SEARCH · Search NASA
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.
This dataset is generated using a Bayesian Hidden Markov Model trained on the DOE 9505 streamflow projection ensemble. A set of 21,000 streamflow realizations are generated for the Colorado, Sabine, and Trinity river basins in Texas. Separate models are trained either using the full 9505 ensemble or a subset based on three hyperparameters: bias correction, downscaling, and hydrological model.
ABSTRACT In this paper, we present photometric redshift (photo-z) estimates for the Dark Energy Spectroscopic Instrument (DESI) Legacy Imaging Surveys, currently the most sensitive optical survey covering the majority of the extragalactic sky. Our photo-z methodology is based on a machine-learning approach, using sparse Gaussian processes augmented with Gaussian mixture models (GMMs) that allow regions of parameter space to be identified and trained separately in a purely data-driven way. The same GMMs are also used to calculate cost-sensitive learning weights that mitigate biases in the spectroscopic training sample. By design, this approach aims to produce reliable and unbiased predictions for all parts of the parameter space present in wide area surveys. Compared to previous literature estimates using the same underlying photometry, our photo-zs are significantly less biased and more accurate at z > 1, with negligible loss in precision or reliability for resolved galaxies at z < 1. Our photo-z estimates offer accurate predictions for rare high-value populations within the parent sample, including optically selected quasars at the highest redshifts (z > 6), as well as X-ray or radio continuum selected populations across a broad range of flux (densities) and redshift. Deriving photo-z estimates for the full Legacy Imaging Surveys Data Release 8, the catalogues provided in this work offer photo-z estimates predicted to be of high quality for ≳9 × 108 galaxies over ${\sim}19\, 400\, \text{deg}^{2}$ and spanning 0 < z ≲ 7, offering one of the most extensive samples of redshift estimates ever produced.
The US Nuclear Regulatory Commission relies on the lattice physics analysis capabilities of the SCALE code system to perform confirmatory licensing analyses. Either SCALE lattice physics code—TRITON/NEWT or Polaris—can be used to generate cross section data used by the PARCS nodal core simulator for full-core neutronics calculations. This report presents an assessment of the accuracy of SCALE lattice physics codes for preparation of lattice physics data that are used to support simulator codes such as the NRC’s PARCS, for UO 2 -mixed oxide (MOX)/Zr fueled light water reactor (LWR) analyses. Due to the nature of lattice physics calculations, critical reactor experiment benchmarks cannot be modeled in explicit detail in a lattice physics code. However, this limitation does not mean that these measurement data are not usable for lattice physics studies. Therefore, either geometry approximations or axial buckling must be implemented to determine the critical water height. These modeling limitations have led to development of a three-phase assessment strategy. In the first phase, selected critical experiment benchmarks are modeled using the SCALE 3D continuous-energy (CE) Monte Carlo (MC) code KENO, which is the most rigorous neutron transport method available in SCALE, with no approximations in the spatial, angle, or energy treatments. Biases and statistical uncertainties in quantities of interest such as k eff and pin power distributions are determined by comparing CE KENO results to experimental data. This first phase in the accuracy assessment is to demonstrate and establish the use of CE KENO as a reference solution for the second phase. In the second phase, 14 numerical test suites are used to compare SCALE lattice physics calculations with CE KENO as a reference solution. In the third phase, SCALE lattice physics depletion calculations are performed, and the spent fuel isotopic results are compared with available radiochemical assay measurements. Isotopic measurement comparisons provide quantitative assessment of isotopic density distribution predictions with the depletion models in SCALE lattice physics codes. This report documents results for all test suites. The assessment was performed using standard production techniques unless otherwise noted. Both TRITON/NEWT and Polaris exhibited acceptable accuracy for most test cases. For the few test cases in which acceptable accuracy criteria were not met, further code and data development are planned. The computer codes used in this assessment are as follows: SCALE 6.2 rev19189 (pre-release of 6.2.1) was used for CE KENO, TRITON/NEWT and Polaris (PWR only) calculations; SCALE 6.2.2 was used for Polaris calculations to address several updates in support of support boiling water reactor (BWR) lattice geometries. For this work, there is no difference between SCALE 6.2.1 and SCALE 6.2.2 beyond the inclusion of the Polaris BWR analysis capability; The ENDF/B-VII.1 continuous energy and 252 group libraries deployed in SCALE 6.2 were generated with AMPX 6.2 that is distributed with SCALE.
A new method (constrained sinusoidal crossover adjustment) for removing the orbit error in satellite altimetry is tested (using crossovers accumulated in the first 91 days of the Geosat non-repeat era in the tropical Pacific) and found to have excellent qualities. Two features distinguish the new method from the conventional bias-and-tilt crossover adjustment. First, a sine wave (with wavelength equaling the circumference of the Earth) is used to represent the orbit error for each satellite revolution, instead of the bias-and-tilt (and curvature, if necessary) approach for each segment of the satellite ground track. Secondly, the indeterminacy of the adjustment process is removed by a simple constraint minimizing the amplitudes of the sine waves, rather than by fixing selected tracks. Overall the new method is more accurate, more efficient, and much less cumbersome than the old. The idea of restricting the crossover adjustment to crossovers between tracks that are less than certain days apart in order to preserve the large-scale long-term oceanic variability is also tested with inconclusive results because the orbit error was unusually nonstationary in the initial 91 days of the GEOSAT mission.
A new method (constrained sinusoidal crossover adjustment) for removing the orbit error in satellite altimetry is tested (using crossovers accumulated in the first 91 days of the Geosat non-repeat era in the tropical Pacific) and found to have excellent qualities. Two features distinguish the new method from the conventional bias-and-tilt crossover adjustment. First, a sine wave (with wavelength equaling the circumference of the Earth) is used to represent the orbit error for each satellite revolution, instead of the bias-and-tilt (and curvature, if necessary) approach for each segment of the satellite ground track. Secondly, the indeterminacy of the adjustment process is removed by a simple constraint minimizing the amplitudes of the sine waves, rather than by fixing selected tracks. Overall the new method is more accurate, more efficient, and much less cumbersome than the old. The idea of restricting the crossover adjustment to crossovers between tracks that are less than certain days apart in order to preserve the large-scale long-term oceanic variability is also tested with inconclusive results because the orbit error was unusually nonstationary in the initial 91 days of the GEOSAT mission.
A backside illuminated CCD imaging sensor for reading out image charges from wells of the array of pixels is significantly improved for blue, UV, far UV and low energy x-ray wavelengths (1-5000.ANG.) by so overthinning the backside as to place the depletion edge at the surface and depositing a thin transparent metal film of about 10.ANG. on a native-quality oxide film of less than about 30.ANG. grown on the thinned backside. The metal is selected to have a higher work function than that of the semiconductor to so bend the energy bands (at the interface of the semiconductor material and the oxide film) as to eliminate wells that would otherwise trap minority carriers. A bias voltage may be applied to extend the frontside depletion edge to the interface of the semiconductor material with the oxide film in the event there is not sufficient thinning. This metal film (flash gate), which improves and stabilizes the quantum efficiency of a CCD imaging sensor, will also improve the QE of any p-n junction photodetector.
A backside illuminated CCD imaging sensor for reading out image charges from wells of the array of pixels is significantly improved for blue, UV, far UV and low energy x-ray wavelengths (1-5000.ANG.) by so overthinning the backside as to place the depletion edge at the surface and depositing a thin transparent metal film of about 10.ANG. on a native-quality oxide film of less than about 30.ANG. grown on the thinned backside. The metal is selected to have a higher work function than that of the semiconductor to so bend the energy bands (at the interface of the semiconductor material and the oxide film) as to eliminate wells that would otherwise trap minority carriers. A bias voltage may be applied to extend the frontside depletion edge to the interface of the semiconductor material with the oxide film in the event there is not sufficient thinning. This metal film (flash gate), which improves and stabilizes the quantum efficiency of a CCD imaging sensor, will also improve the QE of any p-n junction photodetector.
To find New Physics or to refine our knowledge of the Standard Model at the LHC is an enterprise that involves many factors, such as the capabilities and the performance of the accelerator and detectors, the use and exploitation of the available information, the design of search strategies and observables, as well as the proposal of new models. We focus on the use of the information and pour our effort in re-thinking the usual data-driven ABCD method to improve it and to generalize it using Bayesian Machine Learning techniques and tools. We propose that a dataset consisting of a signal and many backgrounds is well described through a mixture model. Signal, backgrounds and their relative fractions in the sample can be well extracted by exploiting the prior knowledge and the dependence between the different observables at the event-by-event level with Bayesian tools. We show how, in contrast to the ABCD method, one can take advantage of understanding some properties of the different backgrounds and of having more than two independent observables to measure in each event. In addition, instead of regions defined through hard cuts, the Bayesian framework uses the information of continuous distribution to obtain soft-assignments of the events which are statistically more robust. To compare both methods we use a toy problem inspired by pp\to hh\to b\bar b b \bar b p p → h h → b b ‾ b b ‾ , selecting a reduced and simplified number of processes and analysing the flavor of the four jets and the invariant mass of the jet-pairs, modeled with simplified distributions. Taking advantage of all this information, and starting from a combination of biased and agnostic priors, leads us to a very good posterior once we use the Bayesian framework to exploit the data and the mutual information of the observables at the event-by-event level. We show how, in this simplified model, the Bayesian framework outperforms the ABCD method sensitivity in obtaining the signal fraction in scenarios with 1% and 0.5% true signal fractions in the dataset. We also show that the method is robust against the absence of signal. We discuss potential prospects for taking this Bayesian data-driven paradigm into more realistic scenarios.
The configuration described is one of four antenna system configurations developed using a variety of monolithic microwave integrated circuit module arrangements and optical systems. A parametric analysis is expected to produce a data base for the selection of design points for a variety of applications. Soon to be accomplished is the design concept of the active (lens) array, which will take into consideration such factors as, coupling effects, the space-fed power divider network design, input bias and control layout, investigation of thermal distribution, and analysis of module failure (graceful degradation).
This article reports on the analysis of the Ka-band Antenna Performance Experiment tipping-curve data acquired at the DSS-13 research and development beam-waveguide (BWG) antenna. By measuring the operating system temperatures as the antenna is moved form zenith to low-elevation angles and fitting a model to the data, one can obtain information on how well the overall temperature model behaves at zenith and approximate the contribution due to the atmosphere. The atmospheric contribution estimated from the data can be expressed in the form of (1) atmospheric noise temperatures that can provide weather statistic information and be compared against those estimated from other methods and (2) the atmospheric loss factor used to refer efficiency measurements to zero atmosphere. This article reports on an analysis performed on a set of 68 8.4-GHz and 67 32-GHz tipping-curve data sets acquired between December 1993 and May 1995 and compares the results with those inferred from a surface model using input meteorological data and from water vapor radiometer (WVR) data. The general results are that, for a selected subset of tip curves, (1) the BWG tipping-curve atmospheric temperatures are in good agreement with those determined from WVR data (the average difference is 0.06 +/- 0.64 K at 32 GHz) and (2) the surface model average values are biased 3.6 K below those of the BWG and WVR at 32 GHz.
This study presents the development of statistical models to efficiently downscale future projections of solar irradiance for solar energy applications. A climate data set simulated from a Regional Climate Model (RCM) obtained from the North American Coordinated Regional Climate Downscaling Experiment (NA-CORDEX) is selected as input to the statistical models to create high-resolution global horizontal irradiance (GHI) over the contiguous United States (CONUS). Our approach builds statistical downscaling models that (1) regrid RCM data (0.22 degree and daily spatiotemporal resolution), (2) correct bias of GHI projections, (3) downscale the future GHI project from daily-scale to hourly-scale, and (4) spatially downscale to generate GHI at 8-km resolution. To calibrate and validate the statistical models, we adapt and use the National Solar Radiation Database (NSRDB). Preliminary results show that the statistical downscaling approach downscales future projections of GHI under two climate scenarios (RCP4.5 and RCP8.5) with a nBIAS of 3%, nMAE of 34% and nRMSE of 46% estimated against NSRDB for the contiguous United State. This presentation will summarize the implemented methodology and validation results as well as future extension of this research.
A complete all-sky sample of objects, flux-limited at 60 microns, has been extracted from the data base of the IRAS. The sample consists of 5014 objects, of which 2649 are galaxies and 13 are not yet identified. In order to study large-scale structure with this sample, it must be free of systematic biases. Corrections are applied for a major systematic effect in the flux densities listed in the IRAS Point Source Catalog: sources resolved by the IRAS beam have flux densities systematically underestimated. In addition, accurate flux densities are obtained for sources flagged as variable, or of moderate flux quality at 60 microns. The IRAS detectors suffered radiation-induced responsivity enhancement (hysteresis) due to crossings of the satellite scans across the Galactic plane; this effect is measured and is shown to be negligible.
Two suites of geomagnetic field models were generated at the request of Los Alamos National Lab. concerning Strategic Defense Initiative (SDI) research. The first is a progression of five models incorporating MAGSAT data and data from a sequence of batches as a priori information. The batch sequence is: post 1979.5 observatory data, post 1980 land survey and selected aeromagnetic and marine survey data, a special White Sands (NM) area survey by Project Magnet with some additional post 1980 marine survey data, and finally DE-2 satellite data. These models are of 13th deg and order in their main field terms, and deg and order 10 in their first derivative temporal terms. The second suite consists of four models based solely upon post 1983.5 observatory and survey data. They are of deg and order 10 in main field and 8 in a first deg Taylor series. A comprehensive error analysis was applied to both series, which accounted for error sources such as the truncated core and crustal fields, and the neglected Sq and low deg crustal fields. Comparison of the power spectrum of the MGST (10/81) model with those of this series show good agreement.
The training of deep neural network models on large data remains a difficult problem, despite progress towards scalable techniques. In particular, there is a mismatch between the random but predetermined order in which AI flows select training samples and the streaming I/O patterns for which traditional HPC data storage (e.g., parallel file systems) are designed. In addition, as more data are obtained, it is feasible neither simply to train learning models incrementally, due to catastrophic forgetting (i.e., bias towards new samples), nor to train frequently from scratch, due to prohibitive time and/or resource constraints. In this paper, we study data management techniques that combine caching and streaming with rehearsal support in order to enable efficient access to training samples in both offline training and continual learning. We revisit state-of-art streaming approaches based on data pipelines that transparently handle prefetching, caching, shuffling, and data augmentation, and discuss the challenges and opportunities that arise when combining these methods with data-parallel training techniques. We also report on preliminary experiments that evaluate the I/O overheads involved in accessing the training samples from a parallel file system (PFS) under several concurrency scenarios, highlighting the impact of the PFS on the design of the data pipelines.
The tip of the red giant branch (TRGB) is an apparent discontinuity of the luminosity function (LF) due to the end of the red giant evolutionary phase and is used to measure distances in the local universe. In practice, tip localization via edge detection response (EDR) relies on several methods applied on a case-by-case basis. It is hard to evaluate how individual choices affect a distance estimation using only a single host field while also avoiding confirmation bias. To devise a standardized approach, we compare unsupervised, algorithmic analyses of the TRGB in multiple halo fields per galaxy. We first optimize methods for the lowest field-to-field dispersion, including spatial filtering, smoothing, and weighting of LF, color band selection, and tip selection based on the number of likely RGB stars and the ratio of stars below versus above the tip (R). We find R, which we call the tip contrast, to be the most important indicator of the quality of EDR measurements; higher R selection can decrease field-to-field dispersion. Further, since R is found to correlate with the age or metallicity of the stellar population based on theoretical modeling, it might result in a displacement of the detected tip magnitude. We find a tip-contrast relation with a slope of -0.023 ± 0.0046 mag/ratio, an ~5σ result that can be used to correct these variations in the detections. When using TRGB to establish a distance ladder, consistent TRGB standardization using tip-contrast relation across rungs is vital to make robust cosmological measurements.
Several areas were investigated for improving vertical accuracy for a rotorcraft using the differential Global Positioning System (GPS) during a landing approach. Continuous deltaranging was studied and the potential improvement achieved by estimating acceleration was studied by comparing the performance on a constant acceleration turn and a rough landing profile of several filters: a position-velocity (PV) filter, a position-velocity-constant acceleration (PVAC) filter, and a position-velocity-turning acceleration (PVAT) filter. In overall statistics, the PVAC filter was found to be most efficient with the more complex PVAT performing equally well. Vertical performance was not significantly different among the filters. Satellite selection algorithms based on vertical errors only (vertical dilution of precision or VDOP) and even-weighted cross-track and vertical errors (XVDOP) were tested. The inclusion of an altimeter was studied by modifying the PVAC filter to include a baro bias estimate. Improved vertical accuracy during degraded DOP conditions resulted. Flight test results for raw differential results excluding filter effects indicated that the differential performance significantly improved overall navigation accuracy. A landing glidepath steering algorithm was devised which exploits the flexibility of GPS in determining precise relative position. A method for propagating the steering command over the GPS update interval was implemented.
A recently proposed hybrid-neutral-network and fuzzy-logic-control architecture is applied to a fuzzy logic controller developed for attitude control of the Space Shuttle. A model using reinforcement learning and learning from past experience for fine-tuning its knowledge base is proposed. Two main components of this approximate reasoning-based intelligent control (ARIC) model - an action-state evaluation network and action selection network are described as well as the Space Shuttle attitude controller. An ARIC model for the controller is presented, and it is noted that the input layer in each network includes three nodes representing the angle error, angle error rate, and bias node. Preliminary results indicate that the controller can hold the pitch rate within its desired deadband and starts to use the jets at about 500 sec in the run.
One of the high-priority problems in satellite precipitation estimation is developing algorithms for estimating the errors in precipitation retrievals by individual sensors and subsequent multi-satellite combinations. Classically, we distinguish between "random" and "bias" errors, which do and do not, respectively average to zero over a "big enough" time/space sample. The current operational GPCP and TRMM multi-satellite algorithms are nearly unique in estimating random error for the monthly precipitation estimates from individual sensor systems (including gauge), following Huffman, and then making multi-sensor combinations. No routinely operational global precipitation produces estimates of bias error. Subsequently, a similar scheme has been followed to provide random error estimates for the Multi-satellite Precipitation Analysis (MPA) being computed in real time and after real time for TRMM. The Huffman algorithm for random error is briefly reviewed, including a discussion of the limitations imposed on the algorithm by standard monthly precipitation data sets. Starting from a very simple theoretical treatment of the histogram of precipitation samples in a month, an equation is developed that depends on the estimated average precipitation rate for the month, the number of samples in the month, and two constants. The constants are set separately for each source of precipitation estimate (such as "raingauge") by calibration at selected ground sites. We discuss recent work validating the random error estimates to highlight the successes and limitations of this first-generation approach. We then consider what information is needed from the individual sensor algorithms to facilitate additional accuracy in the estimation of random errors across the time/space span of climate regimes which a global estimation system must handle. In addition, the thorny issue of estimating bias is raised. Finally, the role of error estimates (and the qualitative errors!) in creating combinations of precipitation estimates from different individual sensors is discussed. This issue is particularly important when fine scales in space and time are being considered, say the 0.25 x 0.25-deg 3-hourly estimates in the MPA.