How much data is really required for rainfall-runoff model calibration? A summary metric view.

Abstract:

The parameters of conceptual rainfall-runoff models do not represent directly measurable entities and must therefore be estimated from measurements of the watersheds response to rainfall. Various authors have investigated the relationship between the length of the data used for model calibration and the accuracy and precision of the model parameters. In this talk I will revisit this topic but using summary statistics of the discharge data, rather than the discharge data itself. We hypothesize that the use of summary metrics should lead to model realizations that better represent the hydrologic characteristics of the watershed data. Our conclusions with diagnostic model evaluation shed new light on the information content of data, and if properly extracted, support the use of a considerably shorter calibration time period.