Seasonal response function for daily streamflow investigation

Abstract

In this study, a linear model with variable transfer response function has been applied for daily streamflow investigation purposes. The reason for a time-varying approach has to be found in the non-linear features of the rainfall-runoff transformation, with seasonal variability affecting model performance when a linear approach is used to describe it. Depending on the number of seasons which can be identified within the hydrologic year, a time-varying response function would require a number of model parameters greater than in the case of a constant response function. Model parameters along with other difficulties related to the identification procedure, the estimation procedure, the robustness of the solutions and the parameters conceptual meaning are to be considered. A general procedure to identify seasonally variable parameters, generating a seasonal variable response function, is proposed to overcome such difficulties. We refer to a recently proposed linear shot noise approach and to a few case studies, showing the improved model descriptive capabilities and the possibility to investigate relationships between climate characteristics and model parameters variability.