Leveraging ensemble meteorological forcing data to improve parameter estimation of hydrologic models

Abstract

As continental to global scale high‐resolution meteorological datasets continue to be developed, there are sufficient meteorological datasets available now for modellers to construct a historical forcing ensemble. The forcing ensemble can be a collection of multiple deterministic meteorological datasets or come from an ensemble meteorological dataset. In hydrological model calibration, the forcing ensemble can be used to represent forcing data uncertainty. This study examines the potential of using the forcing ensemble to identify more robust parameters through model calibration. Specifically, we compare an ensemble forcing‐based calibration with two deterministic forcing‐based calibrations and investigate their flow simulation and parameter estimation properties and the ability to resist poor‐quality forcings. The comparison experiment is conducted with a six‐parameter hydrological model for 30 synthetic studies and 20 real data studies to provide a better assessment of the average performance of the deterministic and ensemble forcing‐based calibrations. Results show that the ensemble forcing‐based calibration generates parameter estimates that are less biased and have higher frequency of covering the true parameter values than the deterministic forcing‐based calibration does. Using a forcing ensemble in model calibration reduces the risk of inaccurate flow simulation caused by poor‐quality meteorological inputs, and improves the reliability and overall simulation skill of ensemble simulation results. The poor‐quality meteorological inputs can be effectively filtered out via our ensemble forcing‐based calibration methodology and thus discarded in any post‐calibration model applications. The proposed ensemble forcing‐based calibration method can be considered as a more generalized framework to include parameter and forcing uncertainties in model calibration.

Impacts of observational uncertainty on analysis and modelling of hydrological processes: Preface

“…A correct design of the modelling study should consider a proper selection of the input (precipitation) products for hydrological models because the model parameters and modelling uncertainty will largely depend on this forcing product. Based on papers from this issue, it could be recommended to either test different products to choose the most appropriate one for the purpose of the study (Culler et al, 2021), or to use an ensemble of input forcings instead of using only a single product (Liu et al, 2021).…”

“…Three of these papers focused on multiple output datasets, that is, using more than one output variable, with total suspended solid data in addition to streamflow (Wu et al, 2021) or stable isotopes together with streamflow data (Stevenson et al, 2021). One paper recommends using multiple input precipitation products to account for input uncertainty and to better constrain parameter uncertainty of hydrological models (Liu et al, 2021). Nevertheless, assessing the effect of using multiple datasets on model identification and process understanding remains an ongoing research avenue, specifically as new datasets, new techniques for data collection, or new methods for data uncertainty assessment become more available.…”

“…Testing new (hydrological) techniques at non‐optimal locations enables us to test the limits of such techniques and apply them in a broader range of locations, while accounting for uncertainty. Another example is given by Liu et al (2021) who propose using an ensemble of input forcings, instead of only a single input product, to address input uncertainty in model simulations and to better constrain model parameters.…”

“…In our special issue, all three papers dealing with input uncertainty similarly focus largely on precipitation uncertainty. The papers investigate input (precipitation) uncertainty and its impact on parameter identification in hydrological models (Liu et al, 2021), input uncertainty in observed and generated variables for hydrological models and process representation (Beven, 2021b), and precipitation uncertainty impact on the rainfall‐triggered landslide events (Culler et al, 2021). Beven (2021b) also discusses uncertainties in other input variables, that is, temperature and evapotranspiration.…”

“…In detail, Liu et al (2021) investigate the effect of using a forcing ensemble of precipitation time series, represented as an ensemble meteorological dataset or a collection of multiple deterministic meteorological datasets, on the ability to determine robust parameters of a hydrological model. Based on 30 synthetic datasets and 20 real case studies, they find that using an ensemble of forcing inputs is beneficial over a single deterministic input because it improves the overall simulation skills of ensemble‐based flow simulations and reduces the potential effect a poor‐quality input data can have on model calibration.…”