An application-oriented protocol for flood frequency analysis based on botanical evidence

Flood risk assessment and management typically rely on flood frequency analysis (FFA), such that planning and countermeasures can be designed based on the discharge that has to be expected at a given location for a given return period. In mountain streams, systematic flow time series are often very short or completely missing, which significantly reduces the reliability of FFA. In fast-flowing mountain streams, the inclusion of non-systematic data obtained from botanical evidence (BE) is seen as an optimal alternative to extend systematic data back in time. However, no comprehensive protocol has been defined so far to tackle FFA using BE. On the basis of recent case studies, we present here an application-oriented protocol with guidelines on how to combine systematic and non-systematic data in FFA containing BE. This study is based on work realized in different mountain streams located in Spain, Poland and India, representing quite diverse physiographic characteristics and differing hydrological regimes. We organize the protocol along the different steps that are typically realized in BE-based FFA: i) dating of floods from BE; ii) estimation of flood flows from paleostage indicators (PSI) and hydrodynamic modelling; as well as iii) FFA using the expected moments algorithm (EMA). The ubiquity of trees growing along (mountain) streams, their longevity and the often large number of flood-affected trees makes them an almost unbeatable data source that can be employed readily and with reasonable efforts to improve the reliability of FFA, especially in data-scarce regions. In addition, the EMA represents a highly efficient tool for the collection of information contained in BE as it can be used with interval, censored and binomial-censored data and on any distributional family that can be operated with the method of moments. Accordingly, we call for more work incorporating BE into FFA in mountain streams, such that flood hazard and risk assessment can be undertaken more robustly and, therefore, more effective risk mitigation measures can be envisaged.