Search NASA⌕ Search

SEARCH · Search NASA

Results for “data base for model validation”

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 451 records · Page 25

Effects of fuel composition and octane sensitivity on polycyclic aromatic hydrocarbon and soot emissions of gasoline–ethanol blend surrogates

The sooting propensity of a fuel is closely coupled with the fuel composition and chemistry. A detailed understanding of their effects is, therefore, needed to develop next-generation fuels which can minimize particulate emissions. With this overarching goal, the present work numerically investigates the effects of fuel composition and octane sensitivity (S) on polycyclic aromatic hydrocarbons (PAH) and soot emissions, for four-component gasoline-ethanol blend surrogates comprising isooctane, n-heptane, toluene, and ethanol. A partially-premixed counterflow flame is chosen as the canonical configuration for this study and simulations are performed using CHEMKIN-Pro-employing a kinetic mechanism developed by Park et al. (2017). In addition, a, detailed soot model based on the sectional method is used to capture the spatial characteristics of soot emissions. The kinetic mechanism and soot model are validated using available experimental data for various targets. A total of 86 TPRF-E mixtures, spanning a wide range of concentration of each component, and a wide range of S are analyzed. The effect of each non-paraffinic fuel component on the resultant PAH and soot emissions is investigated. PAH and soot emissions are found to vary significantly depending upon the blend composition. Additionally, based on the parametric sweeps, a regression analysis is carried out to identify global parameters that govern the formation of PAHs and soot. The analysis shows that both toluene content and S have a prominent effect on the formation of PAHs and soot, with toluene content having a stronger impact. Moreover, larger PAHs have higher dependency on toluene content and S. Furthermore, a detailed analysis is carried out to understand the physical and chemical phenomena associated with the observed trends of PAH and soot emissions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Connect the Dots: In Situ 4-D Seismic Monitoring of CO 2 Storage With Spatio-Temporal CNNs

4-D seismic imaging has been widely used in CO 2 sequestration projects to monitor the fluid flow in the volumetric subsurface region that is not sampled by wells. Ideally, real-time monitoring and near-future forecasting would provide site operators with great insights to understand the dynamics of the subsurface reservoir and assess any potential risks. However, due to obstacles such as high deployment cost, availability of acquisition equipment, exclusion zones around surface structures, only very sparse seismic imaging data can be obtained during monitoring. That leads to an unavoidable and growing knowledge gap over time. The operator needs to understand the fluid flow throughout the project lifetime and the seismic data are only available at a limited number of times. This is insufficient for understanding reservoir behavior. To overcome those challenges, we have developed spatio-temporal neural-network-based models that can produce high-fidelity interpolated or extrapolated images effectively and efficiently. Specifically, our models are built on an autoencoder, and incorporate the long short-term memory (LSTM) structure with a new loss function regularized by optical flow. We validate the performance of our models using real 4-D post-stack seismic imaging data acquired at the Sleipner CO 2 sequestration field. We employ two different strategies in evaluating our models. Numerically, we compare our models with different baseline approaches using classic pixel-based metrics. We also conduct a blind survey and collect a total of 20 responses from domain experts to evaluate the quality of data generated by our models. Finally, via both numerical and expert evaluation, we conclude that our models can produce high-quality 2-D/3-D seismic imaging data at a reasonable cost, offering the possibility of real-time monitoring or even near-future forecasting of the CO 2 storage reservoir.

4-D seismic imaging↗

Battery state-of-health diagnostics during fast cycling using physics-informed deep-learning

Rapid, in-situ Li-ion battery state-of-health (SOH) quantification is challenging. Li-ion battery aging can vary significantly with chemistry, operating conditions, cycling demands, electrode design, and operation history. As a cell ages, optimal and safe operating conditions need to be adapted to account for battery degradation by tracking critical aging modes such as loss-of-lithium-inventory (LLI), loss-of-active-material (LAM) in either electrode, and/or impedance rise. This manuscript describes a framework for identifying battery aging modes in-operando using fast-rate voltage charge/discharge responses. The framework uses a physically based Li-ion battery model to produce synthetic high-rate responses at aged states. The aging model is calibrated against experimental data from cells with different electrode loadings and cycled under a variety of fast-charging conditions (1 h, 15 min, 10 min, and 7 min charging). The synthetically generated high-rate responses at aged states are then used to train a deep-learning model to identify real cell state-of-health from fast charge/discharge battery voltage responses. The synthetically trained deep-learning model performance is validated by comparing to standard incremental capacity analysis and half-cell measurements. Finally, the framework demonstrates the benefits of using high-rate physics-based models to generate synthetic data for training deep-learning models.

25 ENERGY STORAGE↗

CATMoS: Collaborative Acute Toxicity Modeling Suite

Background: Humans are exposed to tens of thousands of chemical substances that need to be assessed for their potential toxicity. Acute systemic toxicity testing serves as the basis for regulatory hazard classification, labeling, and risk management. However, it is cost- and time-prohibitive to evaluate all new and existing chemicals using traditional rodent acute toxicity tests. In silico models built using existing data facilitate rapid acute toxicity predictions without using animals. Objectives: The U.S. Interagency Coordinating Committee on the Validation of Alternative Methods Acute Toxicity Workgroup organized an international collaboration to develop in silico models for predicting acute oral toxicity based on five different endpoints: LD50 value, U.S. Environmental Protection Agency hazard categories, Globally Harmonized System for Classification and Labelling hazard categories, very toxic chemicals (LD50 =50 mg/kg), and non-toxic chemicals (LD50 >2000 mg/kg). Methods: An acute oral toxicity data inventory for 11,992 chemicals was compiled, split into training and evaluation sets, and made available to 35 participating international research groups that submitted a total of 139 predictive models. Predictions that fell within the applicability domains of the submitted models were evaluated using external validation sets. These were then combined into consensus models to leverage strengths of individual approaches. Results: The resulting consensus predictions, which leverage the collective strengths of each individual model, form the Collaborative Acute Toxicity Modeling Suite (CATMoS). CATMoS demonstrated high performance in terms of accuracy and robustness when compared to in vivo results. Discussion: CATMoS is being evaluated by regulatory agencies for its utility and applicability as a potential replacement for in vivo rat acute oral toxicity studies. CATMoS predictions for over 800,000 chemicals have been made available via the NTP’s Integrated Chemical Environment. The models are also implemented in a free, standalone open-source tool, OPERA, which allows predictions of new and untested chemicals to be made.

63 RADIATION, THERMAL, AND OTHER ENVIRON. POLLUTAN↗

Combustion of petroleum-based transportation fuels and their blends with biofuels: a new approach for developing surrogates and understanding the effects of blending

Accurate and versatile reaction mechanisms are necessary to simulate combustion of transportation fuels blended with biofuels for improving fuel efficiency, reducing fuel consumption and mitigating the formation and emissions of combustion products harmful to the atmosphere. This project involved a combined effort of simulation and experimentation of liquid fuel burning using a combustion configuration amenable to numerical modeling with a level of detail not previously achieved for blends of biofuels with gasoline certification and surrogate fuels. The configuration of an isolated droplet burning under conditions where gas transport arises solely from fuel evaporation was selected as the platform for experiments and numerical modeling. The spherical symmetry and one-dimensional gas transport that results enabled simulating the droplet burning process with an precedented level of detail. Processes associated with unsteady gas and liquid transport, formation of particulate and gaseous products, radiation, multicomponent phase equilibrium at the droplet surface and moving boundary effects from droplet evaporation, were incorporated in a single numerical framework. The numerical model was based on the open source code OPENSmoke++ (OS) adapted to incorporate the effects noted above. In addition to simulation, an experimental design was developed for the isolated droplet configuration to acquire the data needed for validating the numerical model. Several broad accomplishments of the project were the following: demonstrating generally excellent agreement between measured and simulated combustion parameters for the fuel blends examined that included mixtures of heptane and isobutanol as a model system, and surrogates comprised of up to seven miscible components mixed with ethanol or isobutanol as representative biofuel additives; using a new approach to validate reaction mechanisms of biofuel blends which incorporated fuel evaporation into the process along with developing an experimental design for acquiring data to compare with simulations; and showing that a wealth of information could be obtained on the combustion physics of biofuels from experiments requiring volumes on the order of only nanoliters at a time thus opening the way to evaluating biofuels synthesized by new processes early in development.

02 PETROLEUM↗

SUMMARY REPORT ON POTENTIAL IMPACTS OF SMALL COLUMN ION EXCHANGE ON DWPF GLASS FORMULATION

This report summarizes a large amount of experimental work completed to identify the potential impacts of material from Small Column Ion Exchange (SCIX) on glass formulation at the Defense Waste Processing Facility (DWPF). The results show no significant issues with the predicted values of chemical durability and viscosity using the current Product Composition Control System (PCCS) models when the SCIX components are added to projected DWPF glass compositions. No modifications to the viscosity and durability models appear to be necessary at this time in order to incorporate the SCIX streams at DWPF. It is recommended that the Savannah River National Laboratory (SRNL) continue to verify the durability and viscosity models as the projected compositions for DWPF processing evolve. It is also recommended that the data generated thus far be reviewed and a determination be made as to how best to extend the validation ranges of the durability and viscosity models. The liquidus temperatures for the experimental glasses are also reported and discussed in this report. The results show that the measured or estimated (based on measured data) liquidus temperature values for the glasses with SCIX components added are consistently higher than those predicted by the current model. Therefore, the PCCS liquidus temperature model will need to be modified in order to incorporate the SCIX streams at DWPF. It is recommended that SRNL carry out full measurements of the liquidus temperatures for those KT-series glasses where estimates have been made. These data should then be used to support an evaluation of whether a refitting of the liquidus temperature model coefficients will be sufficient to correctly predict the liquidus temperature of glasses containing the SCIX components (particularly higher TiO{sub 2} concentrations), or whether additional modifications to the model are required. While there are prediction issues with the current liquidus temperature model, they are not at this time expected to hamper the incorporation of SCIX streams at DWPF. The estimated liquidus temperatures, while higher than the model predicted values, remain below the current DWPF limit of 1050 C for most of the study glasses. Note that the properties and performance of the glasses in this study are highly dependent on glass composition. Therefore, should significant changes be made to the projected compositions or processing rates for SCIX or DWPF, many of the assessments and experiments may have to be revisited.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

The Analysis Description Language Ecosystem: Latest developments and physics applications

We present latest developments in Analysis Description Language (ADL), a declarative domain-specific language describing the physics algorithm of a HEP data analysis decoupled from software frameworks. Analyses written in ADL can be integrated into any framework for various tasks. ADL is a multipurpose construct with uses ranging from analysis design to preservation, reinterpretation, queries, visualisation, combination, etc. The most advanced infrastructure to execute ADL on events is the CutLang runtime interpreter. Recent technical developments include an automated interface with different data types, generation of the abstract syntax tree, a visualization tool that that auto-converts analysis flows to graphs, incorporation of trained machine learning models and a Jupyter-based plotting tool. We also report physics implications including a large scale LHC analysis implementation and validation effort for beyond the standard model reinterpretation purposes and studies with ATLAS and CMS open data.

Sekmen, Sezen [Kyungpook National Univ., Daegu (Ko↗

Predictive analytics of selections of russet potatoes

We explore the application of machine learning algorithms specifically to enhance the selection process of Russet potato (Solanum tuberosum L.) clones in breeding trials by predicting their suitability for advancement. This study addresses the challenge of efficiently identifying high-yield, disease-resistant, and climate-resilient potato varieties that meet processing industry standards. Leveraging manually collected data from trials in the state of Oregon, we investigate the potential of a wide variety of state-of-the-art binary classification models. The dataset includes 1086 clones, with data on 38 attributes recorded for each clone, focusing on yield, size, appearance, and frying characteristics, with several control varieties planted consistently across four Oregon regions from 2013 to 2021. We conduct a comprehensive analysis of the dataset that includes preprocessing, feature engineering, and imputation to address missing values. We focus on several key metrics such as accuracy, F1-score, and Matthews correlation coefficient (MCC) for model evaluation. The top-performing models, namely a feedforward neural network classifier (Neural Net), a histogram-based gradient boosting classifier (HGBC), and a support vector machine classifier (SVM), demonstrate consistent and significant results. To further validate our findings, we conducted a simulation study using the aims, data-generating mechanisms, estimands, methods, and performance measures (ADEMP) framework, simulating different data-generating scenarios to assess model robustness and performance through true positive, true negative, false positive, and false negative distributions, area under the receiver operating characteristic curve (AUC-ROC) and MCC. The simulation results highlight that non-linear models like SVM and HGBC consistently show higher AUC-ROC and MCC than logistic regression, thus outperforming the traditional linear model across various distributions, and emphasizing the importance of model selection and tuning in agricultural trials. Variable selection further enhances model performance and identifies influential features in predicting trial outcomes. The findings emphasize the potential of machine learning in streamlining the selection process for potato varieties, offering benefits such as increased efficiency, substantial cost savings, and judicious resource utilization. Our study contributes insights into precision agriculture and showcases the relevance of advanced technologies for informed decision-making in breeding programs.

60 APPLIED LIFE SCIENCES↗

Methodical Aspects of the Determination of Carbon Dioxide in Atmosphere Using FTIR Spectroscopy

Monitoring the state of ground-based spectroscopic systems ensures high accuracy of atmospheric gas composition measurements and the possibility of using measurement data for validating satellite measurements and numerical models. Column-averaged carbon dioxide mixing ratios (XCO{sub 2}) can be used to assess the quality of spectroscopic measurements because its content is stable in the atmosphere. The XCO{sub 2} values were retrieved from spectra measured by Fourier transform IR spectrometry (FTIR) using a Bruker 125HR instrument between 2009 and 2017 at the Peterhof station (St. Petersburg, Russia). The total error of the FTIR CO{sub 2} columns retrieval was 4.18 ± 0.02%, with 0.36 ± 0.06% and 4.16 ± 0.02% for random and systematic errors, respectively. The data obtained are compared with model calculations based on long-term observations at the Mauna Loa station. Various schemes solving for the inverse problem were analyzed and compared among themselves, and estimates of empirical random errors in the determination of XCO{sub 2} were obtained. To improve the accuracy of XCO{sub 2} measurements, it is recommended to determine the error of the Fourier-spectrometer phase and the slope of the zero line of the spectrum simultaneously with the atmospheric gas composition measurements taking into account various continuum sources of error.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Liquid Water: Obtaining the right answer for the right reasons

Water is ubiquitous on our planet and plays an essential role in many chemical and biological processes. Accurate models for water are crucial in understanding, controlling and predicting the physical and chemical properties of complex aqueous systems. Over the last few years we have been developing a molecular-level based approach for a macroscopic model for water that is based on the explicit description of the underlying intermolecular interactions between molecules in water clusters. In the absence of detailed experimental data for small water clusters, highly-accurate theoretical results are required to validate and parameterize model potentials. As an example of the benchmarks needed for the development of accurate models for the interaction between water molecules, for the most stable structure of (H2O)20 we ran a coupled-cluster calculation on the ORNL’s Jaguar petaflop computer that used over 100 TB of memory for a sustained performance of 487 TFLOP/s (double precision) on 96,000 processors, lasting for 2 hours. By this summer we will have studied multiple structures of both (H2O)20 and (H2O)30 and completed basis set and other convergence studies and anticipate the sustained performance rising close to 1 PFLOP/s.

aqueous↗

Liquid Water: Obtaining the right answer for the right reasons

Water is ubiquitous on our planet and plays an essential role in many chemical and biological processes. Accurate models for water are crucial in understanding, controlling and predicting the physical and chemical properties of complex aqueous systems. Over the last few years we have been developing a molecular-level based approach for a macroscopic model for water that is based on the explicit description of the underlying intermolecular interactions between molecules in water clusters. In the absence of detailed experimental data for small water clusters, highly-accurate theoretical results are required to validate and parameterize model potentials. As an example of the benchmarks needed for the development of accurate models for the interaction between water molecules, for the most stable structure of (H$_2$O)$_{20}$ we ran a coupled-cluster calculation on the ORNL's Jaguar petaflop computer that used over 100 TB of memory for a sustained performance of 487 TFLOP/s (double precision) on 96,000 processors, lasting for 2 hours. By this summer we will have studied multiple structures of both (H$_2$O)$_{20}$ and (H$_2$O)$_{30}$ and completed basis set and other convergence studies and anticipate the sustained performance rising close to 1 PFLOP/s.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A graph neural network (GNN) approach to basin-scale river network learning: the role of physics-based connectivity and data fusion

Abstract. Rivers and river habitats around the world are under sustained pressure from human activities and the changing global environment. Our ability to quantify and manage the river states in a timely manner is critical for protecting the public safety and natural resources. In recent years, vector-based river network models have enabled modeling of large river basins at increasingly fine resolutions, but are computationally demanding. This work presents a multistage, physics-guided, graph neural network (GNN) approach for basin-scale river network learning and streamflow forecasting. During training, we train a GNN model to approximate outputs of a high-resolution vector-based river network model; we then fine-tune the pretrained GNN model with streamflow observations. We further apply a graph-based, data-fusion step to correct prediction biases. The GNN-based framework is first demonstrated over a snow-dominated watershed in the western United States. A series of experiments are performed to test different training and imputation strategies. Results show that the trained GNN model can effectively serve as a surrogate of the process-based model with high accuracy, with median Kling–Gupta efficiency (KGE) greater than 0.97. Application of the graph-based data fusion further reduces mismatch between the GNN model and observations, with as much as 50 % KGE improvement over some cross-validation gages. To improve scalability, a graph-coarsening procedure is introduced and is demonstrated over a much larger basin. Results show that graph coarsening achieves comparable prediction skills at only a fraction of training cost, thus providing important insights into the degree of physical realism needed for developing large-scale GNN-based river network models.

54 ENVIRONMENTAL SCIENCES↗

Improved computational neutronics methods and validation protocols for the advanced test reactor

The Idaho National Laboratory (INL) is in the process of updating the various reactor physics modeling and simulation tools used to support operation and safety assurance of the Advanced Test Reactor (ATR). Key accomplishments so far have encompassed both computational as well as experimental work. A new suite of stochastic and deterministic transport theory based reactor physics codes and their supporting nuclear data libraries (HELIOS, KENO6/SCALE, NEWT/SCALE, ATTILA, and an extended implementation of MCNP5) has been installed at the INL. Corresponding models of the ATR and ATRC are now operational with all five codes, demonstrating the basic feasibility of the new code packages for their intended purposes. On the experimental side of the project, new hardware was fabricated, measurement protocols were finalized, and the first four of six planned physics code validation experiments based on neutron activation spectrometry have been conducted at the ATRC facility. Data analysis for the first three experiments, focused on characterization of the neutron spectrum in one of the ATR flux traps, has been completed. The six experiments will ultimately form the basis for flexible and repeatable ATR physics code validation protocols that are consistent with applicable national standards. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Availability of Critical Benchmark Experiments for the Pebble Tanker Transportation Model for Nuclear Criticality Safety Validation of TRISO Pebbles

This study addresses the need for comprehensive investigations into TRi-structural ISOtropic (TRISO) fuel pebble transportation validation. In this work, an exploratory model, the pebble tanker(PT), was developed with the aim of facilitating the validation of nuclear criticality safety calculations in the context of industrial-scale transportation of TRISO fuel. The PT model was designed to investigate the availability and applicability of critical benchmark experiments crucial for assessing the transportation of these pebbles. This work incorporated sensitivity/uncertainty (S/U) similarity studies to quantify the applicability of critical benchmark experiments and to address nuclear data uncertainties in the context of TRISO transportation. Two container models were investigated: one for the Hermes-type pebble and one for the Pebble Bed Modular Reactor (PBMR)–type pebble. The models were simplified, considering fuel, containment, and either water or air, to enable a focus on the underlying physics of applications involving TRISO fuel pebbles using the PT model. A crucial aspect under consideration was the capacity of the transport package to hold pebbles while ensuring subcriticality in the flooded state. An approach in the criticality validation process involves assessing the similarity between systems through an integral index parameter evaluation. This involves calculating a correlation coefficient (referred to as c k ) based on shared nuclear data–induced uncertainty between a benchmark experiment and the application of the PT model. To facilitate this analysis, the SCALE tools, particularly the CSAS6-Shift, TSUNAMI-3D-Shift, and TSUNAMI-IP sequences, were employed for comprehensive studies in neutronics and S/U analysis. Our findings showed that there are sufficient critical experimental benchmarks to perform this validation of the PT model in the most reactive state, i.e. when the tanker is flooded. This paper provides valuable insights into validating a transport package for Generation IV TRISO fuel pebbles.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

MOOSE-based Tritium Migration Analysis Program, Version 8 (TMAP8) for advanced open-source tritium transport and fuel cycle modeling

Tritium management is critical for the safety, sustainability, and economics of fusion energy systems, and advanced and reliable modeling tools help accelerate the development of tritium technologies. This paper presents the Tritium Migration Analysis Program, Version 8 (TMAP8), an open-source, MOOSE-based application developed to provide state-of-the-art tritium transport and fuel cycle modeling capabilities. TMAP8 aims to expand the capabilities of previous versions (i.e., TMAP4 and TMAP7) by leveraging modern computational techniques, ensuring high software quality assurance standards (key to building trust), and enabling multispecies, multiscale, and multiphysics simulations for integrated tritium transport modeling in complex geometries. This paper outlines TMAP8’s scope and rigorous development practices, emphasizing its transparency, accessibility, modularity, and reliability. We present the current suite of verification and validation cases based on those from TMAP4, demonstrating TMAP8’s accuracy and reliability against analytical solutions and experimental data. Additionally, the paper showcases TMAP8’s integrated fuel cycle modeling capabilities, highlighting its applicability at various scales and levels. The TMAP8 code and documentation are openly available, promoting collaborative development and widespread adoption within the fusion community. Future work will soon expand TMAP8’s verification and validation suite to include those from TMAP7 and other recent experimental studies for validation.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Intelligent Experiments Through Real-time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future EIC Detectors (Final Report)

The overall vision of this project was to integrate real-time artificial intelligence (AI) directly into the data acquisition and detector-control systems of nuclear physics experiments, including both fast online event selection and an autonomous detector-control feedback loop. The work carried out under the award focused on the fast online event-selection half of that vision: the efficient recording of low-momentum heavy-flavor (HF) hadron decays in proton-proton collisions at the sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC)—an observable that requires fast tracking and topological trigger selection not previously demonstrated at RHIC, and that is essential for QCD studies at future facilities such as the Electron-Ion Collider (EIC). The autonomous detector-control (GPU-based feedback) component named in the project title remained a design concept and was not implemented under this award. The Massachusetts Institute of Technology (MIT) group led the offline simulation and data processing needed to train the machine-learning (ML) models, the translation of trained models to Field-Programmable Gate Array (FPGA) firmware using the hls4ml framework, and the physics validation of heavy-flavor reconstruction. Over the award period, the team developed and hardware-tested the principal components of an AI-based heavy-flavor trigger on simulated and recorded sPHENIX tracker data: a software Bipartite Graph Attention Network (BiGAT) trigger model reaching > 95% signal efficiency at 99% background rejection; an FPGA-native hit clusterizer matching the offline clustering; smaller networks synthesized to FPGA within the required sub-10 µs latency; and an assembled decoder–clusterizer–inference firmware chain exercised on the FELIX readout board. A complete, fully integrated hardware demonstrator was not finished within the award period. This report documents the project goals, the MIT group’s contributions, the technical accomplishments, and the outlook toward applications at the future EIC ePIC detector.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Scalable deep learning for watershed model calibration

Watershed models such as the Soil and Water Assessment Tool (SWAT) consist of high-dimensional physical and empirical parameters. These parameters often need to be estimated/calibrated through inverse modeling to produce reliable predictions on hydrological fluxes and states. Existing parameter estimation methods can be time consuming, inefficient, and computationally expensive for high-dimensional problems. In this paper, we present an accurate and robust method to calibrate the SWAT model (i.e., 20 parameters) using scalable deep learning (DL). We developed inverse models based on convolutional neural networks (CNN) to assimilate observed streamflow data and estimate the SWAT model parameters. Scalable hyperparameter tuning is performed using high-performance computing resources to identify the top 50 optimal neural network architectures. We used ensemble SWAT simulations to train, validate, and test the CNN models. We estimated the parameters of the SWAT model using observed streamflow data and assessed the impact of measurement errors on SWAT model calibration. We tested and validated the proposed scalable DL methodology on the American River Watershed, located in the Pacific Northwest-based Yakima River basin. Our results show that the CNN-based calibration is better than two popular parameter estimation methods (i.e., the generalized likelihood uncertainty estimation [GLUE] and the dynamically dimensioned search [DDS], which is a global optimization algorithm). For the set of parameters that are sensitive to the observations, our proposed method yields narrower ranges than the GLUE method but broader ranges than values produced using the DDS method within the sampling range even under high relative observational errors. The SWAT model calibration performance using the CNNs, GLUE, and DDS methods are compared using R 2 and a set of efficiency metrics, including Nash-Sutcliffe, logarithmic Nash-Sutcliffe, Kling-Gupta, modified Kling-Gupta, and non-parametric Kling-Gupta scores, computed on the observed and simulated watershed responses. The best CNN-based calibrated set has scores of 0.71, 0.75, 0.85, 0.85, 0.86, and 0.91. The best DDS-based calibrated set has scores of 0.62, 0.69, 0.8, 0.77, 0.79, and 0.82. The best GLUE-based calibrated set has scores of 0.56, 0.58, 0.71, 0.7, 0.71, and 0.8. The scores above show that the CNN-based calibration leads to more accurate low and high streamflow predictions than the GLUE and DDS sets. Our research demonstrates that the proposed method has high potential to improve our current practice in calibrating large-scale integrated hydrologic models.

54 ENVIRONMENTAL SCIENCES↗