Estimation of instantaneous peak flow from maximum mean daily flow by regionalization of catchment model parameters

Abstract

AbstractRegionalization methods have been effectively used in many hydrological studies, such as regional flood frequency analysis and low flows. However, there is no study to estimate the instantaneous peak flow (IPF) from maximum mean daily flow (MDF) using hydrological models with regionalized parameters. In this paper, the semidistributed conceptual hydrological model Hydrologiska Byråns Vattenbalansavdelning is operated on a daily time step for 18 catchments in the Aller‐Leine basin, Germany. The model is calibrated on four different flow statistics, including winter/summer extremes distribution and flow duration curves. The model parameter values are predefined with the associated catchment descriptors by a transfer function. Two different regionalization schemes are investigated: one is carried out for all the catchments in the study area; the other one is only performed for several catchments within a cluster. The k‐means algorithm is used to 12 different catchment characteristics from all 18 catchments as the partitional clustering algorithm. Subsequently, the general extreme value distributions are fitted to the modeled MDFs, which are then transferred into IPF quantiles using a multiple regression model.The results show that (a) the uncertainty resulted from model parameter regionalization for the estimation of IPFs is much smaller than the error when using MDFs instead of IPFs; (b) the hydrological responses of the clustered catchments located in the flat areas are, in general, not as homogeneous as the ones in high elevated regions; and (c) the model with the parameters derived from the same regionalization coefficients within a cluster performs better using the corresponding parameters estimated through all the catchments.