Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “small sample size”

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 1,225 records · Page 68
All sourcesNASA NTRSDOE OSTI
DOE OSTI2015

Characterization of plutonium activities in soil - 15429

Plutonium (Pu) contamination in soils at the Nevada National Security Site (NNSS) is characterized using Am-241 activity results from gamma spectroscopy (gamma Am-241) analysis. This is done by inferring the Pu activities using Pu to Am-241 ratios developed from Pu and Am-241 results from alpha spectroscopy isotopic analyses (isotopic Pu and isotopic Am-241). Although the isotopic Pu and Am-241 methods can accurately measure isotopic activities within the sampled material, the amount of soil analyzed is small (typically 1 gram [g] of soil). If there is heterogeneity of particle distribution within the contaminated release site soil, isotopic methods may provide results that are not representative of the contaminated site. The degree of potential heterogeneity needs to be considered before using isotopic analyses to characterize a Pu-contaminated site. Results from various Pu-contaminated sites at the NNSS are presented that demonstrate significant heterogeneity. A larger sample size (∼1,600 g) within a Marinelli container is used for the gamma spectroscopy analysis. This greatly reduces the impact of heterogeneously distributed discrete particles and provides results that are more representative of true contaminant activities at the release site. However, this method cannot provide Pu activities and has the potential to provide less accurate results due to the haphazard location of contaminant particles within the Marinelli container at the time of measurement. The distance of particles from the detector would result in some differential self-absorption of the emissions from radioactive particles. As the magnitude of this problem was not previously understood, a study was conducted to evaluate the variability in Am-241 results due to self-absorption from particle position. Results from this study are shown that demonstrate that this effect provides minimal variability in replicate Am-241 measurements. The method used at the NNSS to characterize Pu activities in soil takes advantage of the accuracy of the alpha spectroscopy isotopic analyses to establish Pu to Am-241 ratios and the more representative gamma spectroscopy results for Am-241 to provide a reliable characterization of Pu activities at contaminant release sites. (authors)

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS2004

New Material for Surface-Enhanced Raman Spectroscopy

A chemical method of synthesis and application of coating materials that are especially suitable for surface-enhanced Raman spectroscopy (SERS) has been developed. The purpose of this development is to facilitate the utilization of the inherently high sensitivity of SERS to detect chemicals of interest (analytes) in trace amounts, without need for lengthy sample preparation. Up to now, the use of SERS has not become routine because the methods available have not been able to reproduce sampling conditions and provide quantitative measurements. In contrast, the coating materials of the present method enable analysis with minimum preparation of samples, and SERS measurements made using these materials are reproducible and reversible. Moreover, unlike in methods investigated in prior efforts to implement SERS, sampling is not restricted to such specific environments as electrolytes or specific solvents. The coating materials of this method are porous glasses, formed in sol-gel processes, that contain small particles of gold or silver metal. Materials of this type can be applied to the sample-contact surfaces of a variety of sampling and sensing devices, including glass slides, glass vials, fiber-optic probes, and glass tubes. Glass vials with their insides coated according to this method are particularly convenient for SERS to detect trace chemicals in solutions: One simply puts a sample solution containing the analyte(s) into a vial, then puts the vial into a Raman spectrometer for analysis. The chemical ingredients and the physical conditions of the sol-gel process have been selected so that the porous glass formed incorporates particles of the desired metal with size(s) to match the wavelength(s) of the SERS excitation laser in order to optimize the generation of surface plasmons. The ingredients and processing conditions have further been chosen to tailor the porosity and polarity of the glass to optimize the sample flow and the interaction between the analyte(s) and the plasmon field that generates Raman photons. The porous silica network of a sol-gel glass creates a unique environment for stabilizing SERS-active metal particles. Relative to other material structures that could be considered for SERS, the porous silica network offers higher specific surface area and thus greater interaction between analyte molecules and metal particles. Efforts to perform SERS measurements with the help of sampling devices coated by this method have been successful. In tests, numerous organic and inorganic chemicals were analyzed in several solvents, including water. The results of the tests indicate that the SERS measurements were reproducible within 10 percent and linear over five orders of magnitude. One measure of the limits of detectability of chemicals in these tests was found to be a concentration of 300 parts per billion. Further development may eventually make it possible to realize the full potential sensitivity of SERS for detecting some analytes in quantities as small as a single molecule.

Farquharson, Stuart↗
DOE OSTI2016

Small Angle X-ray Scattering for Nanoparticle Research

X-ray scattering is a structural characterization tool that has impacted diverse fields of study. It is unique in its ability to examine materials in real time and under realistic sample environments, enabling researchers to understand morphology at nanometer and ångström length scales using complementary small and wide angle X-ray scattering (SAXS, WAXS), respectively. Herein, we focus on the use of SAXS to examine nanoscale particulate systems. We provide a theoretical foundation for X-ray scattering, considering both form factor and structure factor, as well as the use of correlation functions, which may be used to determine a particle’s size, size distribution, shape, and organization into hierarchal structures. The theory is expanded upon with contemporary use cases. Both transmission and reflection (grazing incidence) geometries are addressed, as well the combination of SAXS with other X-ray and non-X ray characterization tools. Furthermore, we conclude with an examination of several key areas of research where X-rays scattering has played a pivotal role, including in situ nanoparticle synthesis, nanoparticle assembly, and in operando studies of catalysts and energy storage materials. Throughout this review we highlight the unique capabilities of X-ray scattering for structural characterization of materials in their native environment.

36 MATERIALS SCIENCE↗
DOE OSTI2011

SNLS3: CONSTRAINTS ON DARK ENERGY COMBINING THE SUPERNOVA LEGACY SURVEY THREE-YEAR DATA WITH OTHER PROBES

We present observational constraints on the nature of dark energy using the Supernova Legacy Survey three-year sample (SNLS3) of Guy et al. and Conley et al. We use the 472 Type Ia supernovae (SNe Ia) in this sample, accounting for recently discovered correlations between SN Ia luminosity and host galaxy properties, and include the effects of all identified systematic uncertainties directly in the cosmological fits. Combining the SNLS3 data with the full WMAP7 power spectrum, the Sloan Digital Sky Survey luminous red galaxy power spectrum, and a prior on the Hubble constant H{sub 0} from SHOES, in a flat universe we find {Omega}{sub m} = 0.269 {+-} 0.015 and w = -1.061{sup +0.069}{sub -} 0{sub .068} (where the uncertainties include all statistical and SN Ia systematic errors)-a 6.5% measure of the dark energy equation-of-state parameter w. The statistical and systematic uncertainties are approximately equal, with the systematic uncertainties dominated by the photometric calibration of the SN Ia fluxes-without these calibration effects, systematics contribute only a {approx}2% error in w. When relaxing the assumption of flatness, we find {Omega}{sub m} = 0.271 {+-} 0.015, {Omega}{sub k} = -0.002 {+-} 0.006, and w = -1.069{sup +0.091}{sub -0.092}. Parameterizing the time evolution of w as w(a) = w{sub 0} + w{sub a} (1 - a) gives w{sub 0} = -0.905 {+-} 0.196, w{sub a} = -0.984{sup +1.094}{sub -1.097} in a flat universe. All of our results are consistent with a flat, w = -1 universe. The size of the SNLS3 sample allows various tests to be performed with the SNe segregated according to their light curve and host galaxy properties. We find that the cosmological constraints derived from these different subsamples are consistent. There is evidence that the coefficient, {beta}, relating SN Ia luminosity and color, varies with host parameters at >4{sigma} significance (in addition to the known SN luminosity-host relation); however, this has only a small effect on the cosmological results and is currently a subdominant systematic.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2020

Deep transfer learning for star cluster classification: I. application to the PHANGS– HST survey

ABSTRACT We present the results of a proof-of-concept experiment that demonstrates that deep learning can successfully be used for production-scale classification of compact star clusters detected in Hubble Space Telescope(HST) ultraviolet-optical imaging of nearby spiral galaxies ($D\lesssim 20\, \textrm{Mpc}$) in the Physics at High Angular Resolution in Nearby GalaxieS (PHANGS)–HST survey. Given the relatively small nature of existing, human-labelled star cluster samples, we transfer the knowledge of state-of-the-art neural network models for real-object recognition to classify star clusters candidates into four morphological classes. We perform a series of experiments to determine the dependence of classification performance on neural network architecture (ResNet18 and VGG19-BN), training data sets curated by either a single expert or three astronomers, and the size of the images used for training. We find that the overall classification accuracies are not significantly affected by these choices. The networks are used to classify star cluster candidates in the PHANGS–HST galaxy NGC 1559, which was not included in the training samples. The resulting prediction accuracies are 70 per cent, 40 per cent, 40–50 per cent, and 50–70 per cent for class 1, 2, 3 star clusters, and class 4 non-clusters, respectively. This performance is competitive with consistency achieved in previously published human and automated quantitative classification of star cluster candidate samples (70–80 per cent, 40–50 per cent, 40–50 per cent, and 60–70 per cent). The methods introduced herein lay the foundations to automate classification for star clusters at scale, and exhibit the need to prepare a standardized data set of human-labelled star cluster classifications, agreed upon by a full range of experts in the field, to further improve the performance of the networks introduced in this study.

Wei, Wei↗
DOE OSTI2012

Using a Simple Binomial Model to Assess Improvement in Predictive Capability: Sequential Bayesian Inference, Hypothesis Testing, and Power Analysis

We present a Bayesian statistical methodology for identifying improvement in predictive simulations, including an analysis of the number of (presumably expensive) simulations that will need to be made in order to establish with a given level of confidence that an improvement has been observed. Our analysis assumes the ability to predict (or postdict) the same experiments with legacy and new simulation codes and uses a simple binomial model for the probability, {theta}, that, in an experiment chosen at random, the new code will provide a better prediction than the old. This model makes it possible to do statistical analysis with an absolute minimum of assumptions about the statistics of the quantities involved, at the price of discarding some potentially important information in the data. In particular, the analysis depends only on whether or not the new code predicts better than the old in any given experiment, and not on the magnitude of the improvement. We show how the posterior distribution for {theta} may be used, in a kind of Bayesian hypothesis testing, both to decide if an improvement has been observed and to quantify our confidence in that decision. We quantify the predictive probability that should be assigned, prior to taking any data, to the possibility of achieving a given level of confidence, as a function of sample size. We show how this predictive probability depends on the true value of {theta} and, in particular, how there will always be a region around {theta} = 1/2 where it is highly improbable that we will be able to identify an improvement in predictive capability, although the width of this region will shrink to zero as the sample size goes to infinity. We show how the posterior standard deviation may be used, as a kind of 'plan B metric' in the case that the analysis shows that {theta} is close to 1/2 and argue that such a plan B should generally be part of hypothesis testing. All the analysis presented in the paper is done with a general beta-function prior for {theta}, enabling sequential analysis in which a small number of new simulations may be done and the resulting posterior for {theta} used as a prior to inform the next stage of power analysis.

42 ENGINEERING↗
DOE OSTI2021

On the Correlation between Atomic Gas and Bars in Galaxies

We analyze the correlation between properties of large-scale bars and atomic gas content of galaxies to explore the role of H i gas on bar evolution in galaxies. We show that the absolute bar size depends strongly on total stellar mass of galaxies and does not change significantly with H i gas fraction at fixed stellar mass. Furthermore, the physical size of the bar is small and nearly constant in high H i gas fraction and low-mass galaxies and becomes larger with increasing galactic stellar mass in low gas fraction galaxies. When the stellar masses are fixed, the relative bar length normalized to the disk shows a decrease with increasing H i gas fraction due to the larger disks in gas-richer galaxies. We measure the gas deficiency of the samples and find that the gas-rich galaxies have longer and stronger bars compared with the H i gas-deficient galaxies at fixed stellar mass, especially for the massive ones. When splitting the samples into star-forming and quiescent subgroups, the star-forming galaxies tend to have longer bars than the quiescent ones at fixed stellar mass and gas deficiency. In addition, our results suggest two different types of bar properties, in which the bars in gas-rich galaxies grow longer but retain similar axial ratio over time, while they grow longer and fatter over time in gas-poor galaxies.

74 ATOMIC AND MOLECULAR PHYSICS↗
NASA NTRS1983

Effect of the eruption of El Chichon stratospheric aerosol size and composition

Dominant effects of the El Chichon eruption on stratospheric aerosols at 19.8 to 20.7 km are: (1) vapor depositional growth of the small-aerosol (background) mode; (2) development of a large-particle mode by sedimentation from the highest altitudes in the cloud; (3) a change in the large-particle mode from sulfate-coated silicates to sulfate aerosols, some with silicate cores; (4) a 100-fold increase in sulfate mass in the large particle mode. Terminal velocities of large silicate particles, maximum r = 2.3 micron, sampled 1 month after eruption, and calibrated with the aid of lidar data, indicate initial injection to 26 to 27 km. Smaller velocities of sulfate aerosols, median r = 0.5 micron, are compatible with major growth in 2 to 3 months at 27 to 28 km. Aerosol settling accounts for the descent of the main lidar return to 26.5 km in August and to 20 to 21 km in December.

Oberbeck, V. R.↗
NASA NTRS1993

Feasibility of Electrostriction for Measurement of the Critical Temperature in the Critical Viscosity Experiment

The large compressibility of a pure fluid near its critical point is a potential means of measuring the critical temperature. At the suggestion of Prof. Ferrell we observed electrostriction in xenon by applying a large voltage across a small, open capacitor immersed in the sample and then measuring the capacitance change caused by the subsequent increase in the xenon's local density. This scheme was attractive because the compressibility is a large, well-understood effect and because we were already familiar with low-voltage capacitance measurements. We found qualitative agreement with our initial expectations based on the fluid's compressibility. However, we also found an additional effect comparable in size to the expected electrostriction but proportional to the applied voltage. We have no explanation for this effect. Thus, given the Science Panel's recommendation against investing extensive effort in this direction, we are abandoning further development of an alternate means for measuring the critical temperature.

Berg, Robert F.↗
NASA NTRS2009

On the Importance of Small Ice Crystals in Tropical Anvil Cirrus

In situ measurements of ice crystal concentrations and sizes made with aircraft instrumentation over the past two decades have often indicated the presence of numerous relatively small (< 50 m diameter) crystals in cirrus clouds. Further, these measurements frequently indicate that small crystals account for a large fraction of the extinction in cirrus clouds. The fact that the instruments used to make these measurements, such as the Forward Scattering Spectrometer Probe (FSSP) and the Cloud Aerosol Spectrometer (CAS), ingest ice crystals into the sample volume through inlets has led to suspicion that the indications of numerous small ]crystals could be artifacts of large ]crystal shattering on the instrument inlets. We present new aircraft measurements in anvil cirrus sampled during the Tropical Composition, Cloud, and Climate Coupling (TC4) campaign with the 2 ] Dimensional Stereo (2D ]S) probe, which detects particles as small as 10 m. The 2D ]S has detector "arms" instead of an inlet tube. Since the 2D ]S probe surfaces are much further from the sample volume than is the case for the instruments with inlets, it is expected that 2D ]S will be less susceptible to shattering artifacts. In addition, particle inter ]arrival times are used to identify and remove shattering artifacts that occur even with the 2D ]S probe. The number of shattering artifacts identified by the 2D ]S interarrival time analysis ranges from a negligible contribution to an order of magnitude or more enhancement in apparent ice concentration over the natural ice concentration, depending on the abundance of large crystals and the natural small ]crystal concentration. The 2D ]S measurements in tropical anvil cirrus suggest that natural small ]crystal concentrations are typically one to two orders of magnitude lower than those inferred from CAS. The strong correlation between the CAS/2D ]S ratio of small ]crystal concentrations and large ]crystal concentration suggests that the discrepancy is likely caused by shattering of large crystals on the CAS inlet. We argue that past measurements with CAS in cirrus with large crystals present may contain errors due to crystal shattering, and past conclusions derived from these measurements may need to be revisited. Further, we present correlations between CAS spurious concentration and 2D ]S large ]crystal mass from spatially uniform anvil cirrus sampling periods as an approximate guide for estimating quantitative impact of large ]crystal shattering on CAS concentrations in previous datasets. We use radiative transfer calculations to demonstrate that in the maritime anvil cirrus sampled during TC4, small crystals indicated by 2D ]S contribute relatively little cloud extinction, radiative forcing, or radiative heating in the anvils, regardless of anvil age or vertical location in the clouds. While 2D ]S ice concentrations in fresh anvil cirrus may often exceed 1 cm.3, and are observed to exceed 10 cm.3 in turrets, they are typically ~0.1 cm.3 and rarely exceed 1 cm.3 (<1.4% of the time) in aged anvil cirrus. We hypothesize that isolated occurrences of higher ice concentrations in aged anvil cirrus may be caused by ice nucleation driven by either small ]scale convection or gravity waves. It appears that the numerous small crystals detrained from convective updrafts do not persist in the anvil cirrus sampled during TC ]4.

Jensen, E. J.↗
NASA NTRS2012

Finding Near-Earth Asteroid (NEA) Destinations for Human Exploration: Implications for Astrobiology

The current number of known potential NEA targets for HSF is limited to those objects whose orbital characteristics are similar to that of the Earth. This is due to the projected capabilities of the exploration systems currently under consideration and development at NASA. However, NEAs with such orbital characteristics often have viewing geometries that place them at low solar elongations and thus are difficult to detect from the vicinity of Earth. While ongoing ground-based surveys and data archives maintained by the NEO Program Observation Program Office and the Minor Planet Center (MPC) have provided a solid basis upon which to build, a more complete catalog of the NEO population is required to inform a robust and sustainable HSF exploration program. Since all the present NEO observing assets are currently confined to the vicinity of the Earth, additional effort must be made to provide capabilities for detection of additional HSF targets via assets beyond Earth orbit. A space-based NEO survey telescope located beyond the vicinity of the Earth, has considerable implications for planetary science and astrobiology. Such a telescope will provide foundational knowledge of our Solar System small body population and detect targets of interest for both the HSF and scientific communities. Data from this asset will yield basic characterization data on the NEOs observed (i.e., albedo, size determination, potential for volatiles and organics, etc.) and help down select targets for future HSF missions. Ideally, the most attractive targets from both HSF and astrobiology perspectives are those NEAs that may contain organic and volatile materials, and which could be effectively sampled at a variety of locations and depths. Presented here is an overview of four space-based survey concepts; any one of which after just a few years of operation will discover many highly accessible NEO targets suitable for robotic and human exploration. Such a space-based survey mission will reveal incredible returns for several disciplines including: exploration, in situ resource utilization, planetary defense, and science. Of particular, interest to the scientific

Landis, Rob↗
NASA NTRS2017

Itokawa's Opposition Surge seen by Hayabusa/AMICA

Using images acquired by the Hayabusa/AMICA instrument, along with Lederer et al.'s (2008) ground-based observations, we re-examine Itokawa's disk-integrated phase curve. The AMICA images provide critical opposition measurements (between 0.7deg - 9.3deg phase at 540 nm). Using Hapke's model (2012), we fit the updated phase curves at 5 different wavelengths. Preliminary modeling results show a range of porosity values commensurate with those in the literature (Ostro et al. 2004, Gundlach and Blum, 2012, Kiuchi and Nakamura 2014) based on an impact-generated grain size distribution function and grain size range evaluations from the AMICA data (Yano et al. 2006). This wide range on a global porosity is indicative of a highly variable porosity across the surface. The derived transport mean free path and the generally forward scattering nature of the global regolith are indicative of scattering centers (such as cracks, bubbles, and inclusions) that are small compared to the observational wavelengths. The derived regolith properties are compared with the imaging and sample analysis results, providing a test of the predictive capabilities of global disk-integrated measurements. This work suggests that the sub-pixel grain information could be extracted from the photometry, especially around opposition.

Dominque, D.↗
NASA NTRS2023

NEID Reveals That the Young Warm Neptune TOI-2076 b Has a Low Obliquity

TOI-2076 b is a sub-Neptune-sized planet (R = 2.39 ± 0.10 Rꚛ) that transits a young (204 ± 50 MYr) bright (V = 9.2) K-dwarf hosting a system of three transiting planets. Using spectroscopic observations obtained with the NEID spectrograph on the WIYN 3.5 m Telescope, we model the Rossiter–McLaughlin effect of TOI-2076 b, and derive a sky-projected obliquity of λ = -3 +16o −15 . Using the size of the star (R = 0.775 ± 0.015 Re), and the stellar rotation period (P rot = 7.27 ± 0.23 days), we estimate an obliquity of 𝜓=18 +10o −9 (𝜓 < 34° at 95% confidence), demonstrating that TOI-2076 b is in a well-aligned orbit. Simultaneous diffuser-assisted photometry from the 3.5 m telescope at Apache Point Observatory rules out flares during the transit. TOI-2076 b joins a small but growing sample of young planets in compact multi-planet systems with well-aligned orbits, and is the fourth planet with an age ≲300 Myr in a multi-transiting system with an obliquity measurement. The low obliquity of TOI-2076 b and the presence of transit timing variations in the system suggest the TOI-2076 system likely formed via convergent disk migration in an initially well-aligned disk.

Robert C Frazier↗
NASA NTRS2016

The Sustainability of Habitability on Terrestrial Planets: Insights, Questions, and Needed Measurements from Mars for Understanding the Evolution of Earth-Like Worlds

What allows a planet to be both within a potentially habitable zone and sustain habitability over long geologic time? With the advent of exoplanetary astronomy and the ongoing discovery of terrestrial-type planets around other stars, our own solar system becomes a key testing ground for ideas about what factors control planetary evolution. Mars provides the solar systems longest record of the interplay of the physical and chemical processes relevant to habitability on an accessible rocky planet with an atmosphere and hydrosphere. Here we review current understanding and update the timeline of key processes in early Mars history. We then draw on knowledge of exoplanets and the other solar system terrestrial planets to identify six broad questions of high importance to the development and sustaining of habitability (unprioritized): (1) Is small planetary size fatal? (2) How do magnetic fields influence atmospheric evolution? (3) To what extent does starting composition dictate subsequent evolution, including redox processes and the availability of water and organics? (4) Does early impact bombardment have a net deleterious or beneficial influence? (5) How do planetary climates respond to stellar evolution, e.g., sustaining early liquid water in spite of a faint young Sun? (6) How important are the timescales of climate forcing and their dynamical drivers? Finally, we suggest crucial types of Mars measurements (unprioritized) to address these questions: (1) in situ petrology at multiple units/sites; (2) continued quantification of volatile reservoirs and new isotopic measurements of H, C, N, O, S, Cl, and noble gases in rocks that sample multiple stratigraphic sections; (3) radiometric age dating of units in stratigraphic sections and from key volcanic and impact units; (4) higher-resolution measurements of heat flux, subsurface structure, and magnetic field anomalies coupled with absolute age dating. Understanding the evolution of early Mars will feed forward to understanding the factors driving the divergent evolutionary paths of the Earth, Venus, and thousands of small rocky extra solar planets yet to be discovered.

rocky exoplanets↗
DOE OSTI2018

Investigating PID shunting in polycrystalline silicon modules via multiscale, multitechnique characterization

We investigated the potential-induced degradation (PID) shunting mechanism in multicrystalline-silicon photovoltaic modules by using a multiscale, multitechnique characterization approach. Both field-stressed modules and laboratory-stressed mini modules were studied. We used photoluminescence, electroluminescence, and dark lock-in thermography imaging to identify degraded areas at the module scale. Small samples were then removed from degraded areas, laser marked, and imaged by scanning electron microscopy. We used simultaneous electron-beam induced current imaging and focused ion beam milling to mark around PID shunts for chemical analysis by time-of-flight secondary-ion mass spectrometry or to isolate individual shunt defects for transmission electron microscopy and atom-probe tomography analysis. By spanning a range of 10 orders of magnitude in size, this approach enabled us to investigate the root-cause mechanisms for PID shunting. We observed a direct correlation between recombination active shunts and sodium content. The sodium content in shunted areas peaks at the SiNX/Si interface and is consistently observed at a concentration of 0.1% to 2% in shunted areas. Analysis of samples subjected to PID recovery, either activated by electron beam or thermal effects only, reveals that recovery of isolated shunts correlates with diffusion of sodium out of the structural defects to the silicon surface. We observed the role of oxygen and chlorine in PID shunting and found that those species - although sometimes present in structural defects where PID shunting was observed - do not play a consistent role in PID shunting.

14 SOLAR ENERGY↗
DOE OSTI2024

Pretraining Billion-Scale Geospatial Foundational Models on Frontier

As AI workloads increase in scope, generalization capability becomes challenging for small task-specific models and their demand for large amounts of labeled training samples increases. On the contrary, Foundation Models (FMs) are trained with internet-scale unlabeled data via self-supervised learning and have been shown to adapt to various tasks with minimal fine-tuning. Although large FMs have demonstrated significant impact in natural language processing and computer vision, efforts toward FMs for geospatial applications have been restricted to smaller size models, as pretraining larger models requires very large computing resources equipped with state-of-the-art hardware accelerators. Current satellite constellations collect 100+TBs of data a day, resulting in images that are billions of pixels and multimodal in nature. Such geospatial data poses unique challenges opening up new opportunities to develop FMs. We investigate billion scale FMs and HPC training profiles for geospatial applications by pretraining on publicly available data. We studied from end-to-end the performance and impact in the solution by scaling the model size. Our larger 3B parameter size model achieves up to 30% improvement in top1 scene classification accuracy when comparing a 100M parameter model. Moreover, we detail performance experiments on the Frontier supercomputer, America's first exascale system, where we study different model and data parallel approaches using PyTorch's Fully Sharded Data Parallel library. Specifically, we study variants of the Vision Transformer architecture (ViT), conducting performance analysis for ViT models with size up to 15B parameters. By discussing throughput and performance bottlenecks under different parallelism configurations, we offer insights on how to leverage such leadership-class HPC resources when developing large models for geospatial imagery applications.

Tsaris, Aristeidis (aris)↗
DOE OSTI2021

Low Temperature Time-to-Explosion Experiments on HMX in the LLNL's ODTX/P-ODTX System (FY 2020 Annual Report)

Over the last few decades, there has been a considerable research effort on the thermal decomposition and thermal explosion violence of energetic materials at elevated temperatures in different sample geometries and confinements. Thermal explosion studies on various energetic materials in two-dimensional geometry such as the Scaled-Thermal-Explosion-Experiment (STEX) system and the Sandia-Instrumented-Thermal-Ignition (SITI) system have been reported. The One-Dimensional Time to Explosion (ODTX) system, designed and built by LLNL, has been used since 1970s for thermal explosion studies. The system is attractive because of the one-dimensional geometry and minimal sample requirement. With the recent integration of a pressure monitoring element, the system can study pressure behavior of materials subjected to and during thermal exposure. Rapid pressure monitoring in µsec intervals allow for enhanced pressure determination in the time right before thermal explosion. The test data can be used for the validation of existing thermal models, particularly for introducing pressure terms. This report summarizes the efforts in performing small-scale safety tests, particle size measurements and conducting 7 ODTX/P-ODTX experiments on the Cl5 HMX material that Hunting uses for oil field applications.

36 MATERIALS SCIENCE↗
NASA NTRS1993

Michelson Interferometer (MINT)

MINT is a Michelson interferometer designed to measure the thermal emission from the earth at high spectral resolution (2/cm) over a broad spectral range (250-1700/cm, 6-40 mu m) with contiguous 3-pixel wide (12 mrad, 8 km field of view) along-track sampling. MINT is particularly well suited for monitoring cloud properties (cloud cover, effective temperature, optical thickness, ice/water phase, and effective particle size) both day and night, as well as tropospheric water vapor, ozone, and temperature. The key instrument characteristics that make MINT ideally suited for decadal monitoring purposes are: high wavelength to wavelength precision across the full IR spectrum with high spectral resolution; space-proven long-term durability and calibration stability; and small size, low cost, low risk instrument incorporating the latest detector and electronics technology. MINT also incorporates simplicity in design and operation by utilizing passively cooled DTGS detectors and nadir viewing geometry (with target motion compensation). MINT measurement objectives, instrument characteristics, and key advantages are summarized in this paper.

Lacis, Andrew↗