scholarly journals MOESHA: A Genetic Algorithm for Automatic Calibration and Estimation of Parameter Uncertainty and Sensitivity of Hydrologic Models

2017 ◽  
Vol 60 (4) ◽  
pp. 1259-1269 ◽  
Author(s):  
Bradley L. Barnhart ◽  
Keith A. Sawicz ◽  
Darren L. Ficklin ◽  
Gerald W. Whittaker
Keyword(s):  
Genetic Algorithm ◽  
Sensitivity Analysis ◽  
Parameter Uncertainty ◽  
Model Uncertainty ◽  
Hydrologic Modeling ◽  
Hydrologic Model ◽  
Sensitivity Analyses ◽  
Model Parameters ◽  
Model Output ◽  
Discharge Data

Abstract. Characterization of the uncertainty and sensitivity of model parameters is an essential facet of hydrologic modeling. This article introduces the multi-objective evolutionary sensitivity handling algorithm (MOESHA) that combines input parameter uncertainty and sensitivity analyses with a genetic algorithm calibration routine to dynamically sample the parameter space. This novel algorithm serves as an alternative to traditional static space-sampling methods, such as stratified sampling or Latin hypercube sampling. In addition to calibrating model parameters to a hydrologic model, MOESHA determines the optimal distribution of model parameters that maximizes model robustness and minimizes error, and the results provide an estimate for model uncertainty due to the uncertainty in model parameters. Subsequently, we compare the model parameter distributions to the distribution of a dummy variable (i.e., a variable that does not affect model output) to differentiate between impactful (i.e., sensitive) and non-impactful parameters. In this way, an optimally calibrated model is produced, and estimations of model uncertainty as well as the relative impact of model parameters on model output (i.e., sensitivity) are determined. A case study using a single-cell hydrologic model (EXP-HYDRO) is used to test the method using river discharge data from the Dee River catchment in Wales. We compare the results of MOESHA with Sobol’s global sensitivity analysis method and demonstrate that the algorithm is able to pinpoint non-impactful parameters, demonstrate the uncertainty of model results with respect to uncertainties in model parameters, and achieve excellent calibration results. A major drawback of the algorithm is that it is computationally expensive; therefore, parallelized methods should be used to reduce the computational burden. Keywords: Genetic algorithm, Hydrologic modeling, Model calibration, Sensitivity analysis, Uncertainty.

scholarly journals Comparing impacts of parameter and spatial data uncertainty for a grid-based distributed watershed model

2016 ◽  
Vol 18 (6) ◽  
pp. 961-974 ◽  
Author(s):  
Younggu Her ◽  
Conrad Heatwole
Keyword(s):  
Land Cover ◽  
Parameter Uncertainty ◽  
Input Data ◽  
Hydrologic Model ◽  
Data Uncertainty ◽  
Model Parameters ◽  
Model Output ◽  
Model Errors ◽  
Land Cover Data ◽  
The Impact

Parameter uncertainty in hydrologic modeling is commonly evaluated, but assessing the impact of spatial input data uncertainty in spatially descriptive ‘distributed’ models is not common. This study compares the significance of uncertainty in spatial input data and model parameters on the output uncertainty of a distributed hydrology and sediment transport model, HYdrology Simulation using Time-ARea method (HYSTAR). The Shuffled Complex Evolution Metropolis (SCEM-UA) algorithm was used to quantify parameter uncertainty of the model. Errors in elevation and land cover layers were simulated using the Sequential Gaussian/Indicator Simulation (SGS/SIS) techniques and then incorporated into the model to evaluate their impact on the outputs relative to those of the parameter uncertainty. This study demonstrated that parameter uncertainty had a greater impact on model output than did errors in the spatial input data. In addition, errors in elevation data had a greater impact on model output than did errors in land cover data. Thus, for the HYSTAR distributed hydrologic model, accuracy and reliability can be improved more effectively by refining parameters rather than further improving the accuracy of spatial input data and by emphasizing the topographic data over the land cover data.

scholarly journals Calibration and evaluation of a semi-distributed watershed model of Sub-Saharan Africa using GRACE data

2012 ◽  
Vol 16 (9) ◽  
pp. 3083-3099 ◽  
Author(s):  
H. Xie ◽  
L. Longuevergne ◽  
C. Ringler ◽  
B. R. Scanlon
Keyword(s):  
Regional Scale ◽  
Water Storage ◽  
Hydrologic Model ◽  
Sub Saharan Africa ◽  
Model Parameters ◽  
Total Water ◽  
Discharge Data ◽  
Model Based ◽  
Grace Data ◽  
Sub Saharan

Abstract. Irrigation development is rapidly expanding in mostly rainfed Sub-Saharan Africa. This expansion underscores the need for a more comprehensive understanding of water resources beyond surface water. Gravity Recovery and Climate Experiment (GRACE) satellites provide valuable information on spatio-temporal variability in water storage. The objective of this study was to calibrate and evaluate a semi-distributed regional-scale hydrologic model based on the Soil and Water Assessment Tool (SWAT) code for basins in Sub-Saharan Africa using seven-year (July 2002–April 2009) 10-day GRACE data and multi-site river discharge data. The analysis was conducted in a multi-criteria framework. In spite of the uncertainty arising from the tradeoff in optimising model parameters with respect to two non-commensurable criteria defined for two fluxes, SWAT was found to perform well in simulating total water storage variability in most areas of Sub-Saharan Africa, which have semi-arid and sub-humid climates, and that among various water storages represented in SWAT, water storage variations in soil, vadose zone and groundwater are dominant. The study also showed that the simulated total water storage variations tend to have less agreement with GRACE data in arid and equatorial humid regions, and model-based partitioning of total water storage variations into different water storage compartments may be highly uncertain. Thus, future work will be needed for model enhancement in these areas with inferior model fit and for uncertainty reduction in component-wise estimation of water storage variations.

scholarly journals Local sensitivity analysis for compositional data with application to soil texture in hydrologic modelling

2013 ◽  
Vol 17 (2) ◽  
pp. 461-478 ◽  
Author(s):  
L. Loosvelt ◽  
H. Vernieuwe ◽  
V. R. N. Pauwels ◽  
B. De Baets ◽  
N. E. C. Verhoest
Keyword(s):  
Sensitivity Analysis ◽  
Soil Texture ◽  
Compositional Data ◽  
High Sensitivity ◽  
Hydrologic Model ◽  
Model Output ◽  
Local Sensitivity ◽  
Hydrologic Modelling ◽  
Compositional Model ◽  
Additional Data Collection

Abstract. Compositional data, such as soil texture, are hard to deal with in the geosciences as standard statistical methods are often inappropriate to analyse this type of data. Especially in sensitivity analysis, the closed character of the data is often ignored. To that end, we developed a method to assess the local sensitivity of a model output with resect to a compositional model input. We adapted the finite difference technique such that the different parts of the input are perturbed simultaneously while the closed character of the data is preserved. This method was applied to a hydrologic model and the sensitivity of the simulated soil moisture content to local changes in soil texture was assessed. Based on a high number of model runs, in which the soil texture was varied across the entire texture triangle, we identified zones of high sensitivity in the texture triangle. In such zones, the model output uncertainty induced by the discrepancy between the scale of measurement and the scale of model application, is advised to be reduced through additional data collection. Furthermore, the sensitivity analysis provided more insight into the hydrologic model behaviour as it revealed how the model sensitivity is related to the shape of the soil moistureretention curve.

scholarly journals Spatial Pattern Oriented Multi-Criteria Sensitivity Analysis of a Distributed Hydrologic Model

2018 ◽  
Author(s):  
Mehmet Cüneyd Demirel ◽  
Julian Koch ◽  
Gorka Mendiguren ◽  
Simon Stisen
Keyword(s):  
Sensitivity Analysis ◽  
Spatial Variability ◽  
Spatial Pattern ◽  
Spatial Patterns ◽  
Performance Measures ◽  
Sampling Strategy ◽  
Behavioral Model ◽  
Hydrologic Model ◽  
Actual Evapotranspiration ◽  
Model Parameters

Hydrologic models are conventionally constrained and evaluated using point measurements of streamflow, which represents an aggregated catchment measure. As a consequence of this single objective focus, model parametrization and model parameter sensitivity are typically not reflecting other aspects of catchment behavior. Specifically for distributed models, the spatial pattern aspect is often overlooked. Our paper examines the utility of multiple performance measures in a spatial sensitivity analysis framework to determine the key parameters governing the spatial variability of predicted actual evapotranspiration (AET). Latin hypercube one-at-a-time (LHS-OAT) sampling strategy with multiple initial parameter sets was applied using the mesoscale hydrologic model (mHM) and a total of 17 model parameters were identified as sensitive. The results indicate different parameter sensitivities for different performance measures focusing on temporal hydrograph dynamics and spatial variability of actual evapotranspiration. While spatial patterns were found to be sensitive to vegetation parameters, streamflow dynamics were sensitive to pedo-transfer function (PTF) parameters. Above all, our results show that behavioral model definition based only on streamflow metrics in the generalized likelihood uncertainty estimation (GLUE) type methods require reformulation by incorporating spatial patterns into the definition of threshold values to reveal robust hydrologic behavior in the analysis.

Comprehensive global sensitivity analysis of a repository model using different types of transformations and metamodeling techniques

2021 ◽  
Author(s):  
Sabine M. Spiessl ◽  
Dirk-A. Becker ◽  
Sergei Kucherenko
Keyword(s):  
Sensitivity Analysis ◽  
Global Sensitivity Analysis ◽  
Higher Order ◽  
Model Parameters ◽  
High Dimensional Model Representation ◽  
Model Output ◽  
Global Sensitivity ◽  
Sparse Polynomial ◽  
Sensitivity Indices ◽  
Highly Nonlinear

<p>Due to their highly nonlinear, non-monotonic or even discontinuous behavior, sensitivity analysis of final repository models can be a demanding task. Most of the output of repository models is typically distributed over several orders of magnitude and highly skewed. Many values of a probabilistic investigation are very low or even zero. Although this is desirable in view of repository safety it can distort the evidence of sensitivity analysis. For the safety assessment of the system, the highest values of outputs are mainly essential and if those are only a few, their dependence on specific parameters may appear insignificant. By applying a transformation, different model output values are differently weighed, according to their magnitude, in sensitivity analysis. Probabilistic methods of higher-order sensitivity analysis, applied on appropriately transformed model output values, provide a possibility for more robust identification of relevant parameters and their interactions. This type of sensitivity analysis is typically done by decomposing the total unconditional variance of the model output into partial variances corresponding to different terms in the ANOVA decomposition. From this, sensitivity indices of increasing order can be computed. The key indices used most often are the first-order index (SI1) and the total-order index (SIT). SI1 refers to the individual impact of one parameter on the model and SIT represents the total effect of one parameter on the output in interactions with all other parameters. The second-order sensitivity indices (SI2) describe the interactions between two model parameters.</p><p>In this work global sensitivity analysis has been performed with three different kinds of output transformations (log, shifted and Box-Cox transformation) and two metamodeling approaches, namely the Random-Sampling High Dimensional Model Representation (RS-HDMR) [1] and the Bayesian Sparse PCE (BSPCE) [2] approaches. Both approaches are implemented in the SobolGSA software [3, 4] which was used in this work. We analyzed the time-dependent output with two approaches for sensitivity analysis, i.e., the pointwise and generalized approaches. With the pointwise approach, the output at each time step is analyzed independently. The generalized approach considers averaged output contributions at all previous time steps in the analysis of the current step. Obtained results indicate that robustness can be improved by using appropriate transformations and choice of coefficients for the transformation and the metamodel.</p><p>[1] M. Zuniga, S. Kucherenko, N. Shah (2013). Metamodelling with independent and dependent inputs. Computer Physics Communications, 184 (6): 1570-1580.</p><p>[2] Q. Shao, A. Younes, M. Fahs, T.A. Mara (2017). Bayesian sparse polynomial chaos expansion for global sensitivity analysis. Computer Methods in Applied Mechanics and Engineering, 318: 474-496.</p><p>[3] S. M. Spiessl, S. Kucherenko, D.-A. Becker, O. Zaccheus (2018). Higher-order sensitivity analysis of a final repository model with discontinuous behaviour. Reliability Engineering and System Safety, doi: https://doi.org/10.1016/j.ress.2018.12.004.</p><p>[4] SobolGSA software (2021). User manual https://www.imperial.ac.uk/process-systems-engineering/research/free-software/sobolgsa-software/.</p>

scholarly journals Global sensitivity analysis of parameter uncertainty in landscape evolution models

2018 ◽  
Vol 11 (12) ◽  
pp. 4873-4888 ◽  
Author(s):  
Christopher J. Skinner ◽  
Tom J. Coulthard ◽  
Wolfgang Schwanghart ◽  
Marco J. Van De Wiel ◽  
Greg Hancock
Keyword(s):  
Sensitivity Analysis ◽  
Objective Function ◽  
Landscape Evolution ◽  
Relative Sensitivity ◽  
Function Approach ◽  
Sensitivity Analyses ◽  
Model Parameters ◽  
Model Function ◽  
Sediment Yields ◽  
Statistical Measures

Abstract. The evaluation and verification of landscape evolution models (LEMs) has long been limited by a lack of suitable observational data and statistical measures which can fully capture the complexity of landscape changes. This lack of data limits the use of objective function based evaluation prolific in other modelling fields, and restricts the application of sensitivity analyses in the models and the consequent assessment of model uncertainties. To overcome this deficiency, a novel model function approach has been developed, with each model function representing an aspect of model behaviour, which allows for the application of sensitivity analyses. The model function approach is used to assess the relative sensitivity of the CAESAR-Lisflood LEM to a set of model parameters by applying the Morris method sensitivity analysis for two contrasting catchments. The test revealed that the model was most sensitive to the choice of the sediment transport formula for both catchments, and that each parameter influenced model behaviours differently, with model functions relating to internal geomorphic changes responding in a different way to those relating to the sediment yields from the catchment outlet. The model functions proved useful for providing a way of evaluating the sensitivity of LEMs in the absence of data and methods for an objective function approach.

scholarly journals Aerosol and physical atmosphere model parameters are both important sources of uncertainty in aerosol ERF

2018 ◽  
Vol 18 (13) ◽  
pp. 9975-10006 ◽  
Author(s):  
Leighton A. Regayre ◽  
Jill S. Johnson ◽  
Masaru Yoshioka ◽  
Kirsty J. Pringle ◽  
David M. H. Sexton ◽  
...  
Keyword(s):  
Model Uncertainty ◽  
Radiative Forcing ◽  
Computational Cost ◽  
Short Wave ◽  
Sensitivity Analyses ◽  
Model Parameters ◽  
Large Set ◽  
Individual Parameter ◽  
Cloud Water Content ◽  
Atmosphere Model

Abstract. Changes in aerosols cause a change in net top-of-the-atmosphere (ToA) short-wave and long-wave radiative fluxes; rapid adjustments in clouds, water vapour and temperature; and an effective radiative forcing (ERF) of the planetary energy budget. The diverse sources of model uncertainty and the computational cost of running climate models make it difficult to isolate the main causes of aerosol ERF uncertainty and to understand how observations can be used to constrain it. We explore the aerosol ERF uncertainty by using fast model emulators to generate a very large set of aerosol–climate model variants that span the model uncertainty due to 27 parameters related to atmospheric and aerosol processes. Sensitivity analyses shows that the uncertainty in the ToA flux is dominated (around 80 %) by uncertainties in the physical atmosphere model, particularly parameters that affect cloud reflectivity. However, uncertainty in the change in ToA flux caused by aerosol emissions over the industrial period (the aerosol ERF) is controlled by a combination of uncertainties in aerosol (around 60 %) and physical atmosphere (around 40 %) parameters. Four atmospheric and aerosol parameters account for around 80 % of the uncertainty in short-wave ToA flux (mostly parameters that directly scale cloud reflectivity, cloud water content or cloud droplet concentrations), and these parameters also account for around 60 % of the aerosol ERF uncertainty. The common causes of uncertainty mean that constraining the modelled planetary brightness to tightly match satellite observations changes the lower 95 % credible aerosol ERF value from −2.65 to −2.37 W m−2. This suggests the strongest forcings (below around −2.4 W m−2) are inconsistent with observations. These results show that, regardless of the fact that the ToA flux is 2 orders of magnitude larger than the aerosol ERF, the observed flux can constrain the uncertainty in ERF because their values are connected by constrainable process parameters. The key to reducing the aerosol ERF uncertainty further will be to identify observations that can additionally constrain individual parameter ranges and/or combined parameter effects, which can be achieved through sensitivity analysis of perturbed parameter ensembles.

Sensitivity analysis of an integrated numerical flow model output to key parameters used in common qualitative vulnerability assessment methods

2020 ◽  
Author(s):  
Joanna Doummar ◽  
Assaad H. Kassem
Keyword(s):  
Sensitivity Analysis ◽  
Vulnerability Assessment ◽  
Groundwater Vulnerability ◽  
Assessment Methods ◽  
Qualitative Assessment ◽  
Model Parameters ◽  
Karst Aquifers ◽  
Model Output ◽  
Local Sensitivity ◽  
The Impact

<p>Qualitative vulnerability assessment methods applied in karst aquifers rely on key factors in the hydrological compartments usually assigned different weights according to their estimated impact on groundwater vulnerability. Based on an integrated numerical groundwater model on a snow-governed karst catchment area (Assal Spring- Lebanon), the aim of this work is to quantify the importance of the most influential parameters on recharge and spring discharge and outline potential parameters that are not accounted for in standard methods, when in fact they do play a role in the intrinsic vulnerability of a system. The assessment of the model sensitivity and the ranking of parameters are conducted using an automatic calibration tool for local sensitivity analysis in addition to a variance-based local sensitivity assessment of model output time series (recharge and discharge)  for two consecutive years (2016-2017) to various model parameters. The impact of each parameter was normalized to estimate standardized weights for each of the process based key-controlling parameters. Parameters to which model was sensitive were factors related to soil, 2) fast infiltration (bypass function) typical of karst aquifers, 3) climatic parameters (melting temperature and degree day coefficient) and 4) aquifer hydraulic properties that play a major role in groundwater vulnerability inducing a temporal effect and varied recession. Other less important parameters play different roles according to different assigned weights proportional to their ranking. Additionally, the effect of slope/geomorphology (e.g., dolines) was further investigated.  In general, this study shows that the weighting coefficients assigned to key vulnerability factors in the qualitative assessment methods can be reevaluated based on this process-based approach.</p><p> </p><p> </p><p> </p>

Sensitivity analysis and the challenges posed by multiple approaches: a multifaceted mess

2020 ◽  
Author(s):  
Monica Riva ◽  
Aronne Dell'Oca ◽  
Alberto Guadagnini
Keyword(s):  
Sensitivity Analysis ◽  
Model Calibration ◽  
Global Sensitivity Analysis ◽  
System Model ◽  
Model Parameters ◽  
Model Output ◽  
Multiple Model ◽  
Total Uncertainty ◽  
Global Sensitivity ◽  
Sensitivity Indices

<p>Modern models of environmental and industrial systems have reached a relatively high level of complexity. The latter aspect could hamper an unambiguous understanding of the functioning of a model, i.e., how it drives relationships and dependencies among inputs and outputs of interest. Sensitivity Analysis tools can be employed to examine this issue.</p><p>Global sensitivity analysis (GSA) approaches rest on the evaluation of sensitivity across the entire support within which system model parameters are supposed to vary. In this broad context, it is important to note that the definition of a sensitivity metric must be linked to the nature of the question(s) the GSA is meant to address. These include, for example: (i) which are the most important model parameters with respect to given model output(s)?; (ii) could we set some parameter(s) (thus assisting model calibration) at prescribed value(s) without significantly affecting model results?; (iii) at which space/time locations can one expect the highest sensitivity of model output(s) to model parameters and/or knowledge of which parameter(s) could be most beneficial for model calibration?</p><p>The variance-based Sobol’ Indices (e.g., Sobol, 2001) represent one of the most widespread GSA metrics, quantifying the average reduction in the variance of a model output stemming from knowledge of the input. Amongst other techniques, Dell’Oca et al. [2017] proposed a moment-based GSA approach which enables one to quantify the influence of uncertain model parameters on the (statistical) moments of a target model output.</p><p>Here, we embed in these sensitivity indices the effect of uncertainties both in the system model conceptualization and in the ensuing model(s) parameters. The study is grounded on the observation that physical processes and natural systems within which they take place are complex, rendering target state variables amenable to multiple interpretations and mathematical descriptions. As such, predictions and uncertainty analyses based on a single model formulation can result in statistical bias and possible misrepresentation of the total uncertainty, thus justifying the assessment of multiple model system conceptualizations. We then introduce copula-based sensitivity metrics which allow characterizing the global (with respect to the input) value of the sensitivity and the degree of variability (across the whole range of the input values) of the sensitivity for each value that the prescribed model output can possibly undertake, as driven by a governing model. In this sense, such an approach to sensitivity is global with respect to model input(s) and local with respect to model output, thus enabling one to discriminate the relevance of an input across the entire range of values of the modeling goal of interest. The methodology is demonstrated in the context of flow and reactive transport scenarios.</p><p> </p><p><strong>References</strong></p><p>Sobol, I. M., 2001. Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates. Math. Comput. Sim., 55, 271-280.</p><p>Dell’Oca, A., Riva, M., Guadagnini, A., 2017. Moment-based metrics for global sensitivity analysis of hydrological systems. Hydr. Earth Syst. Sci., 21, 6219-6234.</p>

scholarly journals Emulation of a complex global aerosol model to quantify sensitivity to uncertain parameters

2011 ◽  
Vol 11 (7) ◽  
pp. 20433-20485 ◽  
Author(s):  
L. A. Lee ◽  
K. S. Carslaw ◽  
K. Pringle ◽  
G. W. Mann ◽  
D. V. Spracklen
Keyword(s):  
Sensitivity Analysis ◽  
Gaussian Process ◽  
Interaction Effects ◽  
Model Parameters ◽  
Uncertain Parameters ◽  
Model Output ◽  
Aerosol Model ◽  
Uncertainty Space ◽  
Gaussian Process Emulation ◽  
Gaussian Process Emulator

Abstract. Sensitivity analysis of atmospheric models is necessary to identify the processes that lead to uncertainty in model predictions, to help understand model diversity, and to prioritise research. Assessing the effect of parameter uncertainty in complex models is challenging and often limited by CPU constraints. Here we present a cost-effective application of variance-based sensitivity analysis to quantify the sensitivity of a 3-D global aerosol model to uncertain parameters. A Gaussian process emulator is used to estimate the model output across multi-dimensional parameter space using information from a small number of model runs at points chosen using a Latin hypercube space-filling design. Gaussian process emulation is a Bayesian approach that uses information from the model runs along with some prior assumptions about the model behaviour to predict model output everywhere in the uncertainty space. We use the Gaussian process emulator to calculate the percentage of expected output variance explained by uncertainty in global aerosol model parameters and their interactions. To demonstrate the technique, we show examples of cloud condensation nuclei (CCN) sensitivity to 8 model parameters in polluted and remote marine environments as a function of altitude. In the polluted environment 95 % of the variance of CCN concentration is described by uncertainty in the 8 parameters (excluding their interaction effects) and is dominated by the uncertainty in the sulphur emissions, which explains 80 % of the variance. However, in the remote region parameter interaction effects become important, accounting for up to 40 % of the total variance. Some parameters are shown to have a negligible individual effect but a substantial interaction effect. Such sensitivities would not be detected in the commonly used single parameter perturbation experiments, which would therefore underpredict total uncertainty. Gaussian process emulation is shown to be an efficient and useful technique for quantifying parameter sensitivity in complex global atmospheric model.

Sign in / Sign up

Export Citation Format

Share Document