Search NASA⌕ Search

SEARCH · Search NASA

Results for “Surrogates”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 145 records · Page 8

ADAPTIVE SURROGATES WITHIN THE RAVEN FRAMEWORK FOR DYNAMIC PROBABILISTIC RISK ASSESSMENT ANALYSIS

The RAVEN code has been under development at the Idaho National Laboratory since 2012. Its main goal is to create a multi-purpose platform for the deploying of all the capabilities needed for Probabilistic Risk Assessment (PRA), Uncertainty Quantification (UQ), data mining analysis and optimization studies. RAVEN has demonstrated a good level of maturity in terms of state-of-art and advanced analysis methodologies. In the past year the RAVEN code has been released as open-source project and is available to download (raven.inl.gov) free of charge. The main subject of this paper is to present a new system within the RAVEN framework to drastically accelerate PRA and UQ analysis using acceleration schemes based on surrogate modeling. This system is based on the heavy usage of surrogate modeling and artificial intelligence algorithms, able to emulate the response of a physical system in closed domains (domains where the surrogate models have been constructed from). The aim of this paper is to show to the community how the employment of surrogate modeling techniques can be a viable solution for the current and future challenges in terms of PRA, UQ and Safety Margins characterizations. In addition, in this manuscript it is presented the construction of a RAVEN new entity named “HybridModel” that is aimed to switch between a high-fidelity code and its linked surrogate, accelerating the PRA and UQ processes without largely reducing the accuracy of the results of the analyses. All these new capabilities will be shown with an application example about a Station Black Out scenario for a prototypical Pressurized Water Reactor (modeled with RELAP5-3D).

97 MATHEMATICS AND COMPUTING↗

Decay Heat Surrogate Modeling for High Temperature Reactors

The accuracy of the safety analysis of Nuclear Power Plants, in general, and the High Temperature Reactors, in particular, is driven by several factors, among which the accuracy of decay heat curves or calculation. This paper is aimed to present a methodology based on the usage of the RAVEN code for the construction of a multi-dimensional decay heat curve for High Temperature Reactors, via calculations performed by ORIGEN. The main goal of this activity has been the identification of a model able to surrogate the phenomenology of the decay heat after shutdown, satisfying the following requirements: 1) Reasonable prediction accuracy till 1 month after the shutdown, 2) Ability to capture the main deviation effects determined by field conditions (e.g. burn-up, fuel temperature, fissile enrichment, etc.), 3) As easy as possible formalism in order to possibly allow the implementation of the curve into system codes (e.g. RELAP5-3D) inputs exploiting, for example, control variable capabilities. Three surrogate models (also known as decay heat curves in the safety analysis nomenclature) have been identified: 1) Spline Exponential surrogate model, 2) A Support Vector Regression model and, 3) A Dynamic Mode Decomposition model. The aim of this paper is to present the mathematical background of the identified surrogate models and their performance in the prediction of decay heat evolution for High Temperature Reactors. A preliminary comparison of the surrogate model performances is finally reported.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Karhunen–Loève deep learning method for surrogate modeling and approximate Bayesian parameter estimation

We evaluate the performance of the Karhunen-Loève Deep Neural Network (KL-DNN) framework for surrogate modeling and approximate Bayesian parameter estimation in partial differential equation models. In the surrogate model, the Karhunen-Loève (KL) expansions are used for the dimensionality reduction of the number of unknown parameters and variables, and a deep neural network is employed to relate the reduced space of parameters to that of the state variables. The KL-DNN surrogate model is used to formulate a maximum-a-posteriori-like least-squares problem, which is randomized to draw samples of the posterior distribution of the parameters. We test the proposed framework for a hypothetical unconfined aquifer via comparison with the forward MODFLOW and inverse PEST++ iterative ensemble smoother (IES) solutions as well as the state-of-the-art Fourier neural operator (FNO) and deep operator networks (DeepONets) operator learning surrogate models. Our results show that the KL-DNN surrogate model outperforms FNO and DeepONet for forward predictions. For solving inverse problems, the randomized algorithm provides the same or more accurate Bayesian predictions of the parameters than IES as evidenced by the higher log-predictive probability of both the estimated parameter field and the forecast hydraulic head. The posterior mean obtained from the randomized algorithm is closer to the reference parameter field than that obtained with FNO as the maximum a posteriori estimate.

Approximate Bayesian inference↗

Comparative analysis of temperature dependent properties of commercial nuclear fuel pellet and surrogates undergoing cracking: A review

A variety of normal and accident scenarios can generate thermal stresses enough to cause cracking in light water reactor (LWR) fuel. To better understand this behavior, cracking experiments have been carried out to induce thermal gradients in fuel pellets via induction heating and direct resistance heating. In this study, Ceria (CeO 2 ) and yttria stabilized zirconia (YSZ) pellets were chosen as a surrogate fuel material for uranium dioxide (UO 2 ). Induction heating was done using copper coils and molybdenum susceptors which heated the surrogates to a threshold temperature that is sufficiently high for the fuel material to conduct current. Thereafter, direct resistance heating was achieved by passing current through the specimen using a DC power supply to introduce volumetric heating to replicate reactor operating conditions. Simultaneous real-time dual imaging of the ceria pellet surface has been designed using optical and infra-red camera system to capture images of cracks and full-field temperature gradients on pellet. It was observed that YSZ pellets need a much higher voltage-current for volumetric heating and do not exhibit the cracking pattern as ceria and UO 2 , hence YSZ was ruled out for further consideration. Ceria was found to be an appropriate surrogate for UO 2 . Furthermore, the experimental set up and test conditions demand the need to primarily understand the thermo-physical, mechanical and optical properties of UO 2 and the surrogates. Thus, it is essential to have an in-depth knowledge about the various temperature dependent properties of UO 2 and CeO 2 for establishing a comparative analysis between the surrogate and UO 2 for conducting experiments.

36 MATERIALS SCIENCE↗

Experimental and computational study of methane counterflow diffusion flames perturbed by trace amounts of either jet fuel or a 6-component surrogate under non-sooting conditions

The chemical structure of a methane counterflow diffusion flame and of the same flame doped with 1000 ppm (molar) of either jet fuel or a 6-component jet fuel surrogate was analyzed experimentally, by gas sampling via quartz microprobes and subsequent GC/MS analysis, and computationally using a semi-detailed kinetic mechanism for the surrogate blend. Conditions were chosen to ensure that all three flames were non-sooting, with identical temperature profiles and stoichiometric mixture fraction, through a judicious selection of feed stream composition and strain rate. The experimental dataset provides a glimpse of the pyrolysis and oxidation behavior of jet fuel in a diffusion flame. The jet fuel initial oxidation is consistent with anticipated chemical kinetic behavior, based on thermal decomposition of large alkanes to smaller and smaller fragments and the survival of ring-stabilized aromatics at higher temperatures. The 6-component surrogate captures the same trend correctly, but the agreement is not quantitative with respect to some of the aromatics such as benzene and toluene. Various alkanes, alkenes and aromatics among the jet fuel components are either only qualitatively characterized or could not be identified, because of the presence of many isomers and overlapping spectra in the chromatogram, leaving 80% of the carbon from the jet fuel unaccounted for in the early pyrolysis history of the parent fuel. Computationally, the one-dimensional code adopted a semi-detailed kinetic mechanism for the surrogate blend that is based on an existing hierarchically constructed kinetic model for alkanes and simple aromatics, extended to account for the presence of tetralin and methylcyclohexane as reference fuels. The computational results are in reasonably good agreement with the experimental ones for the surrogate behavior, with the greatest discrepancy in the concentrations of aromatics and ethylene. (author)

33 ADVANCED PROPULSION SYSTEMS↗

Experimental and fuel-surrogates modeling study of the high-pressure pyrolysis of specialty cetane number fuels: implications for fall-off in ethylene unimolecular dissociation

Single pulse shock tube experiments were conducted at 50 atm nominal pressure and 4 ms nominal reaction time over a temperature range of 900–1800 K, to study the pyrolysis speciation of a multi-component jet fuel, F-24, and six cetane number (CN) specialty fuels - CN30, CN35, CN40, CN45, CN50, and CN55. Gas chromatography (GC) was used to qualitatively and quantitatively analyze the post shock gases. The relationship between the formation of key pyrolysis species and the chemically controlled combustion propensity as reflected by the cetane number of each fuel was examined. A surrogate-based mechanism from the CRECK Modelling Group and chemical-functional group based optimized surrogates (CFGO) were used to simulate the pyrolysis speciation results. The model was able to capture the chemistry of most species except two important pyrolysis intermediates – ethylene and acetylene. Chemical kinetic analyses were performed to identify the important reactions which affect the chemistry of these species; however, the rate parameters of critical reactions were found to be unsuitable for simulating the present high-pressure studies. Here, to address this unsuitability, a theory-based fall-off analysis for three reactions representing the decomposition of ethylene and subsequent formation of acetylene was performed, and these are included in an updated version of the CRECK mechanism. This update resolves discrepancies between the experimental results and simulations for ethylene and acetylene. Reaction flux analyses using the updated surrogate model were also performed to identify the important reaction pathways responsible for the formation of crucial species and to provide an analysis of the chemistry of complex multi-component fuel systems. The fundamental reactions responsible for driving pyrolysis chemistry were greatly influenced by the chemical functional groups present in these fuels. In addition to updating the rate parameters of specific reactions to improve modeling, this study also emphasizes the effectiveness of the fuel-surrogate approach, where surrogates representing the chemical functional group composition of the parent fuel serve as a valuable tool for predicting the combustion chemistry of novel fuels.

Chemical Kinetics↗

Machine learning surrogate of physics-based building-stock simulator for end-use load forecasting

Building energy models are used to simulate heat and mass transfer and estimate end-use load in buildings. With the proliferation of solar photovoltaics on residential and commercial buildings, increasingly, buildings are expected to provide grid services, for which accurate and computationally efficient building energy simulations and end-use load prediction are imperative. Existing building energy simulation tools, however, have significant computational overhead that make them less practical in real-time deployment for optimization, design, uncertainty quantification and control in building energy management systems. Here this article presents a data-driven machine learning model based on light gradient boosting method (LightGBM) as a surrogate for a physics-based simulator for residential buildings to predict end-use load. The machine learning based surrogate model accounts for time-series related variables, seasonality and trend component of end-use load, and history of end-use load. The accuracy of the surrogate model is assessed on the prediction of the load profiles of 100 different houses in Cook County, Illinois, USA. The LightGBM surrogate model is shown to reduce the root-mean-squared error by 53% relative to a reference decision tree (DT) based model reported previously in the literature. Moreover, the model predicts the load spikes and high-ramp rate events throughout the year which are often the Achilles heel of other models in the literature. The machine learning based surrogate model is demonstrated to be computationally efficient, with a ten-fold reduction in the computational time compared to a physics-based building energy simulation, and suitable for uncertainty analysis and real-time control of building characteristics in response to uncertainty.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Mitigating Errors in External Respiratory Surrogate-Based Models of Tumor Position

Purpose: To investigate the effect of tumor site, measurement precision, tumor-surrogate correlation, training data selection, model design, and interpatient and interfraction variations on the accuracy of external marker-based models of tumor position. Methods and Materials: Cyberknife Synchrony system log files comprising synchronously acquired positions of external markers and the tumor from 167 treatment fractions were analyzed. The accuracy of Synchrony, ordinary-least-squares regression, and partial-least-squares regression models for predicting the tumor position from the external markers was evaluated. The quantity and timing of the data used to build the predictive model were varied. The effects of tumor-surrogate correlation and the precision in both the tumor and the external surrogate position measurements were explored by adding noise to the data. Results: The tumor position prediction errors increased during the duration of a fraction. Increasing the training data quantities did not always lead to more accurate models. Adding uncorrelated noise to the external marker-based inputs degraded the tumor-surrogate correlation models by 16% for partial-least-squares and 57% for ordinary-least-squares. External marker and tumor position measurement errors led to tumor position prediction changes 0.3-3.6 times the magnitude of the measurement errors, varying widely with model algorithm. The tumor position prediction errors were significantly associated with the patient index but not with the fraction index or tumor site. Partial-least-squares was as accurate as Synchrony and more accurate than ordinary-least-squares. Conclusions: The accuracy of surrogate-based inferential models of tumor position was affected by all the investigated factors, except for the tumor site and fraction index.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Magnetic Resonance Imaging–Guided versus Surrogate-Based Motion Tracking in Liver Radiation Therapy: A Prospective Comparative Study

Purpose: This study applied automatic feature detection on cine–magnetic resonance imaging (MRI) liver images in order to provide a prospective comparison between MRI-guided and surrogate-based tracking methods for motion-compensated liver radiation therapy. Methods and Materials: In a population of 30 subjects (5 volunteers plus 25 patients), 2 oblique sagittal slices were acquired across the liver at high temporal resolution. An algorithm based on scale invariant feature transform (SIFT) was used to extract and track multiple features throughout the image sequence. The position of abdominal markers was also measured directly from the image series, and the internal motion of each feature was quantified through multiparametric analysis. Surrogate-based tumor tracking with a state-of-the-art external/internal correlation model was simulated. The geometrical tracking error was measured, and its correlation with external motion parameters was also investigated. Finally, the potential gain in tracking accuracy relying on MRI guidance was quantified as a function of the maximum allowed tracking error. Results: An average of 45 features was extracted for each subject across the whole liver. The multi-parametric motion analysis reported relevant inter- and intrasubject variability, highlighting the value of patient-specific and spatially-distributed measurements. Surrogate-based tracking errors (relative to the motion amplitude) were were in the range 7% to 23% (1.02-3.57mm) and were significantly influenced by external motion parameters. The gain of MRI guidance compared to surrogate-based motion tracking was larger than 30% in 50% of the subjects when considering a 1.5-mm tracking error tolerance. Conclusions: Automatic feature detection applied to cine-MRI allows detailed liver motion description to be obtained. Such information was used to quantify the performance of surrogate-based tracking methods and to provide a prospective comparison with respect to MRI-guided radiation therapy, which could support the definition of patient-specific optimal treatment strategies.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Higher-order factorization machine for accurate surrogate modeling in material design

Efficient and robust optimization is important in material science for identifying optimal structural parameters and enhancing material performance. Surrogate-based active learning algorithms have recently gained great attention for their ability to efficiently navigate large, high-dimensional design spaces. Among surrogate models, 2 nd -order factorization machine (FM) models are widely employed as the surrogate model in active learning algorithms due to their balance between simplicity and effectiveness. However, their quadratic nature limits their capacity to capture complex, higher-order interactions among variables, often leading to suboptimal solutions. To overcome this limitation, we propose an active learning scheme integrating a 3 rd -order FM model, capable of modeling three-variable interactions and more intricate relationships in material systems. We comprehensively evaluate the surrogate modeling performance of the 3 rd -order FM case using various objective functions. Furthermore, we examine the optimization reliability and efficiency of the 3 rd -order FM-based active learning in a real-world material design task (e.g., nanophotonic structures for transparent radiative cooling). Our study shows that the 3 rd -order FM outperforms the 2 nd -order model in both surrogate accuracy and optimization performance, highlighting higher-order models’ promises for material design and optimization problems.

Factorization machine↗

General Multifidelity Surrogate Models: Framework and Active-Learning Strategies for Efficient Rare Event Simulation

Estimating the probability of failure for complex real-world systems using high-fidelity computational models is often prohibitively expensive, especially when the probability is small. Exploiting low-fidelity models can make this process more feasible, but merging information from multiple low-fidelity and high-fidelity models poses several challenges. Here, this paper presents a robust multi-fidelity surrogate modeling strategy in which the multi-fidelity surrogate is assembled using an active learning strategy using an on-the-fly model adequacy assessment set within a subset simulation framework for efficient reliability analysis. The multi-fidelity surrogate is assembled by first applying a Gaussian process correction to each low-fidelity model and assigning a model probability based on the model's local predictive accuracy and cost. Three strategies are proposed to fuse these individual surrogates into an overall surrogate model based on model averaging and deterministic/stochastic model selection. The strategies also dictate which model evaluations are necessary. No assumptions are made about the relationships between low-fidelity models, while the high-fidelity model is assumed to be the most accurate and most computationally expensive model. Through two analytical and two numerical case studies, including a case study evaluating the failure probability of Tristructural isotropic-coated (TRISO) nuclear fuels, the algorithm is shown to be highly accurate while drastically reducing the number of high-fidelity model calls (and hence computational cost).

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Constructing a Simulation Surrogate with Partially Observed Output

Gaussian process surrogates are a popular alternative to directly using computationally expensive simulation models. When the simulation output consists of many responses, dimension-reduction techniques are often employed to construct these surrogates. However, surrogate methods with dimension reduction generally rely on complete output training data. This article proposes a new Gaussian process surrogate method that permits the use of partially observed output while remaining computationally efficient. The new method involves the imputation of missing values and the adjustment of the covariance matrix used for Gaussian process inference. The resulting surrogate represents the available responses, disregards the missing responses, and provides meaningful uncertainty quantification. In conclusion, the proposed approach is shown to offer sharper inference than alternatives in a simulation study and a case study where an energy density functional model that frequently returns incomplete output is calibrated.

42 ENGINEERING↗

Acceleration of Power System Dynamic Simulations Using a Deep Equilibrium Layer and Neural ODE Surrogate

The dominant paradigm for power system dynamic simulation is to build system-level simulations by combining physics-based models of individual components. The sheer size of the system along with the rapid integration of inverter-based resources exacerbates the computational burden of running time domain simulations. Here, in this paper, we propose a data-driven surrogate model based on implicit machine learningspecifically deep equilibrium layers and neural ordinary differential equationsto learn a reduced order model of a portion of the full underlying system. The data-driven surrogate achieves similar accuracy and reduction in simulation time compared to a physics-based surrogate, without the constraint of requiring detailed knowledge of the underlying dynamic models. This work also establishes key requirements needed to integrate the surrogate into existing simulation workflows; the proposed surrogate is initialized to a steady state operating point that matches the power flow solution by design.

Neural ordinary differential equations↗

The verification and uncertainty quantification of surrogate models used for structural analysis

High fidelity modeling of complex systems can require large finite element models to capture the physics of interest. Typically these high-order models take an excessively long time to run. For important studies such as model validation and uncertainty quantification, where probabilistic measures of the response are required, a large number of simulations of the high fidelity model with different parameters are necessary. In addition, some environments, such as an extensive random vibration excitation, require a long simulation time to capture the entire event. A process that produces a highly efficient model from the original high order model is necessary to enable these analyses. These highly efficient models are referred to as surrogate models, for their purpose is to represent the main physics that is of importance, but decrease the computational burden. A critical aspect of any surrogate model is how faithfully the efficient model represents the original high-order model. This paper describes the process for verifying a surrogate model using response quantities of interest and quantifying the introduced uncertainties in the use of the surrogate model. This is done to help any analyst if they use any surrogate model; such has, Craig-Bampton Reduc tions, POD, etc.

Ross, Michael↗

Development of an Optimized Gasoline Surrogate Formulation for PACE Experiments and Simulations

New powertrain solutions are needed to address societal challenges stemming in part from a transportation sector which relies heavily on combustion of conventional hydrocarbon fuels like gasoline. One way to advance new solutions is through predictive simulations. Predictive simulations can potentially limit the extent of experimental validations, reduce time spent in engineering design cycles, unlock new strategies for high efficiency combustion with power density, and minimize tailpipe emissions. However, the simulation tools currently available are either too computationally expensive, inadequate in their accuracy, or a combination of the two. A DOE-funded consortium of national laboratories called the Partnership to Advance Combustion Engines (PACE) seeks to address this gap by rapidly delivering new knowledge and tools which enable market-competitive powertrain solutions for light-duty vehicles. Approaches to modeling combustion in powertrain systems typically incorporate computational fluid dynamics (CFD) simulations with chemical kinetic models. These CFD simulations generally use surrogate fuels featuring a limited number of existing components in a reduced kinetic model to limit computational costs. Such surrogate fuels often sacrifice matching several or more combustion and physical properties of the target gasoline fuel. Validation of surrogate fuels is also often only pursued for a small subset of standard metrics such as research or motor octane number (RON, MON). The purpose of this project is to develop an optimal E10 gasoline surrogate fuel suitable for tasks across PACE to facilitate rapid common analysis and progression toward consortium goals. This project received contributions from tasks led by eight PACE principal investigators. The project funding here reflects all tasks under Pitz, whose tasks also contributed to additional PACE projects. Based on the optimal gasoline surrogate fuel composition, a reduced chemical kinetic model will be shared with PACE researchers, industry, and the broader combustion community.

33 ADVANCED PROPULSION SYSTEMS↗

Development of Microreactor Automated Control System (MACS): Surrogate Plant-level Modeling and Control Algorithms Integration

This report discusses progress on the modeling and integration task as part of the development of Microreactor Automated Control System (MACS). What follows is a discussion of the software tools used to develop a surrogate microreactor plant-level model, MACS module development, and a summary of the findings from integration with a hardware-in-the-loop (HIL) setup at Idaho National Laboratory (INL). Oak Ridge National Laboratory worked with INL to understand available hardware at INL (such as existing control platforms). With this information and using a preliminary framework for MACS with software interfaces defined, surrogate models and initial automated control algorithms were created for integration into the hardware to provide a HIL demonstration platform for MACS. Available reduced order reactor models that leverage existing microreactor neutronics and thermal hydraulics models have been integrated in the surrogate plant-level model. The surrogate model has been evaluated for sensitivity of all parameters and the MACS software modules (including the surrogate model) have been demonstrated to interact with the hardware. The integration, however, pointed to the need for further improvements to the framework—and specifically improvements to the interfaces to ensure that the HIL simulator is capable of longer-term stable operation with MACS in the loop. These improvements will be the focus of future research.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Report on 238Pu(n,x) surrogate cross section measurement

The goal of this year's effort is to measure the {sup 238}Pu(n,f) and {sup 238}Pu(n,2n) cross section from 100 keV to 20 MeV. We designed a surrogate experiment that used the reaction {sup 239}Pu(a,a{prime}x) as a surrogate for {sup 238}Pu(n,x). The experiment was conducted using the STARS/LIBERACE experimental facility located at the 88 Inch Cyclotron at Lawrence Berkeley National Laboratory in January 2010. A description of the experiment and status of the data analysis is given. In order to obtain a reliable {sup 238}Pu(n,x) cross section we designed the experiment using the surrogate ratio technique. This technique allows one to measure a desired, unknown, cross section relative to a known cross section. In the present example, the {sup 238}Pu(n,x) cross section of interest is determined relative to the known {sup 235}U(n,x) cross section. To increase confidence in the results, and to reduce overall uncertainties, we are also determining the {sup 238}Pu(n,x) cross section relative to the known {sup 234}U(n,x) cross section. The compound nuclei of interest for this experiment were produced using inelastic alpha scattering. For example, {sup 236}U(a,a{prime}x) served as a surrogate for {sup 235}U(n,x); analogous reactions were considered for the other cross sections. Surrogate experiments determine the probabilities for the decay of the compound nuclei into the various channels of interest (fission, gamma decay) by measuring particle-fission (p-f) or particle?gamma (p?g) reaction spectra. By comparing the decay probabilities associated with the unknown cross section to that of a known cross section it is possible to obtain the ratio of these cross sections and thus determine the unknown, desired cross section.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Using a surrogate-assisted Bayesian framework to calibrate the runoff-generation scheme in the Energy Exascale Earth System Model (E3SM) v1

Abstract. Runoff is a critical component of the terrestrial water cycle, and Earth system models (ESMs) are essential tools to study its spatiotemporal variability. Runoff schemes in ESMs typically include many parameters so that model calibration is necessary to improve the accuracy of simulated runoff. However, runoff calibration at a global scale is challenging because of the high computational cost and the lack of reliable observational datasets. In this study, we calibrated 11 runoff relevant parameters in the Energy Exascale Earth System Model (E3SM) Land Model (ELM) using a surrogate-assisted Bayesian framework. First, the polynomial chaos expansion machinery with Bayesian compressed sensing is used to construct computationally inexpensive surrogate models for ELM-simulated runoff at 0.5∘ × 0.5∘ for 1991–2010. The error metric between the ELM simulations and the benchmark data is selected to construct the surrogates, which facilitates efficient calibration and avoids the more conventional, but challenging, construction of high-dimensional surrogates for the ELM simulated runoff. Second, the Sobol' index sensitivity analysis is performed using the surrogate models to identify the most sensitive parameters, and our results show that, in most regions, ELM-simulated runoff is strongly sensitive to 3 of the 11 uncertain parameters. Third, a Bayesian method is used to infer the optimal values of the most sensitive parameters using an observation-based global runoff dataset as the benchmark. Our results show that model performance is significantly improved with the inferred parameter values. Although the parametric uncertainty of simulated runoff is reduced after the parameter inference, it remains comparable to the multimodel ensemble uncertainty represented by the global hydrological models in ISMIP2a. Additionally, the annual global runoff trend during the simulation period is not well constrained by the inferred parameter values, suggesting the importance of including parametric uncertainty in future runoff projections.

58 GEOSCIENCES↗