.

A special case of the partial duration series is the *annual exceedance series*, in which the base value is selected so that the number of values is equal to the number of years of the record.