SEARCH · Search NASA

Results for “deep ensemble”

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.

SourceResearch connectionFrom yearThrough year

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

At least 91 records · Page 5

All sources NASA NTRS DOE OSTI

DOE OSTI2024

SPY: a conceptual design study of a magnet system for a high-pressure gaseous TPC neutrino detector

Here, we present a novel conceptual design for a magnet system that provides the magnetic field necessary for the analysis of tracks in a high-pressure gaseous argon TPC while simultaneously serving as a pressure vessel to contain the TPC gas volume. The magnet was developed within a Near Detector proposal for the Deep Underground Neutrino Experiment (DUNE). The high-pressure gaseous argon TPC is a component proposed to be one of the elements of an ensemble of near detectors that are needed for DUNE.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

NASA NTRS2020

Linkage among Ice Crystal Microphysics, Mesoscale Dynamics and Cloud and Precipitation Structures Revealed By Collocated Microwave Radiometer and Multi-Frequency Radar Observations

Ice clouds and falling snow are ubiquitous globally and play important roles in the Earth’s radiation budget and precipitation processes. Ice particle microphysical properties (e.g., size, habit and orientation) are not only influenced by ambient environment’s dynamic and thermodynamic conditions, but also intimately connect to the cloud radiative effects and particle fall speeds, which therefore impact up to the future climate projection and down to the details of the surface precipitation (e.g., onset-time, location, type and strength). Our previous work revealed that high-frequency Polarimetric radiance Difference (PD) from passive microwave sensors is a good indicator of the bulk aspect ratio of horizontally oriented ice particles that are often occur inside anvil clouds and/or stratiform precipitations. In this current work, we further investigate the dynamic/thermodynamic mechanisms and cloud/precipitation structures associated with ice-phase microphysics corresponding to different PD signals. In order to do so, collocated CloudSat radar (W-band) and Global Precipitation Measurement Dual-frequency Precipitation Radar (GPM-DPR, Ku/Ka bands) observations as well as European Centre for Medium-Range Weather Forecasts (ECMWF) atmosphere background profiles are grouped according to the magnitude of PD for only stratiform precipitation and/or anvil cloud scenes. We found that horizontally-oriented snow aggregates or large snow particles are likely the major contributor to the high-PD signals at 166 GHz, while low-PD magnitudes can be attributed to small cloud ice, randomly oriented snow aggregates, riming snow or super-cooled water. Further, high (low) PD scenes are found to be associated with stronger (weaker) wind shear and higher (lower) ambient humidity, both of which help promote (prohibit) the growth of frozen particles and the organization of convective systems. An ensemble of squall line cases is studied at the end to demonstrate that the PD asymmetry in the leading and trailing edges of the deep convection line is closely tied to the anvil cloud and stratiform precipitation layers respectively, suggesting the potential usefulness of PD as a proxy of stratiform/convective precipitation flag, as well as a proxy of convection life stage.

Ice crystal microphysics↗

DOE OSTI2025

Dataset for "Machine Learning Ensembles Can Enhance Hydrologic Predictions and Uncertainty Quantification" Willard et al. (2025).

This data release provides all data and code used in the paper " "Machine Learning Ensembles Can Enhance Hydrologic Predictions and Uncertainty Quantifications" Willard et al. (2025)" to model stream temperature, evaluate, and assess results. The associated manuscript explores the effect of different ensemble construction techniques across different common machine learning (ML) architectures for predictions in unmonitored basins. Modeling was done using long short-term memory (LSTM), gated recurrent unit (GRU), temporal convolution network (TCN), and extreme gradient boosting (XGBoost) models, and stream site coverage spans 1362 locations across the conterminous United States. The ensemble construction techniques investigated include ensemble by random weight initialization, differing hyperparameters, different random subsets of training data, different subselections of input features, different architectures, and Monte Carlo Dropout. The data is organized into these items items:Code repository and data for the paper " "Machine Learning Ensembles Can Enhance Hydrologic Predictions and Uncertainty Quantifications" Willard et al. (2025).Code: stream_temp_ml_regionalization.zip contains the code repositoryData to run the code:- data_dir.zip -- contains all files that should be moved to the "DATA_DIR" variable defined in the "set_env_vars.sh" script in the code repository- metadata_dir.zip -- contains all files that should be moved to the "METADATA_DIR" variable defined in the "set_env_vars.sh" script in the code repositoryData produced by the code and used in the paper:- outputs_dir.zip - contains model output and results (outputs_dir/results), model weights (outputs_dir/models), and all other outputs used for the paper including feature importances.To cite this code, please use the following BibTeX or MLA entries:bibtex:@misc{willard2025streamensembles,author = {Jared Willard and Charuleka Varadharajan},title = {Dataset for "Machine Learning Ensembles Can Enhance Hydrologic Predictions and Uncertainty Quantification"},year = {2024},doi = {10.15485/2527393},publisher = {ESS-DIVE Repository},url = {https://data.ess-dive.lbl.gov/datasets/doi:10.15485/2527393}}MLA: Willard, Jared, et al. Dataset for "Machine Learning Ensembles Can Enhance Hydrologic Predictions and Uncertainty Quantification". 2025. ESS-DIVE Repository, doi:10.15485/2448016.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2026

Active operator learning with predictive uncertainty quantification for partial differential equations

With the increased prevalence of neural operators being used to provide rapid solutions to partial differential equations (PDEs), understanding the accuracy of model predictions and the associated error levels is necessary for deploying reliable surrogate models in scientific applications. Existing uncertainty quantification (UQ) frameworks employ ensembles or Bayesian methods, which can incur substantial computational costs during both training and inference. Here, we propose a lightweight predictive UQ method tailored for Deep operator networks (DeepONets) that also generalizes to other operator networks. Numerical experiments on linear and nonlinear PDEs demonstrate that the framework’s uncertainty estimates are unbiased and provide accurate out-of-distribution uncertainty predictions with a sufficiently large training dataset. Our framework provides fast inference and uncertainty estimates that can efficiently drive outer-loop analyses that would be prohibitively expensive with conventional solvers. We demonstrate how predictive uncertainties can be used in the context of Bayesian optimization and active learning problems to yield improvements in accuracy and data-efficiency for outer-loop optimization procedures. In the active learning setup, we extend the framework to Fourier Neural Operators (FNO) and describe a generalized method for other operator networks. To enable real-time deployment, we introduce an inference strategy based on precomputed trunk outputs and a sparse placement matrix, reducing evaluation time by more than a factor of five. Our method provides a practical route to uncertainty-aware operator learning in time-sensitive settings.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

From clutter to clarity: Emergent neural operators via questionnaire metrics

Real-world datasets in chemical engineering and bioengineering processes—such as those from catalytic reactors, multiphase flows, polymerization reactors, bioreactors, and clinical trials—can often be unlabeled or disorganized, rendering the training of existing supervised learning models ineffective at learning the underlying dynamics. To salvage these datasets for decision-making, we first seek to obtain clarity from the cluttered data. Here, we present a framework for developing “structural” generative models, discovering emergent equations, and constructing efficient emulators from scrambled datasets by integrating unsupervised organizational learning techniques (Questionnaires) with advanced deep learning architectures (Deep Hidden Physics Models and Deep Operator Networks). Our approach is demonstrated on two illustrative model systems: (a) a 1D advection–diffusion partial differential equation representing a winding underground pipe and (b) an ensemble of Stuart–Landau oscillators, an agent-based system of coupled ordinary differential equations. In both cases, we successfully reconstruct meaningful spatial, temporal, and parameter embeddings from scrambled data, enabling good predictions of system dynamics. As a result, we highlight the framework’s potential for broader applications, enabling data-driven system identification in fields with inherently disorganized or hidden parameter spaces.

42 ENGINEERING↗

DOE OSTI2024

Summer Convective Precipitation Changes Over the Great Lakes Region Under a Warming Scenario

To understand future summer precipitation changes over the Great Lakes Region (GLR), we performed an ensemble of regional climate simulations through the Pseudo-Global Warming (PGW) approach. We found that different types of convective precipitation respond differently to the PGW signal. Isolated deep convection (IDC), usually concentrated in the southern domain, shows an increase in precipitation to the north of the GLR. Mesoscale convective systems (MCSs), usually concentrated upwind of the GLR, shift to the downwind region with increased precipitation. Thermodynamic variables such as convective available potential energy (CAPE) and convective inhibition energy (CIN) are found to increase across almost the entire studied domain, creating a potential environment more favorable for stronger convection systems and less favorable for weaker ones. Meanwhile, changes in the lifting condensation level (LCL) and level of free convection (LFC) show a strong correlation with variations in convective precipitation, highlighting the significance of these thermodynamic factors in controlling precipitation over the domain. Our results indicate that the decrease in LCL and LCF in areas with increased convective precipitation is mainly due to increased atmospheric moisture. In response to the prescribed warming perturbation, MCSs occur more frequently downwind, while localized IDCs exhibit more intense rain rates, longer durations, and larger rainfall area.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS1986

A study of the response of deep tropical clouds to mesoscale processes - Three-dimensional numerical experiments

A three-dimensional numerical cloud model is utilized to study the collective feedback effects of cloud systems on the large-scale environment. The characteristic features of the cloud ensemble model are described. The model is applied to a case of a well-defined ITCZ rainband over the eastern tropical Atlantic ocean. The location, number, and configuration of the clouds that formed in the model during the simulations are examined. The cloud heating and moistening, environmental wind shear, and vertical transports of momentum are analyzed. The collective feedbacks of cloud systems on the large-scale temperature, moisture, and horizontal momentum fields are computed after each simulation. The effect of the pressure gradient force on the v-momentum flux is investigated. It is observed that there is good correlation between the model simulations and the observed data.

Tao, Wei-Kuo↗

NASA NTRS2021

Groundwater Recharge Estimated by Land Surface Models: An Evaluation in the Conterminous U.S.

Estimating diffuse recharge of precipitation is fundamental to assessing groundwater sustainability. Diffuse recharge is also the process through which climate and climate change directly affect groundwater. In this study, we evaluated diffuse recharge over the conterminous U.S. simulated by a suite of land surface models (LSMs) that were forced using a common set of meteorological input data. Simulated annual recharge exhibited spatial patterns that were similar among the LSMs, with the highest values in the eastern U.S. and Pacific Northwest. However, the magnitudes of annual recharge varied significantly among the models and were associated with differences in simulated ET, runoff and snow. Evaluation against two independent datasets did not answer the question of whether the ensemble mean performs the best, due to inconsistency between those datasets. The amplitude and timing of seasonal maximum recharge differed among the models, influenced strongly by model physics governing deep soil moisture drainage rates and, in cold regions, snowmelt. Evaluation using in situ soil moisture observations suggested that true recharge peaks 1-3 months later than simulated recharge, indicating systematic biases in simulating deep soil moisture. However, recharge from lateral flows and through preferential flows cannot be inferred from soil moisture data, and the seasonal cycle of simulated groundwater storage actually compared well with in situ groundwater observations. Long-term trends in recharge were not consistently correlated with either precipitation trends or temperature trends. This study highlights the need to employ dynamic flow models in LSMs, among other improvements, to enable more accurate simulation of recharge.

snowmelt↗

DOE OSTI2025

Heterogeneous energetic material damage simulator (HEDS): A deep learning approach to simulate damage–sensitivity linkages

Damage in the microstructures of energetic materials (EMs), such as propellants and plastic bonded explosives (PBXs), can significantly alter their response to external loads. Both sensitization and desensitization can occur, causing concerns with safety and performance in the field; predictive models that connect damage and the sensitivity of EMs can enable design and provide confidence in their robustness and reliability. However, modeling of damage evolution is challenging for real microstructures of EMs; samples of damaged EMs are difficult to obtain, thereby hindering experiments and direct numerical simulations to determine the sensitivity of EMs at various stages of damage. Here, we develop an approach to generate synthetic, i.e., in silico produced, damaged microstructures for use in simulations to connect damage levels to sensitivity. The development of the present workflow to generate and impose varying levels of damage in microstructures, known as HEDS (Heterogeneous Energetic Material Damage Simulator), begins with a small set of images of damaged PBXs and combines a collection of deep neural network techniques to generate microstructures with varying levels of damage. By making the synthetic microstructures conform closely to those observed in available real, imaged microstructures, we develop an ensemble of damaged microstructures that can be used for in silico shock experiments. HEDS develops these microstructure ensembles as level set fields, which are directly employed in a sharp interface Eulerian hydrocode where shock simulations are performed to quantify the energy release rate from hotspot fields generated in the microstructure. These capabilities can be useful for the analysis and assessment of changes in the sensitivity of EMs and to design formulations that are less susceptible to damage-induced changes in sensitivity and performance.

Fang, Irene (ORCID:0009000844557122)↗

NASA NTRS2015

High-Resolution NU-WRF Simulations of a Deep Convective-Precipitation System During MC3E: Comparisons Between Goddard Microphysics Schemes and Observations - Part 1

The Goddard microphysics scheme was recently improved by adding a 4th ice class (frozen dropshail). This new 4ICE scheme was implemented and tested in the Goddard Cumulus Ensemble model (GCE) for an intense continental squall line and a moderate,less-organized continental case. Simulated peak radar reflectivity profiles were improved both in intensity and shape for both cases as were the overall reflectivity probability distributions versus observations. In this study, the new Goddard 4ICE scheme is implemented into the regional-scale NASA Unified - Weather Research and Forecasting model (NU-WRF) and tested on an intense mesoscale convective system that occurred during the Midlatitude Continental Convective Clouds Experiment (MC3E). The NU42WRF simulated radar reflectivities, rainfall intensities, and vertical and horizontal structure using the new 4ICE scheme agree as well as or significantly better with observations than when using previous versions of the Goddard 3ICE (graupel or hail) schemes. In the 4ICE scheme, the bin microphysics-based rain evaporation correction produces more erect convective cores, while modification of the unrealistic collection of ice by dry hail produces narrow and intense cores, allowing more slow-falling snow to be transported rearward. Together with a revised snow size mapping, the 4ICE scheme produces a more horizontally stratified trailing stratiform region with a broad, more coherent light rain area. In addition, the NU-WRF 4ICE simulated radar reflectivity distributions are consistent with and generally superior to those using the GCE due to the less restrictive open lateral boundaries

Regional Scale Model↗

DOE OSTI2025

Predicting critical heat flux with uncertainty quantification and domain generalization using conditional variational autoencoders and deep neural networks

Deep generative models (DGMs) can generate synthetic data samples that closely resemble the original dataset, addressing data scarcity. In this work, we developed a conditional variational autoencoder (CVAE) to augment critical heat flux (CHF) data used for the 2006 Groeneveld lookup table. To compare with traditional methods, a fine-tuned deep neural network (DNN) regression model was evaluated on the same dataset. Both models achieved small mean absolute relative errors, with the CVAE showing more favorable results. Uncertainty quantification (UQ) was performed using repeated CVAE sampling and DNN ensembling. The DNN ensemble improved performance over the baseline, while the CVAE maintained consistent results with less variability and higher confidence. Both models achieved small errors inside and outside the training domain, with slightly larger errors outside. Altogether, the CVAE performed better than the DNN in predicting CHF and exhibited better uncertainty behavior.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2026

Leveraging machine learning to enhance aerosol classification using Single-Particle Mass Spectrometry

Advancing automated classification of atmospheric aerosols from Single-Particle Mass Spectrometry (SPMS) data remains challenging due to overlapping ion signatures, compositional diversity, and limited labeled data. This study evaluates supervised and semi-supervised learning frameworks to enhance aerosol identification by jointly leveraging labeled and unlabeled spectra. Four models were compared: a supervised Support Vector Machine (SVM), a self-training SVM, a stacked autoencoder classifier, and a stacked autoencoder trained using a temporal-ensembling Mean Teacher approach. All models achieved high and stable accuracies (90.0 %–91.1 %), surpassing previous results on the same dataset (87 %) and matching the performance of state-of-the-art deep learning methods. Despite small global metric differences (≤ 1 %), semi-supervised variants yielded up to 5 %–10 % improvements for compositionally rare particle types – such as soot (0.77 % of spectra, F1-score: 0.93–0.97) and hazelnut pollen (0.98 % of spectra, F1-score: 0.97–1.00) – equating to roughly ∼ 187 additional correctly classified spectra. These gains are scientifically significant, as such rare particles exert disproportionate influence on radiative absorption and ice nucleation processes; their improved detection reduces modeled uncertainties in aerosol absorption optical depth and mixed-phase cloud ice nucleation rates. The models' residual misclassifications (≈ 9 %) largely arise from true spectral overlap among chemically adjacent species (e.g., Na- vs. K-feldspar, coated vs. uncoated feldspars), reflecting physical compositional continuity rather than algorithmic error. Collectively, these findings demonstrate that leveraging unlabeled data to learn robust spectral representations and refine classification enhances both fidelity and interpretability, bridging data-driven analysis with aerosol–climate process understanding.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

Deep learning model for fast, science-based forecasting of fluid migration along faults in geologic carbon storage scenarios

Effective long-term geologic storage depends on robust site selection and credible, science-based forecasting of subsurface behavior to ensure storage integrity. For this work, we develop a deep learning–based reduced-order model (ROM) to quantify potential carbon dioxide (CO₂) and brine migration through geological faults. The ROM combines a Transformer model for binary classification and a Stacked Ensemble for regression, trained on a comprehensive dataset generated from 1400 physics-based reservoir simulations. Key geologic and operational parameters—including fault geometry, reservoir structure, and injection conditions—were systematically varied to capture a wide range of fluid migration scenarios. The ROM accurately predicts the onset of migration, cumulative migration volumes of both CO₂ and brine, and associated migration rates, as compared to an independent set of validation simulations, while significantly reducing computational cost compared to traditional simulation methods. Model performance was evaluated across diverse fault configurations, revealing that shallow reservoir geometry and fault angle are among the most influential factors governing migration behavior. Sensitivity analysis using SHapley Additive exPlanations (SHAP) provided interpretability, revealing distinct patterns in how geological and operational features drive transient versus cumulative migration outcomes. The ROM’s ability to rapidly simulate fault migration scenarios enables efficient sensitivity analyses, scenario evaluations, and decision support for site selection and monitoring design. This approach enhances the safety, scalability, and long-term operational performance of geologic carbon storage (GCS) systems by providing a robust, interpretable tool for predicting subsurface fluid migration and assessing fault-related migration potential.

42 ENGINEERING↗

NASA NTRS1992

Long-term evolution of 1991 DA: A dynamically evolved extinct Halley-type comet

The long-term dynamical evolution of 21 variational orbits for the intermediate-period asteroid 1991 DA was followed for up to +/-10(exp 5) years from the present. 1991 DA is close to the 2:7 resonance with Jupiter; it has avoided close encounters, within 1 AU, with this planet for at least the past 30,000 years, even at the node crossing. The future evolution typically shows no close encounters with Jupiter within at least 50,000 years. This corresponds to the mean time between node crossings with either Jupiter or Saturn. Close encounters with Saturn and Jupiter lead to a chaotic evolution for the whole ensemble, while secular perturbations cause large-amplitude swings in eccentricity and inclination (the latter covering the range 15 deg approximately less than i approximately less than 85 deg) which correlate with deep excursions of the perihelion distance to values much less than 1 AU. These variations are similar to those found in P/Machholz and a variety of other high-inclination orbits, e.g., P/Hartley-IRAS. We emphasize the connection between the orbital evolution of 1991 DA and that of Halley-type comets. If 1991 DA was once a comet, it is not surprising that it is now extinct.

Hahn, Gerhard↗

DOE OSTI2025

Generalizable Image Segmentation for Microstructure Characterization Through Integrated SEM and EBSD Analysis

We demonstrate generalizable semantic segmentation using minimal ground truth data. Correlated scanning electron microscopy (SEM) images and electron backscatter diffraction (EBSD) measurements of frictionstir processed 316L stainless steel plates were used to train deep learning models for grain boundary segmentation. Secondary electron (SE) imaging taken at an accelerating voltage of 10 keV correlated to EBSD-derived grain boundaries produced the best performing model. Notably, an ensemble of three models trained on a single SE image produced accurate segmentation over a series of BSE images of samples manufactured under different processing parameters, with a resultant mean absolute error in grain size of 0.34 µm. The striking generalizability of the models likely results from the similar escape depths of the SE training input and the EBSD training output and the reduced probability of dislocation artifacts appearing in the image. This finding highlights the importance of considering the physical principles behind imaging in the development of robust segmentation models for microstructure characterization.

Taufique, Mohammad Fuad Nur↗

NASA NTRS2011

Using Enhanced Grace Water Storage Data to Improve Drought Detection by the U.S. and North American Drought Monitors

NASA's Gravity Recovery and Climate Experiment (GRACE) satellites measure time variations of the Earth's gravity field enabling reliable detection of spatio-temporal variations in total terrestrial water storage (TWS), including groundwater. The U.S. and North American Drought Monitors rely heavily on precipitation indices and do not currently incorporate systematic observations of deep soil moisture and groundwater storage conditions. Thus GRACE has great potential to improve the Drought Monitors by filling this observational gap. GRACE TWS data were assimilating into the Catchment Land Surface Model using an ensemble Kalman smoother enabling spatial and temporal downscaling and vertical decomposition into soil moisture and groundwater components. The Drought Monitors combine several short- and long-term drought indicators expressed in percentiles as a reference to their historical frequency of occurrence. To be consistent, we generated a climatology of estimated soil moisture and ground water based on a 60-year Catchment model simulation, which was used to convert seven years of GRACE assimilated fields into drought indicator percentiles. At this stage we provide a preliminary evaluation of the GRACE assimilated moisture and indicator fields.

Houborg, Rasmus↗

DOE OSTI2026

The Role of Subcloud Mesoscale Convergence in Sculpting Convective Updraft Width and Depth

The initiation of deep moist convection is governed in part by the horizontal width of updrafts near cloud base, which limits the deleterious effects of entrainment-driven dilution on buoyant thermals ascending through the free troposphere. However, the factors controlling cloud-base updraft width, which in turn dictates cloud depth, are not well understood. We track the evolving three-dimensional structure of the mesoscale subcloud forcing for vertical motion and near-cloud thermodynamic ingredients within a high-resolution ensemble of simulations of seven realistic daytime orographic convection initiation events to determine their relative roles in controlling cloud width and depth. Statistical analysis of approximately 5000 cloudy updraft samples indicates that the most important contributors to the width of cloudy updrafts across the ensemble are the depth and magnitude of the subcloud mesoscale ascent. However, the depth achieved by clouds is more consistently predicted by the near-cloud ambient relative humidity within the lower to middle free troposphere and convective available potential energy. Therefore, although the width of cloudy updrafts may be partly set at low levels by the mesoscale vertical mass and moisture flux, the likelihood of deep moist convection is governed by the generation of positive buoyancy within cumulus thermals and entrainment-driven dilution that reduces it. The persistence of the low-level mesoscale vertical forcing locally consolidates and vertically transports boundary layer moisture, helping to reduce updraft dilution. However, these factors vary in relative impacts on cloudy updrafts across individual cases, indicating multiple pathways for deep convection initiation.

Convective storms↗

NASA NTRS1978

The modification and application of Kuo's parameterization of cumulus convection in middle latitudes

An evaluation to reveal physical insight into the involved mechanisms through a combination of the large scale heat and moisture budgets and formulation of the latent heat released was performed. To improve upon the treatment of the interaction of deep convection with the environment, modifications consisting of considerations of the large scale moisture supply and of the vertical transport of moisture and of dry static energy inside the cloud are made. Also, a two layer cloud ensemble model is combined with the modified scheme. An examination of the modified scheme is performed by combining the large scale heat and moisture budgets and the modified formulation of the latent heat released. The modified parameterization procedures are compared with the original scheme, and the results of tests of both schemes are discussed. A variational optimization scheme is developed to analyze the wind field such that the integrated continuity equation and the global boundary condition are satisfied.

Liang, W. J.↗