A fresh look at the limits of acceptability approach for the calibration of hydrological models

This study aims to look into the references and methods for calibrating hydrological models. Due to climate change and its effects on the hydrological cycle, it is more important than ever to understand how the various parts of the process behave. Hydrological models and their parameters are first presented. Starting with the Generalized likelihood uncertainty estimation (GLUE) methodology, a statistical method used in hydrology to quantify the uncertainty of model predictions (Beven and Binley 1992), I will then explain in greater detail how uncertainties and errors in data and hydrological models may be described. After that, different model calibration approaches, i.e.: i) single objective minimisation, ii) the Generalized Likelihood Uncertainty Estimation (GLUE) method, and iii) multi-objective optimisation using the SAFE (Sensitivity Analysis For Everybody) Matlab toolbox (Pianosi, Sarrazin, et al. 2015) are applied to a case study. The applied model is the rainfall-runoff model Hymod (Boyle 2001; Wagener, Boyle, et al. 2001) and the study basin is the Leaf catchment in Mississippi, USA. After addressing the shortcomings of existing approaches, the “limit of acceptability approach” (Liu, Freer, et al. 2009) is utilized to calibrate the hydrological model to improve the calibration results identifying a set of behavioural parameterisations. In the concluding section, a novel criterion for identifying behavioral parameter sets is proposed, and its application results are compared to those of the previously described methods.