Hydrological model calibration with uncertain discharge data

ABSTRACT

Discharge data used to calibrate and evaluate hydrological models can be highly uncertain and this uncertainty affects the conclusions that we can draw from modelling results. We investigated the role of discharge data uncertainty and its representation in hydrological model calibration to give recommendations on methods to account for data uncertainty. We tested five different representations of discharge data uncertainty in calibrating the HBV-model for three Swiss catchments, ranging from using no information to using full empirical probability distributions for each time step. We developed a new objective function to include discharge data uncertainty, as quantified by these distributions directly in calibration to hydrological time series. This new objective function provided more reliable results than using no data uncertainty or multiple realizations of discharge time series. We recommend using the new objective function in combination with empirical or triangular distributions of the discharge data uncertainty.

KEYWORDS:

• discharge data uncertainty
• model calibration
• rating curve uncertainty
• objective function
• HBV model
• model evaluation

Editor A. Castellarin Guest Editor S. Uhlenbrook

Editor A. Castellarin Guest Editor S. Uhlenbrook

Acknowledgements

The ScienceCloud computational and storage infrastructure provided by S3IT at the University of Zurich enabled the computation-intensive simulations. We thank the Swiss Federal Office for the Environment (FOEN) and MeteoSwiss for providing hydrological and meteorological data making this study possible. Special thanks to Hanspeter Hodel at FOEN for comprehensive personal communications on rating curves and site characteristics. We thank Tanja de Boer-Euser and an anonymous reviewer for constructive comments that helped to clarify and improve the paper. The new objective function that incorporates different discharge data distributions is implemented in the HBV-light model software, available at https://www.geo.uzh.ch/en/units/h2k/Services/HBV-Model.html

Disclosure statement

The authors declare that there is no conflict of interest regarding the publication of this article.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This research has been conducted as a part of the EXAR project (Hazard information for extreme flood events on the rivers Aare and Rhine) funded by the Swiss Federal Office for the Environment (FOEN), the Swiss Federal Nuclear Safety Inspectorate (ENSI), the Swiss Federal Office of Energy (SFOE), the Swiss Federal Office for Civil Protection (FOCP) and the Swiss Federal Office of Meteorology and Climatology (MeteoSwiss) under Project No. 15.0054.PJ/O503-1381. IW’s time was partly funded by Svenska Forskningsrådet Formas (2018-02900, ERA-NET).