SEARCH · Engineering Papers

Results for “ensemble prediction”

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 991 records · Page 55

All sources NASA NTRS DOE OSTI

DOE OSTI2020

Capturing multireference excited states by constrained-density-functional theory

The computation of excited electronic states with commonly employed (approximate) methods is challenging, typically yielding states of lower quality than the corresponding ground state for a higher computational cost. In this work, we present a mean-field method that extends the previously proposed excited constrained-density-functional theory (XCDFT) from single Slater determinants to ensemble one-body reduced density matrices for computing low-lying excited states. The method still retains an associated computational complexity comparable to a semilocal density functional theory (DFT) calculation while at the same time it is capable of approaching states with multireference character. Further, we benchmark the quality of this method on well-established test sets, finding good descriptions of the electronic structure of multireference states and maintaining an overall accuracy for the predicted excitation energies comparable to semilocal time-dependent DFT.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

DOE OSTI2020

MFNets: Multifidelity data-driven networks for Bayesian learning and prediction

This paper presents a multifidelity uncertainty quantification framework called MFNets. We seek to address three existing challenges that arise when experimental and simulation data from different sources are used to enhance statistical estimation and prediction with quantified uncertainty. Specifically, we demonstrate that MFNets can (1) fuse heterogeneous data sources arising from simulations with different parameterizations, e.g simulation models with different uncertain parameters or data sets collected under different environmental conditions; (2) encode known relationships among data sources to reduce data requirements; and (3) improve the robustness of existing multi-fidelity approaches to corrupted data. MFNets construct a network of latent variables (LVs) to facilitate the fusion of data from an ensemble of sources of varying credibility and cost. These LVs are posited as explanatory variables that provide the source of correlation in the observed data. Furthermore, MFNets provide a way to encode prior physical knowledge to enable efficient estimation of statistics and/or construction of surrogates via conditional independence relations on the LVs. We highlight the utility of our framework with a number of theoretical results which assess the quality of the posterior mean as a frequentist estimator and compare it to standard sampling approaches that use single fidelity, multilevel, and control variate Monte Carlo estimators. We also use the proposed framework to derive the Monte Carlo-based control variate estimator entirely from the use of Bayes rule and linear-Gaussian models -- to our knowledge the first such derivation. Finally, we demonstrate the ability to work with different uncertain parameters across different models.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

Intermediate-mass Early-type Disk Galaxies in the Virgo Cluster. II. Near-Infrared Spectra and Evidence for Differences in Evolution

We discuss near-infrared (NIR) slit spectra of six early-type disk galaxies in the Virgo Cluster that were examined previously at visible/red wavelengths. Despite having similar intrinsic luminosities, colors, and morphologies, the NIR spectrophotometric properties of these galaxies indicate that they are not a homogeneous ensemble differing only in terms of luminosity-weighted age and metallicity. While the depth of the CO(2,0) band is consistent with the centers of these galaxies having solar or slightly sub-solar luminosity-weighted metallicities, galaxy-to-galaxy variations in the depth of the Na i 2.21 μm doublet are found, with Na i2.21 μm lines in three galaxies (NGC 4491, NGC 4584, and NGC 4620) that are deeper than those predicted for a solar chemical mixture and a solar-neighborhood mass function. In contrast, the Ca i2.26 μm triplet shows good galaxy-to-galaxy agreement but tends to be deeper than the model prediction. Considering the depth of the NaD lines, the deep Na i2.21 μm lines are tentatively attributed to a bottom-heavy mass function. This is counter to observed trends between mass function slope and velocity dispersion, and so the possibility of a super-solar [Na/Fe] is also discussed. Two of the three galaxies with deep Na i 2.21 μm (NGC 4584 and NGC 4620) have Sérsic exponents that are consistent with a classical bulge. As for NGC 4491, its central NIR spectrum contains prominent emission lines. The relative strengths of Brγ and H{sub 2}S(1), the presence of [Fe ii] emission, and the mid-infrared spectral-energy distribution are all consistent with intense recent star formation near the center of that galaxy. The NIR spectrum of NGC 4584 is devoid of line emission in the NIR, suggesting that star formation does not power the emission detected at visible wavelengths from that galaxy. Wavelengths that contain the Ballick–Ramsey C{sub 2} band at 1.76 μm are matched by moderately metal-poor E-MILES model spectra with an age of 2 Gyr. The radial age trends in these galaxies are in the opposite sense to those in late-type disk galaxies, and it is concluded that they have evolved in a cluster environment for at least several Gyr.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRSDate not supplied

Economic Impact Assessments (EIA) of application of GEOGLOWS in Ecuador: Data Gaps, Limitations and Recommendations

In 2022, the United Nations launched the Early Warnings for All (EW4ALL) Program to establish global early warning systems by 2027. To assess the impact of the substantial $3.1 billion annual investment over five years, EW4ALL will consider factors that will require national coordination for the data needed for these assessments. In 2023, Ecuador was identified as one of the world's most climate-vulnerable countries, emphasizing the need to enhance its early warning systems. In 2020, the SERVIR Amazonia hub implemented the GEOGLOWS streamflow forecast service in collaboration with Ecuador's national meteorological agency (INAMHI). GEOGloWS provides 15-day ensemble forecasts and 80 years of historical streamflow data for every river worldwide through a free web service. The World Meteorological Organization has recognized this initiative as essential in contributing to the UN's call to ensure an 'Early Warning for All' by 2027. In 2023, as part of NASA's continuous efforts to fund research for Policy-Relevant Implementations, an economic impact assessment (EIA) was performed to understand the potential socioeconomic benefits of Early streamflow predictions in Ecuador using the GEOGLOWS service. Preliminary findings highlighted that gaps remain in effectively integrating socioeconomic and Earth observation (EO) data to capture the total value of these predictions. Implementing GEOGLOWS has led to valuable hydrological forecasts; however, the total economic benefits have yet to be documented. This study addresses the gaps and makes recommendations for future work that should focus on capturing the socio-economic benefits and costs associated with these forecasts, including their impact on decision-making at national and local levels. Despite the daily use of GEOGLOWS by key figures, including the President of Ecuador, the need for comprehensive recommendations and assessments is urgent.

Reetwika Basu↗

DOE OSTI2024

Effective optimization of atomic decoration in giant and superstructurally ordered crystals with machine learning

Crystals with complicated geometry are often observed with mixed chemical occupancy among Wyckoff sites, presenting a unique challenge for accurate atomic modeling. Similar systems possessing exact occupancy on all the sites can exhibit superstructural ordering, dramatically inflating the unit cell size. In this work, a crystal graph convolutional neural network (CGCNN) is used to predict optimal atomic decorations on fixed crystalline geometries. This is achieved with a site permutation search (SPS) optimization algorithm based on Monte Carlo moves combined with simulated annealing and basin-hopping techniques. Our approach relies on the evidence that, for a given chemical composition, a CGCNN estimates the correct energetic ordering of different atomic decorations, as predicted by electronic structure calculations. This provides a suitable energy landscape that can be optimized according to site occupation, allowing the prediction of chemical decoration in crystals exhibiting mixed or disordered occupancy, or superstructural ordering. Verification of the procedure is carried out on several known compounds, including the superstructurally ordered clathrate compound Rb8Ga27Sb16 and vacancy-ordered perovskite Cs2SnI6, neither of which was previously seen during the neural network training. In addition, the critical temperature of an order–disorder phase transition in solid solution CuZn is probed with our SPS routines by sampling site configuration trajectories in the canonical ensemble. This strategy provides an accurate method for determining favorable decoration in complex crystals and analyzing site occupation at unprecedented speed and scale.

Chemistry↗

DOE OSTI2022

Differentiable, Learnable, Regionalized Process-Based Models With Multiphysical Outputs can Approach State-Of-The-Art Hydrologic Prediction Accuracy

Predictions of hydrologic variables across the entire water cycle have significant value for water resources management as well as downstream applications such as ecosystem and water quality modeling. Recently, purely data-driven deep learning models like long short-term memory (LSTM) showed seemingly insurmountable performance in modeling rainfall runoff and other geoscientific variables, yet they cannot predict untrained physical variables and remain challenging to interpret. Here, we show that differentiable, learnable, process-based models (called δ models here) can approach the performance level of LSTM for the intensively observed variable (streamflow) with regionalized parameterization. We use a simple hydrologic model HBV as the backbone and use embedded neural networks, which can only be trained in a differentiable programming framework, to parameterize, enhance, or replace the process-based model's modules. Without using an ensemble or post-processor, δ models can obtain a median Nash-Sutcliffe efficiency of 0.732 for 671 basins across the USA for the Daymet forcing data set, compared to 0.748 from a state-of-the-art LSTM model with the same setup. For another forcing data set, the difference is even smaller: 0.715 versus 0.722. Meanwhile, the resulting learnable process-based models can output a full set of untrained variables, for example, soil and groundwater storage, snowpack, evapotranspiration, and baseflow, and can later be constrained by their observations. Both simulated evapotranspiration and fraction of discharge from baseflow agreed decently with alternative estimates. The general framework can work with models with various process complexity and opens up the path for learning physics from big data.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2020

A novel active optimization approach for rapid and efficient design space exploration using ensemble machine learning

In this work, a novel design optimization technique based on active learning, which involves dynamic exploration and exploitation of the design space of interest using an ensemble of machine learning algorithms, is presented. In this approach, a hybrid methodology incorporating an explorative weak learner (regularized basis function model) which fits high-level information about the response surface, and an exploitative strong learner (based on committee machine) that fits finer details around promising regions identified by the weak learner, is employed. For each design iteration, an aristocratic approach is used to select a set of nominees, where points that meet a threshold merit value as predicted by the weak learner are selected to be evaluated using expensive function evaluation. In addition to these points, the global optimum as predicted by the strong learner is also evaluated to enable rapid convergence to the actual global optimum once the most promising region has been identified by the optimizer. This methodology is first tested by applying it to the optimization of a two-dimensional multi-modal surface. The performance of the new active learning approach is compared with traditional global optimization methods, namely micro-genetic algorithm (pGA) and particle swarm optimization (PSO). It is demonstrated that the new optimizer is able to reach the global optimum much faster with a significantly fewer number of function evaluations. Subsequently, the new optimizer is also applied to a complex internal combustion (IC) engine combustion optimization case with nine control parameters related to fuel injection, initial thermodynamic conditions, and in-cylinder flow. It is again found that the new approach significantly lowers the number of function evaluations that are needed to reach the optimum design configuration (by up to 80%) when compared to particle swarm and genetic algorithm-based optimization techniques.

Machine learning, active learning, design optimiza↗

NASA NTRS1982

Modeling human target acquisition in ground-to-air weapon systems

The problems associated with formulating and validating mathematical models for describing and predicting human target acquisition response are considered. In particular, the extension of the human observer model to include the acquisition phase as well as the tracking segment is presented. Relationship of the Observer model structure to the more complex Standard Optimal Control model formulation and to the simpler Transfer Function/Noise representation is discussed. Problems pertinent to structural identifiability and the form of the parameterization are elucidated. A systematic approach toward the identification of the observer acquisition model parameters from ensemble tracking error data is presented.

Phatak, A. V.↗

NASA NTRSDate not supplied

Medium-Range River Flood Forecasts Using a Long Short-Term Memory Network

River flooding and the impacts are a concern for decision makers throughout the United States. Accurate medium-range forecasts (~3-7 days) are critical for providing advanced outlooks to emergency management officials. Unfortunately, accurately forecasting rainfall-runoff and the subsequent rise and fall within rivers remain a challenge in hydrological modeling. While complex physical modeling systems are the standard for representing the hydrological processes, they are computationally demanding and can require extensive calibration. Further, uncertainties remain in the model parameters and input data. The use of machine learning can reduce some of the computational demand while maintaining high accuracy. Therefore, this project makes use of a Long Short-Term Memory (LSTM) network which explicitly accounts for the time-dependent nature of rainfall-runoff modeling. The developed LSTM was trained to predict river gauge height, or stage height, based on time-lagged input features which include: gauge height to initialize the model, the NASA Short-term Prediction Research and Transition Center’s instance of the Land Information System (SPoRT-LIS) relative soil moisture to describe the rainfall infiltration rate, and 6-hr Multi-Radar Multi-Sensor quantitative precipitation estimate (MRMS QPE). The developed LSTM based system is then used to produce 7-day forecasts with a 6-hr temporal resolution using three different quantitative precipitation forecasts (QPF) from the NWS’s Weather Prediction Center (WPC), the NCEP Global Forecast System (GFS) model and the National Blend of Models (NBM). This trained modeling system has been implemented as an experimental product at over 100 different rivers in collaboration with at multiple National Weather Service (NWS) Forecast Offices and River Forecast Centers (RFC) across the eastern half of the United States. The developed LSTM model achieved average Nash-Sutcliffe efficiency (NSE) 0.89 higher than the equivalent medium-range National Water Model ensemble member forecast over a 7-day forecast. In addition to the initial development and evaluation, this project has continued to expand. While the initial model was developed for precipitation dominated basins, expansion of the project has taken it to basins effected by snow melt. This presentation will provide an overview of the project with focus on recent developments on incorporating snow melt processes into the model.

Andrew T. White↗

DOE OSTI2022

Adaptive experimental design for multi-fidelity surrogate modeling of multi-disciplinary systems

We present an adaptive algorithm for constructing surrogate models of multi-disciplinary systems composed of a set of coupled components. With this goal we introduce “coupling” variables with a priori unknown distributions that allow surrogates of each component to be built independently. Once built, the surrogates of the components are combined to form an integrated-surrogate that can be used to predict system-level quantities of interest at a fraction of the cost of the original model. The error in the integrated-surrogate is greedily minimized using an experimental design procedure that allocates the amount of training data, used to construct each component-surrogate, based on the contribution of those surrogates to the error of the integrated-surrogate. Here, the multi-fidelity procedure presented is a generalization of multi-index stochastic collocation that can leverage ensembles of models of varying cost and accuracy, for one or more components, to reduce the computational cost of constructing the integrated-surrogate. Extensive numerical results demonstrate that, for a fixed computational budget, our algorithm is able to produce surrogates that are orders of magnitude more accurate than methods that treat the integrated system as a black-box.

42 ENGINEERING↗

DOE OSTI2025

Using feature importance as an exploratory data analysis tool on Earth system models

Abstract. Machine learning (ML) models are commonly used to generate predictions, but these models can also support the discovery of new science. Generating accurate predictions necessitates that a model captures the structure of the underlying data. If the structure is properly extracted, ML could be a useful exploratory and evidential tool. In this paper, we present a case study that demonstrates the use of ML for exploratory data analysis (EDA) in the climate space. We apply the ML explainability method of spatiotemporal zeroed feature importance (stZFI) to understand how climate-variable associations evolve over space and time. Our analyses focus on data from ensembles of Earth system models (ESMs) which provide data on different climate states and conditions. We elect to work with ESM ensembles since they allow us to compare feature importance across alternative scenarios not available with observed data. The ensembles also account for natural variability so that we can distinguish between signal and noise due to natural climate variability when computing feature importance. The use of perturbed initial condition ensembles introduces variability mimicking the natural variability in the atmosphere; thus the signals emerging using feature importance (FI) can be evaluated against the natural variability in the climate system. For our analyses, we consider the 1991 volcanic eruption of Mount Pinatubo, which was a large stratospheric aerosol injection. We explore the climate pathway associated with the eruption from aerosols to radiation to temperature at both the near-surface and stratospheric levels. In addition to applying the method to data generated from two different ESMs, we apply stZFI to reanalysis data to compare the associations identified by stZFI. We show how stZFI tracks the importance of aerosol optical depth over time on forecasting temperatures. This case study illustrates usefulness of an ML tool (stZFI) for EDA on a well-studied climate exemplar.

Ries, Daniel (ORCID:0000000250294647)↗

NASA NTRS2010

Role of Subsurface Physics in the Assimilation of Surface Soil Moisture Observations

Root zone soil moisture controls the land-atmosphere exchange of water and energy and exhibits memory that may be useful for climate prediction at monthly scales. Assimilation of satellite-based surface soil moisture observations into a land surface model is an effective way to estimate large-scale root zone soil moisture. The propagation of surface information into deeper soil layers depends on the model-specific representation of subsurface physics that is used in the assimilation system. In a suite of experiments we assimilate synthetic surface soil moisture observations into four different models (Catchment, Mosaic, Noah and CLM) using the Ensemble Kalman Filter. We demonstrate that identical twin experiments significantly overestimate the information that can be obtained from the assimilation of surface soil moisture observations. The second key result indicates that the potential of surface soil moisture assimilation to improve root zone information is higher when the surface to root zone coupling is stronger. Our experiments also suggest that (faced with unknown true subsurface physics) overestimating surface to root zone coupling in the assimilation system provides more robust skill improvements in the root zone compared with underestimating the coupling. When CLM is excluded from the analysis, the skill improvements from using models with different vertical coupling strengths are comparable for different subsurface truths. Finally, the skill improvements through assimilation were found to be sensitive to the regional climate and soil types.

Reichle, R. H.↗

DOE OSTI2022

Quantum signatures in nonlinear gravitational waves

The effective quantum field theory description of gravity, despite its non-renormalizability, allows for predictions beyond classical general relativity. As we enter the age of gravitational wave astronomy, an important and timely question is whether measurable quantum predictions that depart from classical gravity, analogous to quantum optics effects which cannot be explained by classical electrodynamics, can be found. In this work, we investigate quantum signatures in gravitational waves using tools from quantum optics. Squeezed-coherent gravitational waves, which can exhibit sub-Poissonian graviton statistics, can enhance or suppress the signal measured by an interferometer, a characteristic effect of quantum squeezing. Moreover, we show that Gaussian gravitational wave quantum states can be reconstructed from measurements over an ensemble of optical fields interacting with a single copy of the gravitational wave, thus opening the possibility of detecting quantum features of gravity beyond classical general relativity.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

NASA NTRS1989

Predicting peculiar interstellar extinction from gaseous abundances

Molecular and atomic abundances are examined for 19 lines of sight through dense clouds, each with a peculiar selective extinction curve. The interstellar clouds appear to fall into two distinct categories: CN-rich, with relatively small amounts of neutral iron, or CN-poor, with large amounts of neutral iron. Lines of sight having a CN/(Fe I) abundance ratio about two or greater, are found to have a shallow 2175 A feature relative to the underlying extinction, while the strength of bump is 3.60 + or - 0.36 for the other dense clouds in the present study. The difference is the strength of the extinction bump between these two ensembles is 1.03 + or - 0.23. Several atomic abundances are examined as potential indicators of peculiar extinction. The Mn I upper limits in the present study support that Mn is being preferentially depleted. Fe I is underabundant relative to K I by 0.7 (dex) in the large (CN)/(Fe I) compared to the small (CN)/(Fe I) lines of sight.

Joseph, Charles L.↗

NASA NTRSDate not supplied

Continuous Habitable Zones: Pairing a GCM and Bayesian Framework to Predict Habitable Zone Evolution

In the near-future, new space telescopes like JWST, LUVOIR, and HabEx will begin attempting to explore the properties of atmospheres of potentially habitable planets. This will require a significant amount of time and resources for even a single planet, which makes it essential to prioritize observations by those most-likely to have detectable life. Here we present a statistical method to estimate the probabilities that specific exoplanets have been continuously in the habitable zone of their host stars for more than 2 billion years, the approximate time it took life on Earth to significantly increase the oxygen content of the atmosphere. We introduce the use of statistics of an ensemble of 3D planetary general circulation models to estimate these probabilities, replacing prior 1D model estimates.

habitable planets↗

NASA NTRSDate not supplied

Pairing a GCM and Bayesian Framework to Predict Habitable Zone Evolution

habitable planets↗

DOE OSTI2025

Radiative, Hydrologic, and Circulation Responses to Warming in Cess‐Potter Simulations Using the Global 3.25‐km SCREAM

Using the global 3.25-km Simple Cloud Resolving E3SM Atmosphere Model (SCREAM 3 km), a pair of 13-month Cess-Potter simulations are performed to quantify the radiative feedbacks and the hydrologic and circulation responses to warming. Large-scale aspects of SCREAM 3 km's top-of-atmosphere radiative fluxes, precipitation rates, and circulations are in good agreement with observations and reanalysis, with notable differences, including a drier lower free-troposphere in the Tropics, reduced precipitation and humidity over the Tropical West Pacific, and poleward shifted Southern Hemisphere midlatitude jet. In response to warming, SCREAM 3 km predicts a total radiative feedback within the top 15% of the CMIP5 and CMIP6 models, which puts it substantially higher than the feedback reported by other kilometer-scale models. SCREAM 3 km's high radiative feedback stems from a strongly positive shortwave cloud feedback, most prominent over the mid- and high-latitudes. SCREAM 3 km's high precipitation response also puts it among the highest of CMIP models, whereas its circulation response is within the spread of CMIP models. An ensemble of five perturbed initial condition Cess-Potter simulations with a 12 km version of SCREAM (SCREAM 12 km) is performed to characterize uncertainty and resolution sensitivity. It suggests that the uncertainty from analyzing a pair of 1-year simulations is small compared to the inter-model spread in feedbacks and precipitation response. SCREAM 12 km also produces a strong precipitation response to warming but a much lower cloud feedback and total radiative feedback. The results from these experiments suggest that the spread in climate feedbacks will likely persist in the next generation of kilometer-scale models.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS2012

Dipole Alignment in Rotating MHD Turbulence

We present numerical results from long-term CPU and GPU simulations of rotating, homogeneous, magnetohydrodynamic (MHD) turbulence, and discuss their connection to the spherically bounded case. We compare our numerical results with a statistical theory of geodynamo action that has evolved from the absolute equilibrium ensemble theory of ideal MHD turbulence, which is based on the ideal MHD invariants are energy, cross helicity and magnetic helicity. However, for rotating MHD turbulence, the cross helicity is no longer an exact invariant, although rms cross helicity becomes quasistationary during an ideal MHD simulation. This and the anisotropy imposed by rotation suggests an ansatz in which an effective, nonzero value of cross helicity is assigned to axisymmetric modes and zero cross helicity to non-axisymmetric modes. This hybrid statistics predicts a large-scale quasistationary magnetic field due to broken ergodicity , as well as dipole vector alignment with the rotation axis, both of which are observed numerically. We find that only a relatively small value of effective cross helicity leads to the prediction of a dipole moment vector that is closely aligned (less than 10 degrees) with the rotation axis. We also discuss the effect of initial conditions, dissipation and grid size on the numerical simulations and statistical theory.

Shebalin, John V.↗