Transferring measured discharge time series: Large‐scale comparison of Top‐kriging to geomorphology‐based inverse modeling

Few methods directly transfer streamflow measurements for continuous prediction of ungauged catchments. Top‐kriging has been used mainly to predict the statistical properties of runoff but has been shown to outperform traditional regionalization approaches of rainfall‐runoff models. We applied the Top‐kriging approach across the Loire River basin and compared predictions to a geomorphology‐based approach. Whereas Top‐kriging uses spatial correlation, the other approach has the advantage of being more physically based by using a well‐known geomorphology‐based hydrological model (WFIUH) and its inversion. Both approaches require an equal degree of calibration and provide similar performances. We also demonstrate that the Ghosh distance, which considers the nested nature of catchments, can be used efficiently to calculate weights and to identify the suitability of gauged catchments for use as donor catchments. This result is particularly relevant for catchments with Strahler orders above five, i.e., where donor catchments are more strongly nested. Key Points: The geomorphological inversion approach and Top‐kriging are equally efficient for continuous streamflow simulation Small upstream catchments have higher uncertainties than larger catchments for both methods A rescaled Ghosh distance provides the best weighting of donor catchments for geomorphological inversion