Search NASA⌕ Search

SEARCH · Search NASA

Results for “inverse parameter inference”

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 109 records · Page 6
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Synthetic modeling of soft x-ray emissivity for magnetic island analysis in LTX-β

We present a synthetic soft x-ray (SXR) forward-modeling framework that characterizes the emissivity structure of rotating magnetic islands in the Lithium Tokamak eXperiment-β across space and time. Magnetic islands associated with tearing modes produce modulations in line-integrated SXR brightness. Traditional tomographic methods struggle to resolve these structures in devices with limited sightlines, resulting in an under-determined inversion problem, or otherwise necessitate equilibrium reconstruction data, preventing their use in active control. Our approach removes this challenge by combining a fully three-dimensional ray-tracing model with a time-dependent emissivity prescription of a $m/n$ = 2/1 magnetic island. The model is validated through helical island geometry, rotation measured by magnetic diagnostics, and equilibrium constraints from PSI-Tri reconstructions. Synthetic brightness signals are generated for each photodiode sightline from a modeled emissivity profile and directly compared with experimental data from a tangential midplane SXR array. By fitting the synthetic diagnostic output to the observed brightness across time, we simultaneously infer key geometric parameters-such as island radial location, island width, and rotation frequency-without requiring full tomographic reconstruction, all with an average absolute deviation of less than 3%. This work demonstrates that forward modeling with a single tangential array can extract key island parameters in a spherical tokamak, providing a computationally efficient alternative to conventional SXR tomography and providing a pathway toward real-time magnetic-island characterization in future devices.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
NASA NTRS2016

Spaceborne Remote Sensing of Aerosol Type: Global Distribution, Model Evaluation and Translation into Chemical Speciation

It is essential to evaluate and refine aerosol classification methods applied to passive satellite remote sensing. We have developed an aerosol classification algorithm (called Specified Clustering and Mahalanobis Classification, SCMC) that assigns an aerosol type to multi-parameter retrievals by spaceborne, airborne or ground-based passive remote sensing instruments [1]. The aerosol types identified by our scheme are pure dust, polluted dust, urban-industrialdeveloped economy, urban-industrialdeveloping economy, dark biomass smoke, light biomass smoke and pure marine. We apply the SCMC method to inversions from the ground-based AErosol RObotic NETwork (AERONET [2]) and retrievals from the space-borne Polarization and Directionality of Earths Reflectances instrument (POLDER, [3]). The POLDER retrievals that we use differ from the standard POLDER retrievals [4] as they make full use of multi-angle, multispectral polarimetric data [5]. We analyze agreement in the aerosol types inferred from both AERONET and POLDER and evaluate GEOS-Chem [6] simulations over the globe. Finally, we use in-situ observations from the SEAC4RS airborne field experiment to bridge the gap between remote sensing-inferred qualitative SCMC aerosol types and their corresponding quantitative chemical speciation. We apply the SCMC method to airborne in-situ observations from the NASA Langley Aerosol Research Group Experiment (LARGE, [7]) and the Differential Aerosol Sizing and Hygroscopicity Spectrometer Probe (DASH-SP, [8]) instruments; we then relate each coarsely defined SCMC type to a sum of percentage of individual aerosol species, using in-situ observations from the Particle Analysis by Laser Mass Spectrometry (PALMS, [9]), the Soluble Acidic Gases and Aerosol (SAGA, [10]), and the High - Resolution Time - of - Flight Aerosol Mass Spectrometer (HR ToF AMS, [11]).

airborne↗
DOE OSTI2025

Characterization of DESI fiber assignment incompleteness effect on 2-point clustering and mitigation methods for DR1 analysis

We present an in-depth analysis of the fiber assignment incompleteness in the Dark Energy Spectroscopic Instrument (DESI) Data Release 1 (DR1). This incompleteness is caused by the restricted mobility of the robotic fiber positioner in the DESI focal plane, which limits the number of galaxies that can be observed at the same time, especially at small angular separations. As a result, the observed clustering amplitude is suppressed in a scale-dependent manner, which, if not addressed, can severely impact the inference of cosmological parameters. We discuss the methods adopted for simulating fiber assignment on mocks and data. In particular, we introduce the fast fiber assignment (FFA) emulator, which was employed to obtain the power spectrum covariance adopted for the DR1 full-shape analysis. We present the mitigation techniques, organised in two classes: measurement stage and model stage. We then use high fidelity mocks as a reference to quantify both the accuracy of the FFA emulator and the effectiveness of the different measurement-stage mitigation techniques. This complements the studies conducted in a parallel paper for the model-stage techniques, namely the θ-cut approach. We find that pairwise inverse probability (PIP) weights with angular upweighting recover the “true” clustering in all the cases considered, in both Fourier and configuration space. Notably, we present the first ever power spectrum measurement with PIP weights from real data.

cosmological simulations↗
NASA NTRS1976

Remote probing of atmospheric particulates from radiation extinction experiments: A review of methods

The existing methodology for reconstructing the particle size distribution and inferring the refractive index of absorbing and scattering atmospheric particulates is critically reviewed. Emphasis is placed on method capabilities and shortcomings and, wherever possible, on achievable accuracy. The nature of the associated remote probing problem is analyzed with regard to the effects of the particulates on EM wave propagation in the atmosphere. The parameterization of size distribution is studied within the unifying framework of Pearson's distribution curves. The inversions of extinction measurements and their ratios are considered separately, and the potentialities of each type of measurement are identified. Work lacking in each of the methods reviewed is indicated. A method of determining both the effective complex refractive index and size distribution model parameters from the same data is also presented. Lastly, determination from extinction ratio data of the complex refractive index independent of size distribution is discussed and error analyzed.

Fymat, A. L.↗
NASA NTRS1973

Determination of Atmospheric Aerosol Characteristics from the Polarization of Scattered Radiation

Aerosols affect the polarization of radiation in scattering, hence measured polarization can be used to infer the nature of the particles. Size distribution, particle shape, real and absorption parts of the complex refractive index affect the scattering. From Lorenz-Mie calculations of the 4-Stokes parameters as a function of scattering angle for various wavelengths the following polarization parameters were plotted: total intensity, intensity of polarization in plane of observation, intensity perpendicular to the plane of observation, polarization ratio, polarization (using all 4-Stokes parameters), plane of the polarization ellipse and its ellipticity. A six-component log-Gaussian size distribution model was used to study the effects of the nature of the polarization due to variations in the size distribution and complex refractive index. Though a rigorous inversion from measurements of scattering to detailed specification of aerosol characteristics is not possible, considerable information about the nature of the aerosols can be obtained. Only single scattering from aerosols was used in this paper. Also, the background due to Rayleigh gas scattering, the reduction of effects as a result of multiple scattering and polarization effects of possible ground background (airborne platforms) were not included.

Harris, F. S., Jr.↗
NASA NTRS2021

Radar-Based Bayesian Estimation of Ice Crystal Growth Parameters within a Microphysical Model

The potential for polarimetric Doppler radar measurements to improve predictions of ice microphysical processes within an idealized model–observational framework is examined. In an effort to more rigorously constrain ice growth processes (e.g., vapor deposition) with observations of natural clouds, a novel framework is developed to compare simulated and observed radar measurements, coupling a bulk adaptive-habit model of vapor growth to a polarimetric radar forward model. Bayesian inference on key microphysical model parameters is then used, via a Markov chain Monte Carlo sampler, to estimate the probability distribution of the model parameters. The statistical formalism of this method allows for robust estimates of the optimal parameter values, along with (non-Gaussian) estimates of their uncertainty. To demonstrate this framework, observations from Department of Energy radars in the Arctic during a case of pristine ice precipitation are used to constrain vapor deposition parameters in the adaptive habit model. The resulting parameter probability distributions provide physically plausible changes in ice particle density and aspect ratio during growth. A lack of direct constraint on the number concentration produces a range of possible mean particle sizes, with the mean size inversely correlated to number concentration. Consistency is found between the estimated inherent growth ratio and independent laboratory measurements, increasing confidence in the parameter PDFs and demonstrating the effectiveness of the radar measurements in constraining the parameters. The combined Doppler and polarimetric observations produce the highest-confidence estimates of the parameter PDFs, with the Doppler measurements providing a stronger constraint for this case.

Robert S. Schrom↗
DOE OSTI2024

Selecting Appropriate Model Complexity: An Example of Tracer Inversion for Thermal Prediction in Enhanced Geothermal Systems

Abstract A major challenge in the inversion of subsurface parameters is the ill‐posedness issue caused by the inherent subsurface complexities and the generally spatially sparse data. Appropriate simplifications of inversion models are thus necessary to make the inversion process tractable and meanwhile preserve the predictive ability of the inversion results. In this study, we investigate the effect of model complexity on fracture aperture inversion and thermal performance prediction in a field‐scale EGS model. Principal component analysis was used to map the aperture field to a low‐dimensional latent space. The complexity of the inversion model was quantitatively represented by the percentage of total variance in the original aperture fields preserved by the latent space. Tracer, pressure and flow rate data were used to invert for fracture aperture through an ensemble‐based inversion method, and the inferred aperture field was used to predict thermal performance. With an over‐simplified aperture model, ensemble collapse occurred. The inverted aperture models failed to resolve necessary flow and transport features, leading to a biased thermal performance prediction. A complex aperture model involved excessive features and was prone to overinterpreting the inversion data. Both the tracer/pressure/flow rate data reproduction and thermal prediction showed significant uncertainties, making it difficult to properly estimate long‐term thermal performance. Fortunately, our results indicate that there exists an appropriate model complexity which can simultaneously match inversion data and predict thermal performance with an acceptable uncertainty. The quality of the fit of tracer data appears to be a useful indicator of such an appropriate model complexity.

15 GEOTHERMAL ENERGY↗
NASA NTRSDate not supplied

Asteroseismic Inversions of Mixed Acoustic-Gravity Modes to Probe the Stellar Core Structure

The discovery of mixed acoustic-gravity modes of oscillations of moderate mass stars opens a unique opportunity to infer the structure of the inner energy-generating cores and thus test the stellar evolution theory. The mixed modes have properties of internal gravity waves (g-modes) in the convectively stable helium core and properties of acoustic modes outside the core. We select several sets of the oscillation mode frequencies in the mass range from about 1.3 to 1.6 solar masses from the Kepler Legacy database, and apply the optimally localized averaging inversion technique previously developed for low-degree helioseismology. The inversion technique takes into account the uncertainties in the determination of the mass and radius of the stars, as well as the surface effects. The methodology provides sensitivity kernels for various structure properties, including the sound speed, density, and Ledoux parameter of convective stability, and, thus, the direct relationship between the stellar properties and the deviation of observed frequencies from the reference models. The inversion results reveal significant deviations in the core structure from the reference models calculated using the MESA evolutionary code for the stellar parameters obtained by the asteroseismic model grid fitting. Our analysis shows that the best resolution of the inner helium core and surrounding shell is achieved in inversions for the Ledoux parameter.

SMD↗
DOE OSTI2025

MOOSE ProbML: Parallelized probabilistic machine learning and uncertainty quantification for computational energy applications

Here, this paper presents the development and demonstration of massively parallel probabilistic machine learning (ML) and uncertainty quantification (UQ) capabilities within the Multiphysics Object-Oriented Simulation Environment (MOOSE), an open-source computational platform for parallel finite element and finite volume analyses. In addressing the computational expense and uncertainties inherent in complex multiphysics simulations, this paper integrates Gaussian process (GP) variants, active learning, Bayesian inverse UQ, adaptive forward UQ, Bayesian optimization, evolutionary optimization, and Markov chain Monte Carlo (MCMC) within MOOSE. It also elaborates on the interaction among key MOOSE systems — Sampler, MultiApp, Reporter, and Surrogate — in enabling these capabilities. The modularity offered by these systems enables development of a multitude of probabilistic ML and UQ algorithms in MOOSE. Example code demonstrations include parallel active learning and parallel Bayesian inference via active learning. The impact of these developments is illustrated through five applications relevant to computational energy applications: UQ of nuclear fuel fission product release, using parallel active learning Bayesian inference; very rare events analysis in nuclear microreactors using active learning; advanced manufacturing process modeling using multi-output GPs (MOGPs) and dimensionality reduction; fluid flow using deep GPs (DGPs); and tritium transport model parameter optimization for fusion energy, using batch Bayesian optimization. These capabilities are part of the MOOSE framework.

97 - MATHEMATICS AND COMPUTING↗
NASA NTRS2001

Inductive Measurement of Plasma Jet Electrical Conductivity (MSFC Center Director's discretionary Fund)

Measurement of plasma jet electrical conductivity has utility in the development of explosively driven magnetohydrodynamic (MHD) energy converters as well as magnetic flux compression reaction chambers for nuclear/chemical pulse propulsion and power. Within these types of reactors, the physical parameter of critical importance to underlying MHD processes is the magnetic Reynolds number, the value of which depends upon the product of plasma electrical conductivity and velocity. Therefore, a thorough understanding of MHD phenomena at high magnetic Reynolds number is essential, and methods are needed for the accurate and reliable measurement of electrical conductivity in high-speed plasma jets. It is well known that direct measurements using electrodes suffer from large surface resistance, and an electrodeless technique is desired. To address this need, an inductive probing scheme, originally developed for shock tube studies, has been adapted. In this method, the perturbation of an applied magnetic field by a plasma jet induces a voltage in a search coil, which, in turn, can be used to infer electrical conductivity through the inversion of a Fredholm integral equation of the first kind. A 1-in.-diameter probe using a light-gas gun. Exploratory laboratory experiments were carried out using plasma jets expelled from 15-g shaped charges. Measured conductivities were in the range of 4 kS/m for unseeded octol charges and 26 kS/m for seeded octol charges containing 2-percent potassium carbonate by mass.

Turner, M. W.↗
NASA NTRS2016

Secure Mass Measurements From Transit Timing: 10 Kepler Exoplanets Between 3 and 8M⊕ with Diverse Densities and Incident Fluxes

We infer dynamical masses in eight multiplanet systems using transit times measured from Keplerʼs complete dataset, including short-cadence data where available. Of the 18 dynamical masses that we infer, 10 pass multiple tests for robustness. These are in systems Kepler-26 (KOI-250), Kepler-29 (KOI-738), Kepler-60 (KOI-2086), Kepler-105 (KOI-115), and Kepler-307 (KOI-1576). Kepler -105 c has a radius of 1.3R⊕ and a density consistent with an Earth-like composition. Strong transit timing variation (TTV) signals were detected from additional planets, but their inferred masses were sensitive to outliers or consistent solutions could not be found with independently measured transit times, including planets orbiting Kepler-49 (KOI-248), Kepler-57 (KOI-1270), Kepler-105 (KOI-115), and Kepler-177 (KOI-523). Nonetheless, strong upper limits on the mass of Kepler-177 cimply an extremely low density of 0.1 g cm−3. In most cases, individual orbital eccentricities were poorly constrained owing to degeneracies in TTV inversion. For five planet pairs in our sample, strong secular interactions imply a moderate to high likelihood of apsidal alignment over a wide range of possible eccentricities. We also find solutions for the three planets known to orbit Kepler-60 in a Laplace-like resonance chain. However, nonlibrating solutions also match the transit timing data. For six systems, we calculate more precise stellar parameters than previously known, enabling useful constraints on planetary densities where we have secure mass measurements. Placing these exoplanets on the mass–radius diagram, we find that a wide range of densities is observed among sub-Neptune-mass planets and that the range in observed densities is anticorrelated with incident flux.

Stars individual↗
NASA NTRS2016

Secure Mass Measurements From Transit Timing: 10 Kepler Exoplanets Between 3 and 8 M⊕ With Diverse Densities and Incident Fluxes

We infer dynamical masses in eight multiplanet systems using transit times measured from Keplerʼs complete data set, including short-cadence data where available. Of the 18 dynamical masses that we infer, 10 pass multiple tests for robustness. These are in systems Kepler-26 (KOI-250), Kepler-29 (KOI-738), Kepler-60 (KOI-2086), Kepler-105 (KOI-115), and Kepler-307 (KOI-1576). Kepler-105 c has a radius of 1.3 R(sub ⨁) and a density consistent with an Earth-like composition. Strong transit timing variation (TTV) signals were detected from additional planets, but their inferred masses were sensitive to outliers or consistent solutions could not be found with independently measured transit times, including planets orbiting Kepler-49 (KOI-248), Kepler-57 (KOI-1270), Kepler-105 (KOI-115), and Kepler-177 (KOI-523). Nonetheless, strong upper limits on the mass of Kepler-177 c imply an extremely low density of ∽0.1 g cm(exp −3). In most cases, individual orbital eccentricities were poorly constrained owing to degeneracies in TTV inversion. For five planet pairs in our sample, strong secular interactions imply a moderate to high likelihood of apsidal alignment over a wide range of possible eccentricities. We also find solutions for the three planets known to orbit Kepler-60 in a Laplace-like resonance chain. However, nonlibrating solutions also match the transit timing data. For six systems, we calculate more precise stellar parameters than previously known, enabling useful constraints on planetary densities where we have secure mass measurements. Placing these exoplanets on the mass–radius diagram, we find that a wide range of densities is observed among sub-Neptune-mass planets and that the range in observed densities is anticorrelated with incident flux.

Individual stars↗
NASA NTRS2013

Monte Carlo Bayesian Inference on a Statistical Model of Sub-gridcolumn Moisture Variability Using High-resolution Cloud Observations : Sensitivity Tests and Results - Part II

Part I presented a Monte Carlo Bayesian method for constraining a complex statistical model of GCM sub-gridcolumn moisture variability using high-resolution MODIS cloud data, thereby permitting large-scale model parameter estimation and cloud data assimilation. This part performs some basic testing of this new approach, verifying that it does indeed significantly reduce mean and standard deviation biases with respect to the assimilated MODIS cloud optical depth, brightness temperature and cloud top pressure, and that it also improves the simulated rotational-Ramman scattering cloud optical centroid pressure (OCP) against independent (non-assimilated) retrievals from the OMI instrument. Of particular interest, the Monte Carlo method does show skill in the especially difficult case where the background state is clear but cloudy observations exist. In traditional linearized data assimilation methods, a subsaturated background cannot produce clouds via any infinitesimal equilibrium perturbation, but the Monte Carlo approach allows finite jumps into regions of non-zero cloud probability. In the example provided, the method is able to restore marine stratocumulus near the Californian coast where the background state has a clear swath. This paper also examines a number of algorithmic and physical sensitivities of the new method and provides guidance for its cost-effective implementation. One obvious difficulty for the method, and other cloud data assimilation methods as well, is the lack of information content in the cloud observables on cloud vertical structure, beyond cloud top pressure and optical thickness, thus necessitating strong dependence on the background vertical moisture structure. It is found that a simple flow-dependent correlation modification due to Riishojgaard (1998) provides some help in this respect, by better honoring inversion structures in the background state.

Monte Carlo Bayesian inference↗
NASA NTRS1996

On the Relationship Between Transit Velocity of Interplanetary Shocks and Solar Active Processes

Recently, it was reported that preferential relationships exist between the transit velocity V(sub T) of earthward-directed interplanetary shocks and solar active processes, in particular, eruptive filaments outside active regions (the size of the erupting filament L(sub f)) and solar flares (the value of the X-ray characteristic J). Unfortunately, statistical testing of the proposed associations was not accomplished, nor was the 'geo-effectiveness' of the events adequately described. Reported here are the results of a re-examination of the 21 eruptive filaments (SSC-EF events) and 26 X-ray flares (SSC-F events) that have been associated with storm sudden commencements (SSCs) at Earth. Simple statistical testing refutes the claim that a preferential relationship exists between V(sub T) and L(sub F), while it supports the claim that one exists between V(sub T) and J. More importantly, the inferred relationship between V(sub T) and J is found to be more complicated than previously thought. In particular, it now appears that SSC-F events may be separable into two groups, based on the value of J: a low-J group (J less than 56), in which V(sub T) varies directly with J, and a high-J group (J greater than 56), in which V(sub T) varies inversely with J. As a whole, high-J events are associated with shocks of higher average transit velocity than those of low-J events, and SSC-F events with shocks of higher average transit velocity than those of SSC-EF events. Further, high-J events tend to be of greater X-ray class ( greater than M3), longer duration (greater then 80 min), and are more likely to be associated with type II/IV radio emission (9 of 12) than low-J events. They also tend to occur in magnetically complex (gamma/delta configuration) active regions (10 of 12) that are large in area extent (area greater than 445 millionths of a solar hemisphere) on the day of flaring (9 of 12). Of the 9 solar proton events that affected the Earth's environment that were found to be associated with SSC-F events, six were high-J events. Concerning 'geo-effectiveness', there appears to be no preferential relationship between the value of the J-parameter and the most negative value of the Dst geomagnetic index Dst(min) following the SSC, which is found to usually occur at 6-14 h after SSC onset (18 of 26) and which ranged in value from -1 to -249 (having a median value of about -75). Of the 26 SSC-F events, only 14 can be associated with a Dst(min) less than or equal to -75, and of these only 7 were high-J events. Of the 14 storm-related events (i.e. Dst(min) less than or equal to -75), three have previously been identified as being either 'magnetic clouds' or 'bidirectional flows', both manifestations of earthward-directed coronal mass ejections (CMEs). Superposed epoch analyses of selected solar wind parameters and Dst during the interval of storm-related SSC-F events demonstrate that geoeffective SSC-F events tend to be associated with solar wind flows that are faster, greater in magnetic field strength, and have a rotating field which has a strong southward component shortly after SSC onset, in comparison to SSC-F events that do not have Dst(min) less than or equal to 75. Therefore, it is inferred that geoeffective SSC-F events are probably fast earthward-directed CMEs. Although no single parameter is found that can serve as a predictor of high-skill level for determining the geoeffectiveness of an SSC-F event prior to its occurrence at Earth, one finds that knowledge of the flare's hemispheric location and appearance or lack of appearance of a two-ribbon structure is sufficient to correctly predict the geoeffectiveness of 20 out of 25 of the SSC-F events (80%). Surprisingly, the association or lack of association of metric type II/IV radio emission as a characteristic for determining the geoeffectiveness of the SSC-F events proved unfruitful, as did, to a lesser extent, the duration of the X-ray emission.

Wilson, Robert M.↗
NASA NTRS1994

Inverse Compton X-ray emission from the superluminal quasar 3C 345

In quasars with strong radio cores, the inverse-Compton process is believed to be the dominant source X-ray emission. For objects with parsec-scale radio jets, simple models have predicted that components in the jet emerging from the quasar nucleus generate the observed X-ray emission. We have tested this hypothesis in detail for the quasar 3C 345 using a ROSAT X-ray observation in 1990 July, together with quasi-simultaneous very long base interferometry (VLBI) imaging of the parsec-scale jet at five frequencies. The ROSAT spectrum is well fitted by a power law with index alpha = -0.96 +/- -0.13, consistent with models in which the X-ray emission results from inverse-Compton scattering of radio radiation from high-energy electrons in compact components. We show that the radio properties of brightest `knot' in the jet (`C5') can be fitted with a homogeneous sphere model whose parameters require bulk relativistic motion of the emitting material; otherwise the predicted model whose parameters require bulk relativistic motion of the emitting material; otherwise the predicted inverse-Compton X-ray emission exceeds the observed flux. If C5 is the origin of the X-ray emission, then it has a Doppler factor delta = 7.5((sup +3 sub -2)). If the nucleus or other components contribute to the X-ray emission, then this becomes a firm lower limit to delta. The inhomogeneous jet model of Koenigl is a good fit both to the barely resolved (less than 1 pc) flat-spectrum nucleus in the radio, and also to the ROSAT X-ray spectrum. The synchrotron and inverse-Compton emitting fluid moves down a narrow cone (opening angle 2 phi approximately 5 deg) nucleus relativistically, with delta approximately 4.6. Doppler factors for the nucleus and C5, derived from our ROSAT observation, provide evidence for bulk relativistic motion in the jet. By combining these constraints with well-known superluminal motion of jet components, we can deduce geometry. For epoch 1990.5 we infer the Lorentz factor gamma = 7.5 ((sup +1.0 sub -1.5)) and angle to the line of sight theta = 8((sup +2 deg sub -3 deg)) for H(sub 0) = 100 km/s/Mpc. These values are the most reliable yet derived using this method, because of the near-simultaneity of our X-ray and VLBI observations and the quality of the multifrequency of VLBI images and component radio spectra.

Unwin, S. C.↗
DOE OSTI2025

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

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

Approximate Bayesian inference↗
NASA NTRS1999

A Multipole Expansion Method for Analyzing Lightning Field Changes

Changes in the surface electric field are frequently used to infer the locations and magnitudes of lightning-caused changes in thundercloud charge distributions. The traditional procedure is to assume that the charges that are effectively deposited by the flash can be modeled either as a single point charge (the Q model) or a point dipole (the P model). The Q model has four unknown parameters and provides a good description of many cloud-to-ground (CG) flashes. The P model has six unknown parameters and describes many intracloud (IC) discharges. In this paper we introduce a new analysis method that assumes that the change in the cloud charge can be described by a truncated multipole expansion, i.e., there are both monopole and dipole terms in the unknown source distribution, and both terms are applied simultaneously. This method can be used to analyze CG flashes that are accompanied by large changes in the cloud dipole moment and complex IC discharges. If there is enough information content in the measurements, the model can also be generalized to include quadrupole and higher order terms. The parameters of the charge moments are determined using a dme-dimensional grid search in combination with a linear inversion, and because of this, local minima in the error function and the associated solution ambiguities are avoided. The multipole method has been tested on computer-simulated sources and on natural lightning at the NASA Kennedy Space Center and U.S. Air Force Eastern Range.

Koshak, William J.↗
NASA NTRS1998

A Multipole Expansion Method for Analyzing Lightning Field Changes

Changes in the surface electric field are frequently used to infer the locations and magnitudes of lightning-caused changes in thundercloud charge distributions. The traditional procedure is to assume that the charges that are effectively deposited by the flash can be modeled either as a single point charge (the Q-model) or a point dipole (the P-model). The Q-model has 4 unknown parameters and provides a good description of many cloud-to-ground (CG) flashes. The P-model has 6 unknown parameters and describes many intracloud (IC) discharges. In this paper, we introduce a new analysis method that assumes that the change in the cloud charge can be described by a truncated multipole expansion, i.e., there are both monopole and dipole terms in the unknown source distribution, and both terms are applied simultaneously. This method can be used to analyze CG flashes that are accompanied by large changes in the cloud dipole moment and complex IC discharges. If there is enough information content in the measurements, the model can also be generalized to include quadrupole and higher order terms. The parameters of the charge moments are determined using a 3-dimensional grid search in combination with a linear inversion, and because of this, local minima in the error function and the associated solution ambiguities are avoided. The multipole method has been tested on computer simulated sources and on natural lightning at the NASA Kennedy Space Center and USAF Eastern Range.

Koshak, William J.↗