Innovative methodologies for enhancing the regionalisation of rainfall-runoff model parameters

Abstract

The present research work focuses on the regionalisation of rainfall-runoff model parameters, fundamental for the implementation of hydrological models in ungauged basins and needed to reproduce the actual sequence of river discharge in time. Regionalisation of parameters is based on the transfer of information from hydrologically similar gauged basins to ungauged basins. This work provides further insights on parameter regionalisation and catchment similarity through the application of innovative methodologies to support the existing knowledge. The first experiment develops a methodology to test the robustness of regionalisation procedures to the availability of data in the study region. In particular, the effect of the density of streamflow gauging stations and the topological relationships between their corresponding drainage basins on the different regionalisation techniques is investigated. Such work provides useful information for the choice of the most appropriate method, based on data availability in the region. The focus is then moved to the value of hydrological similarity at sub-basin scale. Driven by the fact that similarity is generally defined between entire catchments, neglecting some significant differences in the within-basin rainfall-runoff transformation processes, a methodology to differentiate hydrological processes at sub-basin scale and to transfer model parameters from similar sub-basins is proposed. The analysis is based on the diversification of the parameter values, and therefore of the corresponding hydrological dynamics, across elevation, one of the main factors influencing the runoff generation processes. Finally, an innovative catchment signature is proposed for improving our knowledge about hydrological similarity, meant for the delineation of hydrologically similar regions. A new methodology for identifying the dominant rainfall-runoff transformation dynamics is presented: the interaction between the entire time-series of runoff generation forcings and runoff itself is quantified taking advantage of the concepts of the Information Theory and used to characterise catchments with promising results.