The FEH 2025 statistical method update

This report presents updates to the FEH statistical method resulting from improvements to four catchment descriptors used prominently in the method: SAAR, FARL, BFIHOST and URBEXT, and 19 years of additional data collection and period-of-record review at UK river gauging stations. The structure and application procedure of the FEH statistical method are unchanged; this update concerns the choice of catchment descriptors to use in the method and their corresponding numerical coefficients. Due to the high performance of the current QMED equation, the new equation offers only a marginally improved fit to the data. However, estimates of the current flood frequency in ungauged catchments should be improved, due to better digital representation of those catchments through improved catchment descriptors. Pooled estimates of rarer floods are improved more substantially due to the increased typical record length; uncertainty in L-moments continues to reduce as record length increases far beyond the point at which uncertainty in QMED becomes small. This method update reiterates that donor stations should be selected by centroid-centroid distance only, since model error, which the donor transfer is intended to correct, is the part of QMED that cannot be estimated by the equation and its associated catchment descriptors i.e. AREA, SAAR(9120), BFIHOST19 and FARL(2015).