SEARCH · Engineering Papers

Results for “Radiation Reliability”

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 1,477 records · Page 82

All sources NASA NTRS DOE OSTI

DOE OSTI2024

Data Imbalance, Uncertainty Quantification, and Transfer Learning in Data‐Driven Parameterizations: Lessons From the Emulation of Gravity Wave Momentum Transport in WACCM

Abstract Neural networks (NNs) are increasingly used for data‐driven subgrid‐scale parameterizations in weather and climate models. While NNs are powerful tools for learning complex non‐linear relationships from data, there are several challenges in using them for parameterizations. Three of these challenges are (a) data imbalance related to learning rare, often large‐amplitude, samples; (b) uncertainty quantification (UQ) of the predictions to provide an accuracy indicator; and (c) generalization to other climates, for example, those with different radiative forcings. Here, we examine the performance of methods for addressing these challenges using NN‐based emulators of the Whole Atmosphere Community Climate Model (WACCM) physics‐based gravity wave (GW) parameterizations as a test case. WACCM has complex, state‐of‐the‐art parameterizations for orography‐, convection‐, and front‐driven GWs. Convection‐ and orography‐driven GWs have significant data imbalance due to the absence of convection or orography in most grid points. We address data imbalance using resampling and/or weighted loss functions, enabling the successful emulation of parameterizations for all three sources. We demonstrate that three UQ methods (Bayesian NNs, variational auto‐encoders, and dropouts) provide ensemble spreads that correspond to accuracy during testing, offering criteria for identifying when an NN gives inaccurate predictions. Finally, we show that the accuracy of these NNs decreases for a warmer climate (4 × CO 2 ). However, their performance is significantly improved by applying transfer learning, for example, re‐training only one layer using ∼1% new data from the warmer climate. The findings of this study offer insights for developing reliable and generalizable data‐driven parameterizations for various processes, including (but not limited to) GWs.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS1986

AMTEC: High efficiency static conversion for space power

Future manned and unmanned space missions will require reliable, high efficiency energy conversion systems. For a manned Mars mission, power levels in the range of 10 to 100 kWe will be needed. The Alkali Metal Thermoelectric Converter (AMTEC) is a direct energy conversion technology with the potential to meet these needs. The AMTEC is a thermally regenerative electrochemical device that derives its operation from the sodium ion conducting properties of beta-alumina solid electrolyte (BASE). To date, an efficiency of 19%, area power density of 1 W/sq cm, and a lifetime of 10,000 hours at high temperature were demonstrated in laboratory devices. Systems studies show that projected AMTEC systems equal or surpass the performance of other static or dynamic systems in applications of 1 kWe-1 MWe. Thus, the laboratory experiments and applications studies conducted to date have shown that the AMTEC posseses great potential. In order to bring this technology to the stage where prototype units can be built and operated, several technical issues must be addressed. These include the need for long life, high power electrodes, minimization of radiative parasitic losses, and high temperature seals. In summary, the evidence shows that if AMTEC is developed, it can play a significant role in future space power applications.

Bankston, C. P.↗

NASA NTRS2001

Global Surface Solar Energy Anomalies Including El Nino and La Nina Years

This paper synthesizes past events in an attempt to define the general magnitude, duration, and location of large surface solar anomalies over the globe. Surface solar energy values are mostly a function of solar zenith angle, cloud conditions, column atmospheric water vapor, aerosols, and surface albedo. For this study, solar and meteorological parameters for the 10-yr period July 1983 through June 1993 are used. These data were generated as part of the Release 3 Surface meteorology and Solar Energy (SSE) activity under the NASA Earth Science Enterprise (ESE) effort. Release 3 SSE uses upgraded input data and methods relative to previous releases. Cloud conditions are based on recent NASA Version-D International Satellite Cloud Climatology Project (ISCCP) global satellite radiation and cloud data. Meteorological inputs are from Version-I Goddard Earth Observing System (GEOS) reanalysis data that uses both weather station and satellite information. Aerosol transmission for different regions and seasons are for an 'average' year based on historic solar energy data from over 1000 ground sites courtesy of Natural Resources Canada (NRCan). These data are input to a new Langley Parameterized Shortwave Algorithm (LPSA) that calculates surface albedo and surface solar energy. That algorithm is an upgraded version of the 'Staylor' algorithm. Calculations are performed for a 280X280 km equal-area grid system over the globe based on 3-hourly input data. A bi-linear interpolation process is used to estimate data output values on a 1 X 1 degree grid system over the globe. Maximum anomalies are examined relative to El Nino and La Nina events in the tropical Pacific Ocean. Maximum year-to-year anomalies over the globe are provided for a 10-year period. The data may assist in the design of systems with increased reliability. It may also allow for better planning for emergency assistance during some atypical events.

Whitlock, C. H.↗

NASA NTRS1998

SiC Technology

Silicon carbide (SiC)-based semiconductor electronic devices and circuits are presently being developed for use in high-temperature, high-power, and/or high-radiation conditions under which conventional semiconductors cannot adequately perform. Silicon carbide's ability to function under such extreme conditions is expected to enable significant improvements to a far-ranging variety of applications and systems. These range from greatly improved high-voltage switching [1- 4] for energy savings in public electric power distribution and electric motor drives to more powerful microwave electronics for radar and communications [5-7] to sensors and controls for cleaner-burning more fuel-efficient jet aircraft and automobile engines. In the particular area of power devices, theoretical appraisals have indicated that SiC power MOSFET's and diode rectifiers would operate over higher voltage and temperature ranges, have superior switching characteristics, and yet have die sizes nearly 20 times smaller than correspondingly rated silicon-based devices [8]. However, these tremendous theoretical advantages have yet to be realized in experimental SiC devices, primarily due to the fact that SiC's relatively immature crystal growth and device fabrication technologies are not yet sufficiently developed to the degree required for reliable incorporation into most electronic systems [9]. This chapter briefly surveys the SiC semiconductor electronics technology. In particular, the differences (both good and bad) between SiC electronics technology and well-known silicon VLSI technology are highlighted. Projected performance benefits of SiC electronics are highlighted for several large-scale applications. Key crystal growth and device-fabrication issues that presently limit the performance and capability of high temperature and/or high power SiC electronics are identified.

Neudeck, Philip G.↗

DOE OSTI2023

Impact of Wildfires on Solar Resource Availability in California in a Changing Climate

Wildfires can emit large amounts of atmospheric particulate matters and influence not only air quality but also availability of photovoltaic (PV) generation due to scattering and absorption of solar radiation. Under anthropogenic changing climate, wildfire activity is projected to increase over western North America due to drier and warmer climate, implying increasing impact on solar resource and larger uncertainty in solar generation especially in regions with faster PV penetration. This study focuses on quantifying the impact of wildfires on aerosol optical depth (AOD) and thus solar resource over California using National Solar Radiation Database (NSRDB), developed by the National Renewable Energy Laboratory (NREL). This assessment includes historical analysis and estimation of solar resource under wildfire scenarios (2020 wildfire and an enhanced wildfire scenario based on 2020 wildfires). Historical analysis for the period of July-October 2019/2020 (low/high fire activity period) shows that the averaged global horizontal irradiance (GHI) and direct normal irradiance (DNI) are reduced by about 30 and 90 W m-2 (5.6 and 15.3%), respectively, during high fire activity period. To create AOD dataset for enhanced fire scenario, 165% increase in burn area (in around 2050) is selected based on comprehensive literature review, which is further applied to the Fire INventory from NCAR (FINN) fire emission. WRF-Chem model with the enhanced FINN emissions is used to simulate and represent a preserving spatial distribution of burned areas and wildfire-emitted aerosols. Our initial analysis suggests that the"enhanced 2020 wildfire" AOD can increase by a factor of 1.3 - 2, which can significantly reduce solar irradiance and increase uncertainty in generation and reliability of power system in extreme wildfire events under a high solar penetration scenario. Overall, this study provides an estimate of the impacts of wildfires on solar resource to make informed decisions on reserve planning, generation scheduling, and reliability investments.

aerosol optical depth↗

DOE OSTI2025

Leveraging transfer learning and leaf spectroscopy for leaf trait prediction with broad spatial, species, and temporal applicability

Accurate and reliable prediction of leaf traits is crucial for understanding plant adaptations to environmental variation, monitoring terrestrial ecosystems, and enhancing comprehension of functional diversity and ecosystem functioning. Currently, various approaches (e.g., statistical, physical models) have been developed to estimate leaf traits through hyperspectral remote sensing and leaf spectroscopy. However, the absence of high-performing, transferable, and stable models across various domains of space, plant functional types (PFTs) and seasons hinder our ability to quantify and comprehend spatiotemporal variations in leaf traits. This study proposes robust and highly transferable models for better predicting leaf traits with hyperspectral reflectance. Initially, three datasets were assembled, pairing common leaf traits — chlorophyll (Chla+b), carotenoids (Ccar), leaf mass per area (LAM), equivalent water thickness (EWT) — with leaf spectra measurements collected across diverse geographic locations in the U.S. and Europe, PFTs, and seasons. Measurements were acquired using spectroradiometers (e.g., ASD FieldSpec 3/4/Pro and SVC HR-1024i) with integrating spheres, leaf clips, and contact probes. Here, we then developed transfer learning-based hybrid models that incorporated the domain knowledge of radiative transfer models (RTMs) through pretraining processes and were well-constrained by fine-tuning with field measurements. Through comparison with other state-of-the-art statistical models, including partial-least squares regression (PLSR) and Gaussian Process Regression (GPR), as well as pure physical models, we found that the proposed transfer learning models achieved better predictive performance and higher transferability. Specifically, compared to other statistical models and pure RTMs, the transfer learning model exhibited higher coefficient of determination (R 2 ) values with range of 0.01 to 0.79, lower normalized root mean square error (NRMSE) with range of 0.06 % to 33.25 % in model performance. Additionally, the models exhibited improved transferability, with higher R 2 values range from 0.04 to 0.32, lower NRMSE range from 0.08 % to 30.81 %. The findings underscore that transfer learning models through integrating domain knowledge from RTMs and limited observations, can harness the advantages of both RTMs and statistical models and serve as a promising approach for effectively predicting leaf traits.

59 BASIC BIOLOGICAL SCIENCES↗

NASA NTRS1990

VIPER project

The VIPER project has so far produced a formal specification of a 32 bit RISC microprocessor, an implementation of that chip in radiation-hard SOS technology, a partial proof of correctness of the implementation which is still being extended, and a large body of supporting software. The time has now come to consider what has been achieved and what directions should be pursued in the future. The most obvious lesson from the VIPER project was the time and effort needed to use formal methods properly. Most of the problems arose in the interfaces between different formalisms, e.g., between the (informal) English description and the HOL spec, between the block-level spec in HOL and the equivalent in ELLA needed by the low-level CAD tools. These interfaces need to be made rigorous or (better) eliminated. VIPER 1A (the latest chip) is designed to operate in pairs, to give protection against breakdowns in service as well as design faults. We have come to regard redundancy and formal design methods as complementary, the one to guard against normal component failures and the other to provide insurance against the risk of the common-cause failures which bedevil reliability predictions. Any future VIPER chips will certainly need improved performance to keep up with increasingly demanding applications. We have a prototype design (not yet specified formally) which includes 32 and 64 bit multiply, instruction pre-fetch, more efficient interface timing, and a new instruction to allow a quick response to peripheral requests. Work is under way to specify this device in MIRANDA, and then to refine the spec into a block-level design by top-down transformations. When the refinement is complete, a relatively simple proof checker should be able to demonstrate its correctness. This paper is presented in viewgraph form.

Kershaw, John↗

NASA NTRS2011

A Variational Method to Retrieve the Extinction Profile in Liquid Clouds Using Multiple Field-of-View Lidar

While liquid clouds playa very important role in the global radiation budget, it's been very difficult to remotely determine their internal cloud structure. Ordinary lidar instruments (similar to radars but using visible light pulses) receive strong signals from such clouds, but the information is limited to a thin layer near the cloud boundary. Multiple field-of-view (FOV) lidars offer some new hope as they are able to isolate photons that were scattered many times by cloud droplets and penetrated deep into a cloud before returning to the instrument. Their data contains new information on cloud structure, although the lack of fast simulation methods made it challenging to interpret the observations. This paper describes a fast new technique that can simulate multiple-FOV lidar signals and can even estimate the way the signals would change in response to changes in cloud properties-an ability that allows quick refinements in our initial guesses of cloud structure. Results for a hypothetical airborne three-FOV lidar suggest that this approach can help determine cloud structure for a deeper layer in clouds, and can reliably determine the optical thickness of even fairly thick liquid clouds. The algorithm is also applied to stratocumulus observations by the 8-FOV airborne "THOR" lidar. These tests demonstrate that the new method can determine the depth to which a lidar provides useful information on vertical cloud structure. This work opens the way to exploit data from spaceborne lidar and radar more rigorously than has been possible up to now.

Pounder, Nicola L.↗

DOE OSTI2020

Quasielastic lepton scattering and back-to-back nucleons in the short-time approximation

Understanding quasielastic electron- and neutrino-scattering from nuclei has taken on new urgency with current and planned neutrino oscillation experiments, and with electron scattering experiments measuring specific final states, such as those involving nucleon pairs in “back-to-back” configurations. Accurate many-body methods are available for calculating the response of light (A ≤ 12) nuclei to electromagnetic and weak probes, but they are computationally intensive and only applicable to the inclusive response. In the present work we introduce a novel approach, based on realistic models of nuclear interactions and currents, to evaluate the short-time (high-energy) inclusive and exclusive response of nuclei. The approach accounts reliably for crucial two-nucleon dynamics, including correlations and currents, and provides information on back-to-back nucleons observed in electron and neutrino scattering experiments. We demonstrate that in the quasielastic regime and at moderate momentum transfers both initial- and final-state correlations, and two-nucleon currents are important for a quantitatively successful description of the inclusive response and final state nucleons. Finally, the approach can be extended to include relativistic—kinematical and dynamical—effects, at least approximately in the two-nucleon sector, and to describe the response in the resonance-excitation region.

6 ≤ A ≤ 19↗

DOE OSTI2023

Local position-space two-nucleon potentials from leading to fourth order of chiral effective field theory

We present local, position-space chiral NN potentials through four orders of chiral effective field theory ranging from leading order (LO) to next-to-next-to-next-to-leading order (N 3 LO, fourth order) of the Δ-less version of the theory. The long-range parts of these potentials are fixed by the very accurate πN LECs as determined in the Roy-Steiner equations analysis. At the highest order (N 3 LO), the NN data below 190 MeV laboratory energy are reproduced with the respectable χ 2 /datum of 1.45. A comparison of the N 3 LO potential with the phenomenological Argonne v 18 (AV18) potential reveals substantial agreement between the two potentials in the intermediate range ruled by chiral symmetry, thus, providing a chiral underpinning for the phenomenological AV18 potential. Our chiral NN potentials may serve as a solid basis for systematic ab initio calculations of nuclear structure and reactions that allow for a comprehensive error analysis. In particular, the order by order development of the potentials will make possible a reliable determination of the truncation error at each order. Our new family of local position-space potentials differs from existing potentials of this kind by a weaker tensor force as reflected in relatively low D-state probabilities of the deuteron (P D ≲ 4.0 % for our N 3 LO potentials) and predictions for the triton binding energy above 8.00 MeV (from two-body forces alone). As a consequence, our potentials may lead to different predictions when applied to light and intermediate-mass nuclei in ab initio calculations and, potentially, help solve some of the outstanding problems in microscopic nuclear structure.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2020

A Novel Machine Learning Approach to Disentangle Multitemperature Regions in Galaxy Clusters

The hot intracluster medium (ICM) surrounding the heart of galaxy clusters is a complex medium that comprises various emitting components. Although previous studies of nearby galaxy clusters, such as the Perseus, the Coma, or the Virgo cluster, have demonstrated the need for multiple thermal components when spectroscopically fitting the ICM’s X-ray emission, no systematic methodology for calculating the number of underlying components currently exists. In turn, underestimating or overestimating the number of components can cause systematic errors in the emission parameter estimations. In this paper, we present a novel approach to determining the number of components using an amalgam of machine learning techniques. Synthetic spectra containing a various number of underlying thermal components were created using well-established tools available from the Chandra X-ray Observatory. The dimensions of the training set was initially reduced using principal component analysis and then categorized based on the number of underlying components using a random forest classifier. Our trained and tested algorithm was subsequently applied to Chandra X-ray observations of the Perseus cluster. Our results demonstrate that machine learning techniques can efficiently and reliably estimate the number of underlying thermal components in the spectra of galaxy clusters, regardless of the thermal model (MEKAL versus APEC). We also confirm that the core of the Perseus cluster contains a mix of differing underlying thermal components. We emphasize that although this methodology was trained and applied on Chandra X-ray observations, it is readily portable to other current (e.g., XMM-Newton, eROSITA) and upcoming (e.g., Athena, Lynx, XRISM) X-ray telescopes. The code is publicly available at https://github.com/XtraAstronomy/Pumpkin.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS2000

A Deep Space Power System Option Based on Synergistic Power Conversion Technologies

Deep space science missions have typically used radioisotope thermoelectric generator (RTG) power systems. The RTG power system has proven itself to be a rugged and highly reliable power system over many missions, however the thermal-to-electric conversion technology used was approximately 5% efficient. While the relatively low efficiency has some benefits in terms of system integration, there are compelling reasons why a more efficient conversion system should be pursued. The cost savings alone that are available as a result of the reduced isotope inventory are significant. The Advanced Radioisotope Power System (ARPS) project was established to fulfill this goal. Although it was not part of the ARPS project, Stirling conversion technology is being demonstrated with a low level of funding by both NASA and DOE. A power system with Stirling convertors. although intended for use with an isotope heat source. can be combined with other advanced technologies to provide a novel power system for deep space missions. An inflatable primary concentrator would be used in combination with a refractive secondary concentrator (RSC) as the heat source to power the system. The inflatable technology as a structure has made great progress for a variety of potential applications such as communications reflectors, radiators and solar arrays. The RSC has been pursued for use in solar thermal propulsion applications, and it's unique properties allow some advantageous system trades to be made. The power system proposed would completely eliminate the isotope heat source and could potentially provide power for science missions to planets as distant as Uranus. This paper will present the background and developmental status of the technologies and will then describe the power system being proposed.

Schreiber, Jeffrey G.↗

DOE OSTI2023

Revealing the strength of three-nucleon interactions with the proposed Einstein Telescope

Nuclear systems, ranging from atomic nuclei to dense matter probed in neutron stars, are governed by strong interactions. Three-nucleon forces have been found to be a crucial ingredient for the reliable description of these systems. Here, in this work, we explore how astrophysical data on neutron stars and their mergers from current and next-generation observatories will enable us to distinguish nuclear Hamiltonians. In particular, we investigate two different nuclear Hamiltonians that have been adjusted to reproduce two-nucleon scattering data and properties of light nuclei, but differ in the three-nucleon interactions among neutrons. We find that no significant constraints can be obtained from current data, but that the proposed Einstein Telescope could provide strong evidence to distinguish among these Hamiltonians.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2021

Constructing angular distributions of neutrinos in core-collapse supernovae from zeroth and first moments calibrated by full Boltzmann neutrino transport

Two-moment neutrino transport methods have been widely used for developing theoretical models of core-collapse supernovae (CCSN), since they substantially reduce the computational burden inherent in the multidimensional neutrino-radiation hydrodynamical simulations. The approximation, however, comes at a price; the detailed structure of angular distribution of neutrinos is sacrificed, that is the main drawback of this approach. In this paper, we develop a novel method by which to construct angular distributions of neutrinos from the zeroth and first angular moments. In our method, the angular distribution is expressed with two quadratic functions of the neutrino angle in a piecewise fashion. We determine the best parameters in the fitting function by comparing to the neutrino data in a spherically symmetric CCSN model with full Boltzmann neutrino transport. We demonstrate the capability of our method by using our recent 2D CCSN model. In this work, we find that the essential features of the angular distributions can be well reconstructed, whereas the angular distributions of incoming neutrinos tend to have large errors that increase with flux factor ($\textit{κ}$). This issue originates from the insensitiveness of incoming neutrinos to $\textit{κ}$, that is an intrinsic limitation in moment methods. Based on the results of the demonstration, we assess the reliability of electron- neutrinos lepton number (ELN)-crossing searches with the two-moment neutrino transport. This analysis is complementary to our [L. Johns and H. Nagakura arXiv:2104.04106] that scrutinizes the limitation of crossing searches with a few moments. We find that the systematic errors of angular distributions for incoming neutrinos lead to misjudgements of the crossing at $\textit{κ}$ ≳ 0.5. This casts doubt on the results of ELN-crossing searches based on two-moment methods in some previous studies.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1994

Numerical aperture limits on efficient ball lens coupling of laser diodes to single-mode fibers with defocus to balance spherical aberration

The potential capabilities and limitations of single ball lenses for coupling laser diode radiation to single-mode optical fibers have been analyzed; parameters important to optical communications were specifically considered. These parameters included coupling efficiency, effective numerical apertures, lens radius, lens refractive index, wavelength, magnification in imaging the laser diode on the fiber, and defocus to counterbalance spherical aberration of the lens. Limiting numerical apertures in object and image space were determined under the constraint that the lens perform to the Rayleigh criterion of 0.25-wavelength (Strehl ratio = 0.80). The spherical aberration-defocus balance to provide an optical path difference of 0.25 wavelength units was shown to define a constant coupling efficiency (i.e., 0.56). The relative numerical aperture capabilities of the ball lens were determined for a set of wavelengths and associated fiber-core diameters of particular interest for single-mode fiber-optic communication. The results support general continuing efforts in the optical fiber communications industry to improve coupling links within such systems with emphasis on manufacturing simplicity, system packaging flexibility, relaxation of assembly alignment tolerances, cost reduction of opto-electronic components and long term reliability and stability.

Wilson, R. Gale↗

NASA NTRSDate not supplied

Enabling Reliable, Fault-Tolerant Autonomous Lunar Habitats with High-Performance Spaceflight Computing

The lunar surface presents unfavorable constraints and harsh living conditions. To address these challenges, autonomous habitats will require complex integrated systems that combine advanced software, high-performance hardware, and cutting-edge sensors to ensure sustainability, safety, and operational efficiency. Consequently, maintaining a sustainable presence on the Moon requires reliable infrastructure and efficient development, precise monitoring, and utilization of resources within a lunar installation. These elements are essential not only to ensure that lunar settlement can be long-term, self-sustaining, and resource-efficient, but also to serve as a foundation for future missions and eventual human habitation on Mars. Humans are not native to the Moon; therefore, our survival and ability to thrive will depend on autonomous systems that can foster safety and resilience through high-availability architectures, graceful degradation, and highly fault-tolerant spaceflight hardware capable of continuing operation during failures. This requires advanced human-rated distributed systems architectures with specialized electronics, scalable capabilities, and an integrated design approach. Unlike current practices focused on short-term missions and regularly maintained components, permanent lunar compute systems must be designed for extended operations beyond mission durations. This paper explores the necessity of transitioning toward fault- tolerant, highly autonomous hardware systems designed for multi-year missions. It also identifies critical subsystems that require high levels of autonomy, supported by radiation-hardened processors and extreme thermal loads, which are essential to mitigate long-term degradation and ensure sustainable lunar habitation. Finally, the paper aligns with NASA’s identified Civil Space Shortfalls, particularly in high-performance onboard computing, advanced data acquisition, extreme-environment avionics, radiation monitoring and countermeasures, and autonomous health management. It proposes NASA’s new High-Performance Spaceflight Computing (HPSC) processor as a turnkey solution, delivering 100 times the performance-per-watt of legacy rad-hard CPUs and enabling onboard AI, edge computing, and fault-tolerant features essential for sustained lunar autonomy and beyond.

Sarkis S Mikaelian↗

DOE OSTI2025

Navigating the Noise: Bringing Clarity to ML Parameterization Design With O $\boldsymbol{\mathcal{O}}$(100) Ensembles

Abstract Machine‐learning (ML) parameterizations of subgrid processes (here of turbulence, convection, and radiation) may one day replace conventional parameterizations by emulating high‐resolution physics without the cost of explicit simulation. However, uncertainty about the relationship between offline and online performance (i.e., when integrated with a large‐scale general circulation model) hinders their development. Much of this uncertainty stems from limited sampling of the noisy, emergent effects of upstream ML design decisions on downstream online hybrid simulation. Our work rectifies the sampling issue via the construction of a semi‐automated, end‐to‐end pipeline for size ensembles of hybrid simulations, revealing important nuances in how systematic reductions in offline error manifest in changes to online error and online stability. For example, removing dropout and switching from a Mean Squared Error to a Mean Absolute Error loss both reduce offline error, but they have opposite effects on online error and online stability. Other design decisions, like incorporating memory, converting moisture input from specific humidity to relative humidity, using batch normalization, and training on multiple climates do not come with any such compromises. Finally, we show that ensemble sizes of may be necessary to reliably detect causally relevant differences online. By enabling rapid online experimentation at scale, we can empirically settle debates regarding subgrid ML parameterization design that would have otherwise remained unresolved in the noise.

Lin, Jerry [Department of Earth System Sciences Un↗

DOE OSTI2021

Plutonium in Manhattan Project workers: Using autopsy data to evaluate organ content and dose estimates based on urine bioassay with implications for radiation epidemiology

Purpose Radiation dose estimates in epidemiology typically rely on intake predictions based on urine bioassay measurements. The purpose of this article is to compare the conventional dosimetric estimates for radiation epidemiology with the estimates based on additional post-mortem tissue radiochemical analysis results. Methods The comparison was performed on a unique group of 11 former Manhattan Project nuclear workers, who worked with plutonium in the 1940s, and voluntarily donated their bodies to the United States Transuranium and Uranium Registries. Results Post-mortem organ activities were predicted using different sets of urine data and compared to measured activities. Use of urinalysis data collected during the exposure periods overestimated the systemic (liver+skeleton) deposition of 239 Pu by 155±134%, while the average bias from using post-exposure urinalyses was –4±50%. Committed effective doses estimated using early urine data differed from the best estimate by, on average, 196±193%; inclusion of follow-up urine measurements in analyses decreased the mean bias to 0.6±36.3%. Cumulative absorbed doses for the liver, red marrow, bone surface, and brain were calculated for the actual commitment period. Conclusion On average, post-exposure urine bioassay results were in good agreement with post-mortem tissue analyses and were more reliable than results of urine bioassays collected during the exposure.

60 APPLIED LIFE SCIENCES↗