Investigating the impact of predictive uncertainty in rainfall-runoff modelling on storage reliability estimates using Bayesian total error analysis

It is a common approach to extend streamflow records using longer term rainfall data and a calibrated rainfall-runoff model. However, the impact that the predictive uncertainty in the simulated streamflow has on evaluating water resource systems, e.g. storage reliability, is rarely assessed. This is likely to be due to the lack of a robust framework for quantifying the uncertainty in the parameters and predictions of conceptual rainfall runoff (CRR) models - which remains a key challenge for hydrological science. The Bayesian total error analysis (BATEA) provides a systematic approach to hypothesize, infer and evaluate probability models describing input, output and model structural error. This study utilizes results from a recent case study which critically evaluated the predictive uncertainty of BATEA compared to traditional calibration approaches (standard least squares (SLS) and weighted least squares (WLS)) and found that BATEA provided probabilistic predictions that are more consistent with the observations than both WLS and SLS. In this study, the sequent peak algorithm is used to estimate the storage reliability while accounting for the predictive uncertainty in the streamflow simulations. Comparing the results from BATEA, WLS and SLS it was found that for low demands the difference in the storage reliability estimates was minimal. However, as the demand increased, SLS and WLS tended to overestimate the storage reliability, relative to BATEA. This result has implications for common streamflow extension techniques and highlights that the predictive uncertainty in simulations needs to be taken into account when assessing water resource management options.