Search NASA⌕ Search

SEARCH · Search NASA

Results for “random sampling”

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 343 records · Page 19

Monte Carlo Transport: Computational Physics Summer Workshop [Slides]

In general, Monte Carlo methods simulate large numbers of random trials in order to observe numerical behavior of systems described by probabilistic behavior. In radiation transport, pseudo-random number generators are used to randomly sample individual particle lives. Information about the particles are tallied.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

ENHANCING STAKEHOLDER ACCEPTANCE OF BIOREMEDIATION TECHNOLOGIES

This project inquired into the judgments and beliefs of people living near DOE reservations and facilities at Oak Ridge, Tennessee; Hanford, Washington; and Los Alamos, Tennessee about bioremediation of subsurface contamination. The purpose of the investigation was to identify strategies based on these judgments and beliefs for enhancing public support of bioremediation. Several methods were used to collect and analyze data including content analysis of transcripts of face-to-face personal interviews, factor analysis of subjective perspectives using Q methodology, and statistical analysis of results from a large-sample randomized telephone survey. Content analysis of interview transcripts identified themes about public perceptions and constructions of contamination risk, risk management, and risk managers. This analysis revealed that those who have no employment relationship at the sites and are not engaged in technical professions are most concerned about contamination risks. We also found that most interviewees are unfamiliar with subsurface contamination risks and how they can be reduced, believe they have little control over exposure, are frustrated with the lack of progress in remediation, are concerned about a lack of commitment of DOE to full remediation, and distrust site managers to act in the public interest. Concern is also expressed over frequent site management turnover, excessive secrecy, ineffective and biased communication, perceived attempts to talk the public into accepting risk, and apparent lack of concern about community welfare. In the telephone survey, we asked respondents who were aware of site contamination about their perceptions of risk from exposure to subsurface contamination. Response analysis revealed that most people believe that they are at significant risk from subsurface contamination but they acknowledge that more education is needed to calibrate risk perceptions against scientific risk assessments. Most rate their personal control over exposure as low. Slightly more than half believe that risk reduction should be balanced against cost. We also found that distrust of DOE and its contractors exists, primarily due to the perception that site managers do not share public values; hence, the public is generally unwilling to defer to DOE in its decision-making. The concomitant belief of inefficacy confounds distrust by generating frustration that DOE does not care. Moreover, the public is split with respect to trust of each other, primarily because of the belief that citizens lack technical competence. With respect to bioremediation support, we found that more than 40% of the public has no opinion. However, of those who do, 3 of 4 are favorably disposed – particularly among those who believe that risk is lower and who are more trusting of site management. We presented survey respondents with four alternative participation strategies based on the results of the Q analysis and asked their judgments of each. The public prefers strategies that shifts power to them. The least empowered strategy (feedback) was supported by 46%; support grew as public power increased, reaching 66% support for independently facilitated deliberation. More DOE distrust generates more support for high power strategies. We offer the following recommendations to enhance public acceptance. First, and perhaps most importantly, site managers should pursue robust trust-building efforts to gain public confidence in DOE risk management that meets public expectations. Public trust decreases risk perception, which increases public willingness to defer to site managers’ discretion in decision-making, which in turn increases public acceptance of the decisions that result. Second, site managers should address public concerns about bioremediation such as its effectiveness in reducing risk, performance compared to other remediation alternatives, costs compared against benefits, time required to start and complete remediation, level of risk that is currently posed by contamination, and scope of application. Third, more should be done to involve the public in bioremediation decision-making. We recommend a two-stage process: independent facilitated deliberation to build trust and address concerns about the motives and competence of site managers, followed by consultation to maintain that trust. Both stages should be inclusive, transparent, and respectful. Participation objectives and roles of participants should be well specified by the participants. A record of discussion should be published that codifies the concerns raised and how they were addressed, which will facilitate progress by averting the need to reconsider the same issues repeatedly. It is most important that the processes convince the public that its participation influences decision outcomes and that participants genuinely (informed and voluntarily) consent to risk exposure.

07 ISOTOPE AND RADIATION SOURCES↗

Evaluation of the Super ESPC Program: Level 2 -- Recalculated Cost Savings

This report presents the results of Level 2 of a three-tiered evaluation of the U.S. Department of Energy Federal Energy Management Program's Super Energy Savings Performance Contract (Super ESPC) Program. Level 1 of the analysis studied all of the Super ESPC projects for which at least one Annual Measurement & Verification (M&V) Report had been produced by April 2006. For those 102 projects in aggregate, we found that the value of cost savings reported by the energy service company (ESCO) in the Annual M&V Reports was 108% of the cost savings guaranteed in the contracts. We also compared estimated energy savings (which are not guaranteed, but are the basis for the guaranteed cost savings) to the energy savings reported by the ESCO in the Annual M&V Report. In aggregate, reported energy savings were 99.8% of estimated energy savings on the basis of site energy, or 102% of estimated energy savings based on source energy. Level 2 focused on a random sample of 27 projects taken from the 102 Super ESPC projects studied in Level 1. The objectives were, for each project in the sample, to: repeat the calculations of the annual energy and cost savings in the most recent Annual M&V Report to validate the ESCO's results or correct any errors, and recalculate the value of the reported energy, water, and operations and maintenance (O&M) savings using actual utility prices paid at the project site instead of the 'contract' energy prices - the prices that are established in the project contract as those to be used by the ESCO to calculate the annual cost savings, which determine whether the guarantee has been met. Level 3 analysis will be conducted on three to five projects from the Level 2 sample that meet validity criteria for whole-building or whole-facility data analysis. This effort will verify energy and cost savings using statistical analysis of actual utility use, cost, and weather data. This approach, which can only be used for projects meeting particular validity criteria, is described in Shonder and Florita (2003) and Shonder and Hughes (2005). To address the first objective of the Level 2 analysis, we first assembled all the necessary information, and then repeated the ESCOs' calculations of reported annual cost savings. Only minor errors were encountered, the most common being the use of incorrect escalation rates to calculate utility prices or O&M savings. Altogether, our corrected calculations of the ESCO's reported cost savings were within 0.6% of the ESCOs' reported cost savings, and errors found were as likely to favor the government as they were the ESCO. To address the second objective, we gathered data on utility use and cost from central databases maintained by the Department of Defense and the General Services Administration, and directly from some of the sites, to determine the prices of natural gas and electricity actually paid at the sites during the periods addressed by the annual reports. We used these data to compare the actual utility costs at the sites to the contract utility prices. For natural gas, as expected, we found that prices had risen much faster than had been anticipated in the contracts. In 17 of the 18 projects for which the comparison was possible, contract gas prices were found to be lower than the average actual prices being paid. We conclude that overall in the program, the estimates of gas prices and gas price escalation rates used in the Super ESPC projects have been conservative. For electricity, it was possible to compare contract prices with the actual (estimated) marginal prices of electricity in 20 projects. In 14 of these projects, the overall contract electricity price was found to be lower than the marginal price of electricity paid to the serving utility. Thus it appears that conservative estimates of electricity prices and escalation rates have been used in the program as well. Finally we calculated the value of the reported energy savings using the prices of utilities actually paid by the sites instead of the contract prices. In 16 of the 22 projects (where this calculation was possible), the recalculated annual cost savings were greater than the annual cost savings reported by the ESCO. In the aggregate for the 22 projects analyzed, the annual cost savings calculated using actual energy prices were found to be 111% of the ESCO-reported savings. Using statistical methods to expand this estimate to the entire 102-project population of Super ESPC projects, we estimate that the total annual cost savings calculated using actual energy prices instead of contract energy prices are about 110% of the reported annual cost savings. We can combine the results of the Level 1 and Level 2 studies to estimate overall cost savings realized by the government for the entire 102-project population of Super ESPC projects examined in Level 1.

03 NATURAL GAS↗

Natural Selection Footprints Among African Chicken Breeds and Village Ecotypes

Natural selection is likely a major factor in shaping genomic variation of the African indigenous rural chicken, driving the development of genetic footprints. Selection footprints are expected to be associated with adaptation to locally prevailing environmental stressors, which may include diverse factors as high altitude, disease resistance, poor nutrition, oxidative and heat stresses. To determine the existence of a selection footprint, 268 birds were randomly sampled from three indigenous ecotypes from East Africa (Rwanda and Uganda) and North Africa (Baladi), and two registered Egyptian breeds (Dandarawi and Fayoumi). Samples were genotyped using the chicken Affymetrix 600K Axiom R Array. A total of 494,332 SNPs were utilized in the downstream analysis after implementing quality control measures. The intra-population runs of homozygosity (ROH) that occurred in >50% of individuals of an ecotype or in >75% of a breed were studied. To identify inter-population differentiation due to genetic structure, FST was calculated for North- vs. East-African populations and Baladi and Fayoumi vs. Dandarawi for overlapping windows (500 kb with a step-size of 250 kb). The ROH and FST mapping detected several selective sweeps on different autosomes. Results reflected selection footprints of the environmental stresses, breed behavior, and management. Intra-population ROH of the Egyptian chickens showed selection footprints bearing genes for adaptation to heat, solar radiation, ion transport and immunity. The high-altitude-adapted East-African populations’ ROH showed a selection signature with genes for angiogenesis, oxygen-heme binding and transport. The neuroglobin gene (GO:0019825 and GO:0015671) was detected on a Chromosome 5 ROH of Rwanda–Uganda ecotypes. The sodium-dependent noradrenaline transporter, SLC6A2 on a Chromosome 11 ROH in Fayoumi breed may reflect its active behavior. Inter-population FST among Egyptian populations reflected genetic mechanisms for the Fayoumi resistance to Newcastle Disease Virus (NDV), while FST between Egyptian and Rwanda–Uganda populations indicated the Secreted frizzled related protein 2, SFRP2, on Chromosome 4, that contributes to melanogenic activity and most likely enhances the Dandarawi chicken adaptation to high-intensity of solar radiation in Southern Egypt. These results enhance our understanding of the natural selection forces role in shaping genomic structure for adaptation to the stressful African conditions.

59 BASIC BIOLOGICAL SCIENCES↗

Simultaneous Discovery of Positive and Negative Interactions Among Rhizosphere Bacteria Using Microwell Recovery Arrays

Understanding microbe-microbe interactions is critical to predict microbiome function and to construct communities for desired outcomes. Investigation of these interactions poses a significant challenge due to the lack of suitable experimental tools available. Here we present the microwell recovery array (MRA), a new technology platform that screens interactions across a microbiome to uncover higher-order strain combinations that inhibit or promote the function of a focal species. One experimental trial generates 10 4 microbial communities that contain the focal species and a distinct random sample of uncharacterized cells from plant rhizosphere. Cells are sequentially recovered from individual wells that display highest or lowest levels of focal species growth using a high-resolution photopolymer extraction system. Interacting species are then identified and putative interactions are validated. Using this approach, we screen the poplar rhizosphere for strains affecting the growth of Pantoea sp. YR343, a plant growth promoting bacteria isolated from Populus deltoides rhizosphere. In one screen, we montiored 3,600 microwells within the array to uncover multiple antagonistic Stenotrophomonas strains and a set of Enterobacter strains that promoted YR343 growth. The later demonstrates the unique ability of the platform to discover multi-membered consortia that generate emergent outcomes, thereby expanding the range of phenotypes that can be characterized from microbiomes. This knowledge will aid in the development of consortia for Populus production, while the platform offers a new approach for screening and discovery of microbial interactions, applicable to any microbiome.

59 BASIC BIOLOGICAL SCIENCES↗

A fast Monte Carlo cell-by-cell simulation for radiobiological effects in targeted radionuclide therapy using pre-calculated single-particle track standard DNA damage data

Introduction: We developed a new method that drastically speeds up radiobiological Monte Carlo radiation-track-structure (MC-RTS) calculations on a cell-by-cell basis. Methods: The technique is based on random sampling and superposition of single-particle track (SPT) standard DNA damage (SDD) files from a “pre-calculated” data library, constructed using the RTS code TOPAS-nBio, with “time stamps” manually added to incorporate dose-rate effects. This time-stamped SDD file can then be input into MEDRAS, a mechanistic kinetic model that calculates various radiation-induced biological endpoints, such as DNA double-strand breaks (DSBs), misrepairs and chromosomal aberrations, and cell death. As a benchmark validation of the approach, we calculated the predicted energy-dependent DSB yield and the ratio of direct-to-total DNA damage, both of which agreed with published in vitro experimental data. We subsequently applied the method to perform a superfast cell-by-cell simulation of an experimental in vitro system consisting of neuroendocrine tumor cells uniformly incubated with 177 Lu. Results and discussion: The results for residual DSBs, both at 24 and 48 h post-irradiation, are in line with the published literature values. Our work serves as a proof-of-concept demonstration of the feasibility of a cost-effective “in silico clonogenic cell survival assay” for the computational design and development of radiopharmaceuticals and novel radiotherapy treatments more generally.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Convergence Rates for Empirical Estimation of Binary Classification Bounds

Bounding the best achievable error probability for binary classification problems is relevant to many applications including machine learning, signal processing, and information theory. Many bounds on the Bayes binary classification error rate depend on information divergences between the pair of class distributions. Recently, the Henze–Penrose (HP) divergence has been proposed for bounding classification error probability. We consider the problem of empirically estimating the HP-divergence from random samples. We derive a bound on the convergence rate for the Friedman–Rafsky (FR) estimator of the HP-divergence, which is related to a multivariate runs statistic for testing between two distributions. The FR estimator is derived from a multicolored Euclidean minimal spanning tree (MST) that spans the merged samples. We obtain a concentration inequality for the Friedman–Rafsky estimator of the Henze–Penrose divergence. We validate our results experimentally and illustrate their application to real datasets.

97 MATHEMATICS AND COMPUTING↗

Coreset Clustering on Small Quantum Computers

Many quantum algorithms for machine learning require access to classical data in superposition. However, for many natural data sets and algorithms, the overhead required to load the data set in superposition can erase any potential quantum speedup over classical algorithms. Recent work by Harrow introduces a new paradigm in hybrid quantum-classical computing to address this issue, relying on coresets to minimize the data loading overhead of quantum algorithms. We investigated using this paradigm to perform k-means clustering on near-term quantum computers, by casting it as a QAOA optimization instance over a small coreset. We used numerical simulations to compare the performance of this approach to classical k-means clustering. We were able to find data sets with which coresets work well relative to random sampling and where QAOA could potentially outperform standard k-means on a coreset. However, finding data sets where both coresets and QAOA work well—which is necessary for a quantum advantage over k-means on the entire data set—appears to be challenging.

42 ENGINEERING↗

A Study of the Properties and Dynamics of the Disk of Satellites in a Milky-Way-like Galaxy System

The dynamics of the satellite systems of Milky-Way-like galaxies offer a useful means by which to study the galaxy formation process in the cosmological context. It has been suggested that the currently observed anisotropic distribution of the satellites in such galaxy systems is inconsistent with the concordance ΛCDM cosmology model on the galactic scale if the observed satellites are random samples of the dark matter (DM) sub-halos that are nearly isotropically distributed around the central galaxy. In this study, we present original high-resolution zoom-in studies of central galaxies and satellite systems based upon initial conditions for the DM distribution from the Aquarius simulations but with substantial high-resolution baryon physics added. We find that the galaxy most like the Milky Way in this study does indeed contain a disk of satellites (DOS). Although one galaxy DOS system does not answer the question of how common such disks are, it does allow the opportunity to explore the properties and dynamics of the DOS system. Our investigation centers on the spatial arrangement (distances, angles, etc.) of satellites in this Milky-Way-like galaxy system with a specific emphasis on identifying and analyzing the disk-like structure along with its dynamical and morphological properties. Among the conclusions from this study, we find that the satellites and DM sub-halos in the galaxy simulations are anisotropically distributed. The dynamical properties of the satellites, however, indicate that the direction of the angular momentum vector of the whole satellite system is different from the normal direction of the fitted DOS and from the normal direction of the velocity dispersion of the system. Hence, the fitted DOS appears to be comprised of infalling sub-halos and is not a rotationally supported system.

Zhao, Xinghai↗

Caffeine Use among Active Duty Navy and Marine Corps Personnel

Data from the National Health and Nutrition Examination Survey (NHANES) indicate 89% of Americans regularly consume caffeine, but these data do not include military personnel. This cross-sectional study examined caffeine use in Navy and Marine Corps personnel, including prevalence, amount of daily consumption, and factors associated with use. A random sample of Navy and Marine Corps personnel was contacted and asked to complete a detailed questionnaire describing their use of caffeine-containing substances, in addition to their demographic, military, and lifestyle characteristics. A total of 1708 service members (SMs) completed the questionnaire. Overall, 87% reported using caffeinated beverages ≥1 time/week, with caffeine users consuming a mean ± standard error of 226 ± 5 mg/day (242 ± 7 mg/day for men, 183 ± 8 mg/day for women). The most commonly consumed caffeinated beverages (% users) were coffee (65%), colas (54%), teas (40%), and energy drinks (28%). Multivariable logistic regression modeling indicated that characteristics independently associated with caffeine use (≥1 time/week) included older age, white race/ethnicity, higher alcohol consumption, and participating in less resistance training. Prevalence of caffeine use in these SMs was similar to that reported in civilian investigations, but daily consumption (mg/day) was higher.

60 APPLIED LIFE SCIENCES↗

Inference of Unresolved Point Sources at High Galactic Latitudes Using Probabilistic Catalogs

The detection of point sources in images is a fundamental operation in astrophysics, and is crucial for constraining population models of the underlying point sources or characterizing the background emission. Standard techniques fall short in the crowded-field limit, losing sensitivity to faint sources and failing to track their covariance with close neighbors. We construct a Bayesian framework to perform inference of faint or overlapping point sources. The method involves probabilistic cataloging, where samples are taken from the posterior probability distribution of catalogs consistent with an observed photon count map. In order to validate our method, we sample random catalogs of the gamma-ray sky in the direction of the North Galactic Pole (NGP) by binning the data in energy and point-spread function classes. Using three energy bins spanning 0.3–1, 1–3, and 3–10 GeV, we identify 270{sub −10}{sup +30} point sources inside a 40{sup ∘}×40{sup ∘} region around the NGP above our point-source inclusion limit of 3×10{sup −11} cm{sup −2} s{sup −1} sr{sup −1} GeV{sup −1} at the 1–3 GeV energy bin. Modeling the flux distribution as a power law, we infer the slope to be −1.92{sub −0.05}{sup +0.07} and estimate the contribution of point sources to the total emission as 18{sub −2}{sup +2}%. These uncertainties in the flux distribution are fully marginalized over the number as well as the spatial and spectral properties of the unresolved point sources. This marginalization allows a robust test of whether the apparently isotropic emission in an image is due to unresolved point sources or of truly diffuse origin.

79 ASTRONOMY AND ASTROPHYSICS↗

On Presolar Stardust Grains from CO Classical Novae

About 30%–40% of classical novae produce dust 20–100 days after the outburst, but no presolar stardust grains from classical novae have been unambiguously identified yet. Although several studies claimed a nova paternity for certain grains, the measured and simulated isotopic ratios could only be reconciled, assuming that the grains condensed after the nova ejecta mixed with a much larger amount of close-to-solar matter. However, the source and mechanism of this potential post-explosion dilution of the ejecta remains a mystery. A major problem with previous studies is the small number of simulations performed and the implied poor exploration of the large nova parameter space. In this paper, we report the results of a different strategy, based on a Monte Carlo technique, that involves the random sampling over the most important nova model parameters: the white dwarf composition; the mixing of the outer white dwarf layers with the accreted material before the explosion; the peak temperature and density; the explosion timescales; and the possible dilution of the ejecta after the outburst. We discuss and take into account the systematic uncertainties for both the presolar grain measurements and the simulation results. Only those simulations that are consistent with all measured isotopic ratios of a given grain are accepted for further analysis. We also present the numerical results of the model parameters. Lastly, we identify 18 presolar grains with measured isotopic signatures consistent with a CO nova origin, without assuming any dilution of the ejecta. Among these, the grains G270_2, M11-334-2, G278, M11-347-4, M11-151-4, and Ag2_6 have the highest probability of a CO nova paternity.

nuclear reactions↗

Long-term Evolution of Relativistic Unmagnetized Collisionless Shocks

Abstract We study a relativistic collisionless electron–positron shock propagating into an unmagnetized ambient medium using 2D particle-in-cell simulations of unprecedented duration and size. The shock generates intermittent magnetic structures of increasingly larger size as the simulation progresses. Toward the end of our simulation, at around 26,000 plasma times, the magnetic coherence scale approaches λ ∼ 100 plasma skin depths, both ahead and behind the shock front. We anticipate a continued growth of λ beyond the time span of our simulation, as long as the shock accelerates particles to increasingly higher energies. The post-shock field is concentrated in localized patches, which maintain a local magnetic energy fraction ε B ∼ 0.1. Particles randomly sampling the downstream fields spend most of their time in low field regions ( ε B ≪ 0.1) but emit a large fraction of the synchrotron power in the localized patches with strong fields ( ε B ∼ 0.1). Our results have important implications for models of gamma-ray burst afterglows.

Astronomy & Astrophysics↗

Testing Classical Properties from Quantum Data

Many properties of Boolean functions can be tested far more efficiently than the function itself can be learned. However, this dramatic advantage often disappears when testers are limited to random samples of ƒ instead of adaptively chosen queries to f. In this work we investigate the quantum version of this restriction: quantum algorithms that test properties of a Boolean function f solely from copies of either the function state |ƒ⟩ ∝ ∑ x |x, ƒ(x)⟩ or the phase state |(-1) ƒ ⟩ ∝ ∑ x (-1) ƒ(x) |x⟩. For monotonicity, symmetry, and triangle-freeness, we show passive quantum testers are unboundedly or super-polynomially better than their classical passive testing counterparts. They are competitive with classic query -based testers in each case. Our new testers use techniques beyond quantum Fourier sampling, and it turns out this is necessary: we show a certain class of bent functions can be tested from 𝒪(1) function states but has a sample complexity lower bound of 2 Ω(n) for any tester relying exclusively on Fourier and classical samples. Our passive quantum testers are competitive with classical query -based testers, but this isn't universal: we exhibit a testing problem that can be solved from 𝒪(1) classical queries but requires Ω(2 n/2 ) function state copies. The Forrelation problem provides a separation of the same magnitude in the opposite direction, so we conclude that quantum data and classical queries are "maximally incomparable" resources for testing. We also begin the study of lower bounds for testing from quantum data. For quantum monotonicity testing, we prove that the ensembles of [Goldreich et al., 2000; Black, 2024], which give exponential lower bounds for classical sample-based testing, do not yield any nontrivial lower bounds for testing from quantum data. New insights specific to quantum data will be required for proving copy complexity lower bounds for testing in this model.

Boolean Functions↗

A Machine Learning-Genetic Algorithm (ML-GA) Approach for Rapid Optimization Using High-Performance Computing

A Machine Learning-Genetic Algorithm (ML-GA) approach was developed to virtually discover optimum designs using training data generated from multi-dimensional simulations. Machine learning (ML) presents a pathway to transform complex physical processes that occur in a combustion engine into compact informational processes. In the present work, a total of over 2000 sector-mesh computational fluid dynamics (CFD) simulations of a heavy-duty engine were performed. These were run concurrently on a supercomputer to reduce overall turnaround time. The engine being optimized was run on a low-octane (RON70) gasoline fuel under partially premixed compression ignition (PPCI) mode. A total of nine input parameters were varied, and the CFD simulation cases were generated by randomly sampling points from this nine-dimensional input space. These input parameters included fuel injection strategy, injector design, and various in-cylinder flow and thermodynamic conditions at intake valve closure (IVC). The outputs (targets) of interest from these simulations included five metrics related to engine performance and emissions. Over 2000 samples generated from CFD were then used to train an ML model that could predict these five targets based on the nine input features. A robust super learner approach was employed to build the ML model, where results from a collection of different ML algorithms were pooled together. Thereafter, a stochastic global optimization genetic algorithm (GA) was used, with the ML model as the objective function, to optimize the input parameters based on a merit function so as to minimize fuel consumption while satisfying CO and NOx emissions constraints. The optimized configuration from ML-GA was found to be very close to that obtained from a sequentially performed CFD-GA approach, where a CFD simulation served as the objective function. In addition, the overall turnaround time was (at least) 75% lower with the ML-GA approach, as the training data was generated from concurrent CFD simulations and employing the ML model as the objective function significantly accelerated the GA optimization. Finally, this study demonstrates the potential of ML-GA and high-performance computing (HPC) to reduce the number of CFD simulations to be performed for optimization problems without loss in accuracy, thereby providing significant cost savings compared to traditional approaches.

97 MATHEMATICS AND COMPUTING↗

The suitability of differentiable, physics-informed machine learning hydrologic models for ungauged regions and climate change impact assessment

As a genre of physics-informed machine learning, differentiable process-based hydrologic models (abbreviated as δ or delta models) with regionalized deep-network-based parameterization pipelines were recently shown to provide daily streamflow prediction performance closely approaching that of state-of-the-art long short-term memory (LSTM) deep networks. Meanwhile, δ models provide a full suite of diagnostic physical variables and guaranteed mass conservation. Here, we ran experiments to test (1) their ability to extrapolate to regions far from streamflow gauges and (2) their ability to make credible predictions of long-term (decadal-scale) change trends. We evaluated the models based on daily hydrograph metrics (Nash–Sutcliffe model efficiency coefficient, etc.) and predicted decadal streamflow trends. For prediction in ungauged basins (PUB; randomly sampled ungauged basins representing spatial interpolation), δ models either approached or surpassed the performance of LSTM in daily hydrograph metrics, depending on the meteorological forcing data used. They presented a comparable trend performance to LSTM for annual mean flow and high flow but worse trends for low flow. For prediction in ungauged regions (PUR; regional holdout test representing spatial extrapolation in a highly data-sparse scenario), δ models surpassed LSTM in daily hydrograph metrics, and their advantages in mean and high flow trends became prominent. In addition, an untrained variable, evapotranspiration, retained good seasonality even for extrapolated cases. The δ models' deep-network-based parameterization pipeline produced parameter fields that maintain remarkably stable spatial patterns even in highly data-scarce scenarios, which explains their robustness. Combined with their interpretability and ability to assimilate multi-source observations, the δ models are strong candidates for regional and global-scale hydrologic simulations and climate change impact assessment.

54 ENVIRONMENTAL SCIENCES↗

A parcel-level evaluation of distributed wind opportunity in the contiguous United States

This study examines the potential for distributed wind (DW) energy across the contiguous United States, leveraging advancements in the National Renewable Energy Laboratory's distributed wind model, dWind. The novel modeling approach described here utilizes a high-resolution dataset and analyzes over 150 million parcels, a significant improvement from prior methods that extrapolated results from a smaller random sample. This achievement is enabled through key model performance improvements, such as transitioning to multiprocessing, which reduces runtime by 97 %. This optimized, high-resolution approach allows the inspection of technology deployment potential and impact on a variety of scales tailored to individual properties and regions. The results here align with prior work showing substantial opportunity for energy generation using DW technologies. Key findings reveal a substantial increase from prior results in estimated technical and economic potential for DW. Metrics tuned to highlight economic potential also show increased incentives supporting rural adoption. Results are spatially aggregated for usability and published via the U.S. Department of Energy Wind Data Portal and a custom scenario visualization platform, aiding policymakers, industry, and property owners in assessing DW viability across various scenarios and spatial scales.

17 WIND ENERGY↗

Hyperspectral Image Classification Using Graph Clustering Methods

Hyperspectral imagery is a challenging modality due to the dimension of the pixels which can range from hundreds to over a thousand frequencies depending on the sensor. Most methods in the literature reduce the dimension of the data using a method such as principal component analysis, however this procedure can lose information. More recently methods have been developed to address classification of large datasets in high dimensions. This paper presents two classes of graph-based classification methods for hyperspectral imagery. Using the full dimensionality of the data, we consider a similarity graph based on pairwise comparisons of pixels. The graph is segmented using a pseudospectral algorithm for graph clustering that requires information about the eigenfunctions of the graph Laplacian but does not require computation of the full graph. We develop a parallel version of the Nyström extension method to randomly sample the graph to construct a low rank approximation of the graph Laplacian. With at most a few hundred eigenfunctions, we can implement the clustering method designed to solve a variational problem for a graph-cut-based semi-supervised or unsupervised classification problem. We implement OpenMP directive-based parallelism in our algorithms and show performance improvement and strong, almost ideal, scaling behavior. The method can handle very large datasets including a video sequence with over a million pixels, and the problem of segmenting a data set into a pre-determined number of classes.

47 OTHER INSTRUMENTATION↗