Addressing subjective decision-making inherent in GLUE-based multi-criteria rainfall–runoff model calibration

Shafii, Mahyar; Tolson, Bryan A.; Matott, L. Shawn

doi:10.1016/j.jhydrol.2015.01.051

Cited by 29 publications

(31 citation statements)

References 88 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The pseudo‐likelihood function of KGE is used to assess model performance. The less subjective selection criteria are a common practice in the literature; thus, we use a behavioural parameter set, which subjectively meets the desired performance criteria (Li & Xu, ; Shafii, Tolson, & Matott, ; Stedinger, Vogel, Lee, & Batchelder, ). These methods fail to account for output uncertainty; therefore, we use a simple method of selecting the top 10% of the model simulations.…”

Section: Methodsmentioning

confidence: 99%

Sensitivity analysis and uncertainty assessment in water budgets simulated by the variable infiltration capacity model for Canadian subarctic watersheds

Lilhare

Pokorny

Déry

et al. 2020

Hydrological Processes

View full text Add to dashboard Cite

In this study, we evaluate uncertainties propagated through different climate data sets in seasonal and annual hydrological simulations over 10 subarctic watersheds of northern Manitoba, Canada, using the variable infiltration capacity (VIC) model. Further, we perform a comprehensive sensitivity and uncertainty analysis of the VIC model using a robust and state-of-the-art approach. The VIC model simulations utilize the recently developed variogram analysis of response surfaces (VARS) technique that requires in this application more than 6,000 model simulations for a 30-year (1981-2010) study period. The method seeks parameter sensitivity, identifies influential parameters, and showcases streamflow sensitivity to parameter uncertainty at seasonal and annual timescales. Results suggest that the Ensemble VIC simulations match observed streamflow closest, whereas global reanalysis products yield high flows (0.5-3.0 mm day −1 ) against observations and an overestimation (10-60%) in seasonal and annual water balance terms. VIC parameters exhibit seasonal importance in VARS, and the choice of input data and performance metrics substantially affect sensitivity analysis. Uncertainty propagation due to input forcing selection in each water balance term (i.e., total runoff, soil moisture, and evapotranspiration) is examined separately to show both time and space dimensionality in available forcing data at seasonal and annual timescales. Reliable input forcing, the most influential model parameters, and the uncertainty envelope in streamflow prediction are presented for the VIC model. These results, along with some specific recommendations, are expected to assist the broader VIC modelling community and other users of VARS and land surface schemes, to enhance their modelling applications. K E Y W O R D S hydrological modelling, lower Nelson River basin, sensitivity analysis, uncertainty assessment, VARS, VIC model, VIC parameters, water balance 1 | INTRODUCTION Numerical modelling of a river basin remains essential for both climate and ecological studies as it provides vital information about the hydrological cycle and water availability for societies and ecosystems. Although recent developments and advances have been achieved in hydrological modelling and computational power, addressing efficiently the uncertainties in hydrological simulation remains a critical

show abstract

Section: Methodsmentioning

confidence: 99%

Sensitivity analysis and uncertainty assessment in water budgets simulated by the variable infiltration capacity model for Canadian subarctic watersheds

Lilhare

Pokorny

Déry

et al. 2020

Hydrological Processes

View full text Add to dashboard Cite

show abstract

“…The Kling–Gupta efficiency (Gupta, Kling, Yilmaz, & Martinez, ; I KGE ) was used as performance metric for the individual objectives:

I_{KGE} = 1 - \sqrt{{(r - 1)}^{2} + {(α - 1)}^{2} + {(β - 1)}^{2}},

where r is the linear correlation coefficient between simulation and observation; α (α = σ m /σ o ) is a measure of relative variability in the simulated and observed values, where σ m is the standard deviation of simulated variables and σ o is the standard deviation of observed variables; β is the ratio between the average value of simulated and observed variables. The 1% (5,000) best performing parameter sets were retained as behavioural and used to establish feasible posterior distributions and to construct model uncertainty ranges, using I KGE_HGS as informal likelihood measure (GLUE; Beven & Binley, ; Freer, McMillan, McDonnell, & Beven, ; Shafii, Tolson, & Shawn, ; Zhang, Li, Guo, & Gong, ).…”

Section: Model Evaluation Calibration Evaluation and Transferabilitymentioning

confidence: 99%

The importance of aspect for modelling the hydrological response in a glacier catchment in Central Asia

Gao

Ding

Zhao

et al. 2017

Hydrological Processes

View full text Add to dashboard Cite

Understanding how explicit consideration of topographic information influences hydrological model performance and upscaling in glacier dominated catchments remains underexplored. In this study, the Urumqi glacier no. 1 catchment in northwest China, with 52% of the area covered by glaciers, was selected as study site. A conceptual glacier‐hydrological model was developed and tested to systematically, simultaneously, and robustly reproduce the hydrograph, separate the discharge into contributions from glacier and nonglacier parts of the catchment, and establish estimates of the annual glacier mass balance, the annual equilibrium line altitude, and the daily catchment snow water equivalent. This was done by extending and adapting a recently proposed landscape‐based semidistributed conceptual hydrological model (FLEX‐Topo) to represent glacier and snowmelt processes. The adapted model, FLEXG, allows to explicitly account for the influence of topography, that is, elevation and aspect, on the distribution of temperature and precipitation and thus on melt dynamics. It is shown that the model can not only reproduce long‐term runoff observations but also variations in glacier and snow cover. Furthermore, FLEXG was successfully transferred and up‐scaled to a larger catchment exclusively by adjusting the areal proportions of elevation and aspect without the need for further calibration. This underlines the value of topographic information to meaningfully represent the dominant hydrological processes in the region and is further exacerbated by comparing the model to a model formulation that does not account for differences in aspect (FLEXG,nA) and which, in spite of satisfactorily reproducing the observed hydrograph, does not capture the influence of spatial variability of snow and ice, which as a consequence reduces model transferability. This highlights the importance of accounting for topography and landscape heterogeneity in conceptual hydrological models in mountainous and snow‐, and glacier‐dominated regions.

show abstract

“…The Nash-Sutcliffe efficiency coefficient used in this study is known to be sensitive to large values (Krause et al, 2005); thus it weighs heavy on peaks rather than recession and baseflow components of hydrographs. Despite the expected bias in model performance evaluation, the Nash-Sutcliffe efficiency coefficient was chosen as a single performance statistics in this study because of its proven applicability as an informal likelihood function (Setegn et al, 2010;Xie and Lian, 2013;Shafii et al, 2015). In addition, it has been commonly used as an objective function in parameter calibration and/or goodness-of-fit statistics in evaluation of hydrologic models (Krause et al, 2005;Engel et al, 2007;Moriasi et al, 2007); thus, use of the coefficient was expected to make the results of this study be more easily interpreted by other researchers and transferred to other modelling studies.…”

Section: Generalized Likelihood Uncertainty Estimation Framework and mentioning

confidence: 99%

Impact of the numbers of observations and calibration parameters on equifinality, model performance, and output and parameter uncertainty

2015

View full text Add to dashboard Cite

The level of model complexity that can be effectively supported by available information has long been a subject of many studies in hydrologic modelling. In particular, distributed parameter models tend to be regarded as overparameterized because of numerous parameters used to describe spatially heterogeneous hydrologic processes. However, it is not clear how parameters and observations influence the degree of overparameterization, equifinality of parameter values, and uncertainty. This study investigated the impact of the numbers of observations and parameters on calibration quality including equifinality among calibrated parameter values, model performance, and output/parameter uncertainty using the Soil and Water Assessment Tool model. In the experiments, the number of observations was increased by expanding the calibration period or by including measurements made at inner points of a watershed. Similarly, additional calibration parameters were included in the order of their sensitivity. Then, unique sets of parameters were calibrated with the same objective function, optimization algorithm, and stopping criteria but different numbers of observations. The calibration quality was quantified with statistics calculated based on the ‘behavioural’ parameter sets, identified using 1% and 5% cut‐off thresholds in a generalized likelihood uncertainty estimation framework. The study demonstrated that equifinality, model performance, and output/parameter uncertainty were responsive to the numbers of observations and calibration parameters; however, the relationship between the numbers, equifinality, and uncertainty was not always conclusive. Model performance improved with increased numbers of calibration parameters and observations, and substantial equifinality did neither necessarily mean bad model performance nor large uncertainty in the model outputs and parameters. Copyright © 2015 John Wiley & Sons, Ltd.

show abstract

Addressing subjective decision-making inherent in GLUE-based multi-criteria rainfall–runoff model calibration

Cited by 29 publications

References 88 publications

Sensitivity analysis and uncertainty assessment in water budgets simulated by the variable infiltration capacity model for Canadian subarctic watersheds

Sensitivity analysis and uncertainty assessment in water budgets simulated by the variable infiltration capacity model for Canadian subarctic watersheds

The importance of aspect for modelling the hydrological response in a glacier catchment in Central Asia

Impact of the numbers of observations and calibration parameters on equifinality, model performance, and output and parameter uncertainty

Contact Info

Product

Resources

About