Search NASA⌕ Search

SEARCH · Search NASA

Results for “Large ensemble”

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 1,009 records · Page 56

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

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

Lin, Jerry [Department of Earth System Sciences Un↗

Evaluations of high-resolution dynamically downscaled ensembles over the contiguous United States

This study uses Weather Research and Forecast (WRF) model to evaluate the performance of six dynamical downscaled decadal historical simulations with 12-km resolution for a large domain (7200 x 6180 km) that covers most of North America. The initial and boundary conditions are from three global climate models (GCMs) and one reanalysis data. The GCMs employed in this study are the Geophysical Fluid Dynamics Laboratory Earth System Model with Generalized Ocean Layer Dynamics component, Community Climate System Model, version 4, and the Hadley Centre Global Environment Model, version 2-Earth System. The reanalysis data is from the National Centers for Environmental Prediction-US. Department of Energy Reanalysis II. We analyze the effects of bias correcting, the lateral boundary conditions and the effects of spectral nudging. We evaluate the model performance for seven surface variables and four upper atmospheric variables based on their climatology and extremes for seven subregions across the United States. The results indicate that the simulation’s performance depends on both location and the features/variable being tested. We find that the use of bias correction and/or nudging is beneficial in many situations, but employing these when running the RCM is not always an improvement when compared to the reference data. The use of an ensemble mean and median leads to a better performance in measuring the climatology, while it is significantly biased for the extremes, showing much larger differences than individual GCM driven model simulations from the reference data. This study provides a comprehensive evaluation of these historical model runs in order to make informed decisions when making future projections.

54 ENVIRONMENTAL SCIENCES↗

A multimodel intercomparison of resolution effects on precipitation: simulations and theory

An ensemble of six pairs of RCM experiments performed at 25 and 50 km for the period 1961–2000 over a large European domain is examined in order to evaluate the effects of resolution on the simulation of daily precipitation statistics. Application of the non-parametric two-sample Kolmorgorov–Smirnov test, which tests for differences in the location and shape of the probability distributions of two samples, shows that the distribution of daily precipitation differs between the pairs of simulations over most land areas in both summer and winter, with the strongest signal over southern Europe. Two-dimensional histograms reveal that precipitation intensity increases with resolution over almost the entire domain in both winter and summer. In addition, the 25 km simulations have more dry days than the 50 km simulations. The increase in dry days with resolution is indicative of an improvement in model performance at higher resolution, while the more intense precipitation exceeds observed values. The systematic increase in precipitation extremes with resolution across all models suggests that this response is fundamental to model formulation. Simple theoretical arguments suggest that fluid continuity, combined with the emergent scaling properties of the horizontal wind field, results in an increase in resolved vertical transport as grid spacing decreases. This increase in resolution-dependent vertical mass flux then drives an intensification of convergence and resolvable-scale precipitation as grid spacing decreases. In conclusion, this theoretical result could help explain the increasingly, and often anomalously, large stratiform contribution to total rainfall observed with increasing resolution in many regional and global models.

54 ENVIRONMENTAL SCIENCES↗

A Novel Emergent Constraint Approach for Refining Regional Climate Model Projections of Peak Flow Timing

Abstract Global climate models (GCMs) are unable to produce detailed runoff conditions at the basin scale. Assumptions are commonly made that dynamical downscaling can resolve this issue. However, given the large magnitude of the biases in downscaled GCMs, it is unclear whether such projections are credible. Here, we use an ensemble of dynamically downscaled GCMs to evaluate this question in the Sierra‐Cascade mountain range of the western US. Future projections across this region are characterized by earlier seasonal shifts in peak flow, but with substantial inter‐model uncertainty (−25 ± 34.75 days, 95% confidence interval (CI)). We apply the emergent constraint (EC) method for the first time to dynamically downscaled projections, leading to a 39% (−28.25 ± 20.75 days, 95% CI) uncertainty reduction in future peak flow timing. While the constrained results can differ from bias corrected projections, the EC is based on GCM biases in historical peak flow timing and has a strong physical underpinning.

54 ENVIRONMENTAL SCIENCES↗

Coherent manipulation of graph states composed of finite-energy Gottesman-Kitaev-Preskill-encoded qubits

Graph states are a central resource in measurement-based quantum information processing. In the photonic qubit architecture based on Gottesman-Kitaev-Preskill (GKP) encoding, the generation of high-fidelity graph states composed of realistic, finite-energy approximate GKP-encoded qubits thus constitutes a key task. We consider the finite-energy approximation of GKP-qubit states given by a coherent superposition of shifted finite-squeezed vacuum states, where the displacements are Gaussian distributed. We present an exact description of graph states composed of such approximate GKP qubits as a coherent superposition of a Gaussian ensemble of randomly displaced ideal GKP-qubit graph states. Using standard Gaussian dynamics, we track the transformation of the covariance matrix and the mean-displacement vector elements of the Gaussian distribution of the ensemble under tools such as GKP-Steane error-correction and fusion operations that can be used to grow large high-fidelity GKP-qubit graph states. The covariance matrix elements capture the noise in the graph state due to the finite-energy approximation of GKP qubits, while the mean displacements relate to the possible absolute shift errors on the individual qubits arising conditionally from the homodyne measurements that are a part of these tools. Our work thus pins down an exact coherent error model for graph states generated from truly finite-energy GKP qubits, which can shed light on their error-correction properties.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Damage and Failure in a Statistical Crack Model

In this paper I will take a close look at a statistical crack model (SCM) as is used in engineering computer codes to simulate fracture at high strain rates. My general goal is to understand the macroscopic behavior effected by the microphysical processes incorporated into an SCM. More specifically, I will assess the importance of including local interactions between cracks into the growth laws of an SCM. My strategy will be to construct a numerical laboratory that represents a single computational cell containing a realization of a statistical distribution of cracks. The cracks will evolve by the microphysical models of the SCM, leading to quantifiable damage and failure of the computational cell. I will use the numerical data generated by randomly generated ensembles of the fracture process to establish scaling laws that will modify and simplify the implementation of the SCM into large scale engineering codes.

42 ENGINEERING↗

Behavior of Filters and Smoothers for Strongly Nonlinear Dynamics

The Kalman filter is the optimal filter in the presence of known gaussian error statistics and linear dynamics. Filter extension to nonlinear dynamics is non trivial in the sense of appropriately representing high order moments of the statistics. Monte Carlo, ensemble-based, methods have been advocated as the methodology for representing high order moments without any questionable closure assumptions. Investigation along these lines has been conducted for highly idealized dynamics such as the strongly nonlinear Lorenz model as well as more realistic models of the means and atmosphere. A few relevant issues in this context are related to the necessary number of ensemble members to properly represent the error statistics and, the necessary modifications in the usual filter situations to allow for correct update of the ensemble members. The ensemble technique has also been applied to the problem of smoothing for which similar questions apply. Ensemble smoother examples, however, seem to be quite puzzling in that results state estimates are worse than for their filter analogue. In this study, we use concepts in probability theory to revisit the ensemble methodology for filtering and smoothing in data assimilation. We use the Lorenz model to test and compare the behavior of a variety of implementations of ensemble filters. We also implement ensemble smoothers that are able to perform better than their filter counterparts. A discussion of feasibility of these techniques to large data assimilation problems will be given at the time of the conference.

Zhu, Yanqui↗

The Behavior of Filters and Smoothers for Strongly Nonlinear Dynamics

The Kalman filter is the optimal filter in the presence of known Gaussian error statistics and linear dynamics. Filter extension to nonlinear dynamics is non trivial in the sense of appropriately representing high order moments of the statistics. Monte Carlo, ensemble-based, methods have been advocated as the methodology for representing high order moments without any questionable closure assumptions (e.g., Miller 1994). Investigation along these lines has been conducted for highly idealized dynamics such as the strongly nonlinear Lorenz (1963) model as well as more realistic models of the oceans (Evensen and van Leeuwen 1996) and atmosphere (Houtekamer and Mitchell 1998). A few relevant issues in this context are related to the necessary number of ensemble members to properly represent the error statistics and, the necessary modifications in the usual filter equations to allow for correct update of the ensemble members (Burgers 1998). The ensemble technique has also been applied to the problem of smoothing for which similar questions apply. Ensemble smoother examples, however, seem to quite puzzling in that results of state estimate are worse than for their filter analogue (Evensen 1997). In this study, we use concepts in probability theory to revisit the ensemble methodology for filtering and smoothing in data assimilation. We use Lorenz (1963) model to test and compare the behavior of a variety implementations of ensemble filters. We also implement ensemble smoothers that are able to perform better than their filter counterparts. A discussion of feasibility of these techniques to large data assimilation problems will be given at the time of the conference.

Zhu, Yanqiu↗

Measurements of Turbulence Attenuation by a Dilute Dispersion of Solid Particles in Homogeneous Isotropic Turbulence

This research addresses turbulent gas flows laden with fine solid particles at sufficiently large mass loading that strong two-way coupling occurs. By two-way coupling we mean that the particle motion is governed largely by the flow, while the particles affect the gas-phase mean flow and the turbulence properties. Our main interest is in understanding how the particles affect the turbulence. Computational techniques have been developed which can accurately predict flows carrying particles that are much smaller than the smallest scales of turbulence. Also, advanced computational techniques and burgeoning computer resources make it feasible to fully resolve very large particles moving through turbulent flows. However, flows with particle diameters of the same order as the Kolmogorov scale of the turbulence are notoriously difficult to predict. Some simple flows show strong turbulence attenuation with reductions in the turbulent kinetic energy by up to a factor of five. On the other hand, some seemingly similar flows show almost no modification. No model has been proposed that allows prediction of when the strong attenuation will occur. Unfortunately, many technological and natural two-phase flows fall into this regime, so there is a strong need for new physical understanding and modeling capability. Our objective is to study the simplest possible turbulent particle-laden flow, namely homogeneous, isotropic turbulence with a uniform dispersion of monodisperse particles. We chose such a simple flow for two reasons. First, the simplicity allows us to probe the interaction in more detail and offers analytical simplicity in interpreting the results. Secondly, this flow can be addressed by numerical simulation, and many research groups are already working on calculating the flow. Our detailed data can help guide some of these efforts. By using microgravity, we can further simplify the flow to the case of no mean velocity for either the turbulence or the particles. In fact the addition of gravity as a variable parameter may help us to better understand the physics of turbulence attenuation. The experiments are conducted in a turbulence chamber capable of producing stationary or decaying isotropic turbulence with nearly zero mean flow and Taylor microscale Reynolds numbers up to nearly 500. The chamber is a 410 mm cubic box with the corners cut off to make it approximately spherical. Synthetic jet turbulence generators are mounted in each of the eight corners of the box. Each generator consists of a loudspeaker forcing a plenum and producing a pulsed jet through a 20 mm diameter orifice. These synthetic jets are directed into ejector tubes pointing towards the chamber center. The ejector tubes increase the jet mass flow and decrease the velocity. The jets then pass through a turbulence grid. Each of the eight loudspeakers is forced with a random phase and frequency. The resulting turbulence is highly Isotropic and matches typical behavior of grid turbulence. Measurements of both phases are acquired using particle image velocimetry (PIV). The gas is seeded with approximately 1 micron diameter seeding particles while the solid phase is typically 150 micron diameter spherical glass particles. A double-pulsed YAG laser and a Kodak ES-1.0 10-bit PIV camera provide the PIV images. Custom software is used to separate the images into individual images containing either gas-phase tracers or large particles. Modern high-resolution PIV algorithms are then used to calculate the velocity field. A large set of image pairs are acquired for each case, then the results are averaged both spatially and over the ensemble of acquired images. The entire apparatus is mounted in two racks which are carried aboard NASA's KC-135 Flying Microgravity Laboratory. The rack containing the turbulence chamber, the laser head, and the camera floats freely in the airplane cabin (constrained by competent NASA personnel) to minimize g-jitter.

Eaton, John↗

Performance Analysis of a Cost-Effective Electret Condenser Microphone Directional Array

Microphone directional array technology continues to be a critical part of the overall instrumentation suite for experimental aeroacoustics. Unfortunately, high sensor cost remains one of the limiting factors in the construction of very high-density arrays (i.e., arrays containing several hundred channels or more) which could be used to implement advanced beamforming algorithms. In an effort to reduce the implementation cost of such arrays, the authors have undertaken a systematic performance analysis of a prototype 35-microphone array populated with commercial electret condenser microphones. An ensemble of microphones coupling commercially available electret cartridges with passive signal conditioning circuitry was fabricated for use with the Langley Large Aperture Directional Array (LADA). A performance analysis consisting of three phases was then performed: (1) characterize the acoustic response of the microphones via laboratory testing and calibration, (2) evaluate the beamforming capability of the electret-based LADA using a series of independently controlled point sources in an anechoic environment, and (3) demonstrate the utility of an electret-based directional array in a real-world application, in this case a cold flow jet operating at high subsonic velocities. The results of the investigation revealed a microphone frequency response suitable for directional array use over a range of 250 Hz - 40 kHz, a successful beamforming evaluation using the electret-populated LADA to measure simple point sources at frequencies up to 20 kHz, and a successful demonstration using the array to measure noise generated by the cold flow jet. This paper presents an overview of the tests conducted along with sample data obtained from those tests.

Humphreys, William M., Jr.↗

Subseasonal Representation and Predictability of North American Weather Regimes Using Cluster Analysis

Abstract This study focuses on assessing the representation and predictability of North American weather regimes, which are persistent large-scale atmospheric patterns, in a set of initialized subseasonal reforecasts created using the Community Earth System Model, version 2 (CESM2). The k -means clustering was used to extract four key North American (10°–70°N, 150°–40°W) weather regimes within ERA5 reanalysis, which were used to interpret CESM2 subseasonal forecast performance. Results show that CESM2 can recreate the climatology of the four main North American weather regimes with skill but exhibits biases during later lead times with overoccurrence of the West Coast high regime and underoccurrence of the Greenland high and Alaskan ridge regimes. Overall, the West Coast high and Pacific trough regimes exhibited higher predictability within CESM2, partly related to El Niño. Despite biases, several reforecasts were skillful and exhibited high predictability during later lead times, which could be partly attributed to skillful representation of the atmosphere from the tropics to extratropics upstream of North America. The high predictability at the subseasonal time scale of these case-study examples was manifested as an “ensemble realignment,” in which most ensemble members agreed on a prediction despite ensemble trajectory dispersion during earlier lead times. Weather regimes were also shown to project distinct temperature and precipitation anomalies across North America that largely agree with observational products. This study further demonstrates that unsupervised learning methods can be used to uncover sources and limits of subseasonal predictability, along with systematic biases present in numerical prediction systems. Significance Statement North American weather regimes are large-scale atmospheric patterns that can persist for several days. Their skillful subseasonal (2 weeks or greater) prediction can provide valuable lead time to prepare for temperature and precipitation anomalies that can stress energy and water resources. The purpose of this study was to assess the climatological representation and subseasonal predictability of four key North American weather regimes using a research subseasonal prediction system and clustering analysis. We found that the Pacific trough and West Coast high regimes exhibited higher predictability than other regimes and that skillful representation of conditions across the tropics and extratropics can increase predictability during later lead times. Future work will quantify causal pathways associated with high predictability.

58 GEOSCIENCES↗

Forecasting forecast skill

It is shown that it is possible to predict the skill of numerical weather forecasts - a quantity which is variable from day to day and region to region. This has been accomplished using as predictor the dispersion (measured by the average correlation) between members of an ensemble of forecasts started from five different analyses. The analyses had been previously derived for satellite-data-impact studies and included, in the Northern Hemisphere, moderate perturbations associated with the use of different observing systems. When the Northern Hemisphere was used as a verification region, the prediction of skill was rather poor. This is due to the fact that such a large area usually contains regions with excellent forecasts as well as regions with poor forecasts, and does not allow for discrimination between them. However, when regional verifications were used, the ensemble forecast dispersion provided a very good prediction of the quality of the individual forecasts.

Kalnay, Eugenia↗

Scaling functions for eigenstate entanglement crossovers in harmonic lattices

For quantum matter, eigenstate entanglement entropies obey an area law or log-area law at low energies and small subsystem sizes and cross over to volume laws for high energies and large subsystems. This transition is captured by crossover functions, which assume a universal scaling form in quantum critical regimes. We demonstrate this for the harmonic lattice model, which describes quantized lattice vibrations and is a regularization for free scalar field theories, modeling, e.g., spin-0 bosonic particles. In one dimension, the ground-state entanglement obeys a log-area law. For dimensions d ≥ 2, it displays area laws, even at criticality. The distribution of excited-state entanglement entropies is found to be sharply peaked around subsystem entropies of corresponding thermodynamic ensembles in accordance with the eigenstate thermalization hypothesis. Numerically, we determine crossover scaling functions for the quantum critical regime of the model and do a large-deviation analysis. We show how infrared singularities of the system can be handled and how to access the thermodynamic limit using a perturbative trick for the covariance matrix. Eigenstates for quasi-free bosonic systems are not Gaussian. Here, we resolve this problem by considering appropriate squeezed states instead. For these, entanglement entropies can be evaluated efficiently.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Is land use producing robust signals in future projections from Earth system models, all else being equal?

Abstract We use six Earth system models (ESMs) run under SSP3-7.0, a scenario characterized by a relatively large land use change (LUC) over the 21st century, and under a variant of the same scenario where a significantly different pattern of LUC, taken from SSP1-2.6, was used, all else being equal. Our goal is to identify changes in climate extremes between the two scenarios that are statistically significant and robust across the ESMs. The motivation for this study is to test a long-held assumption of the shared socio-economic pathway-representative concentration pathway (SSP-RCP) scenario framework: that the signal from LUC can be safely disregarded when pairing different SSPs to the compatible RCPs, where compatibility only considers global radiative forcing, predominantly determined by well-mixed greenhouse gasses emissions. We analyze extremes of daily minimum and maximum temperatures and precipitation, after fitting non-stationary generalized extreme value distributions in a way that borrows strength along the length of the simulation (2015–2100) and across initial condition ensembles. We consider changes in the 20 year return levels (RL20s) of these metrics by 2100, and focus on eight locations where LUC is large within each scenario, and strongly differs between scenarios, averaging the RL20s over a neighborhood characterized by the same LUC to enhance the signal to noise. We find that precipitation extremes do not show significant differences attributable to LUC differences. For temperature extremes (cold and hot) results are mixed, with some location-index combination showing significant results for some of the ESMs but not all, and not many coherent changes appearing for indices across regions, or regions across indices. These ESMs are representative of what is typically adopted as the source of climate information for impact studies, when the SSP-RCP framework is put to use. Overall, our analysis suggests that the hypothesis to pair SSPs to RCPs in a flexible fashion is overall defensible. However, the appearance of some coherence in a few locations and for some indices invites further investigation.

54 ENVIRONMENTAL SCIENCES↗

The Effect of Global Warming on Severe Thunderstorms in the United States

How will warming temperatures influence thunderstorm severity? This question can be explored by using climate models to diagnose changes in large-scale convective instability (CAPE) and wind shear, conditions that are known to be conducive to the formation of severe thunderstorms. First, an ensemble of climate models from phase 5 of the Coupled Model Intercomparison Project (CMIP5) is evaluated on its ability to reproduce a radiosonde climatology of such storm-favorable conditions in the current climate's spring and summer seasons, focusing on the contiguous United States (CONUS). Of the 11 climate models evaluated, a high-performing subset of four (GFDL CM3, GFDL-ESM2M, MRI-CGCM3, and NorESM1-M) is identified. Second, the twenty-first-century changes in the frequency of environments favorable to severe thunderstorms are calculated in these high-performing models as they are forced by the RCP4.5 and RCP8.5 emissions pathways. For the RCP8.5 scenario, the models predict consistent CONUS-mean fractional springtime increases in the range of 50%-180% by the end of the twenty-first century; for the summer, three of the four models predict increases in the range of 40%-120% and one model predicts a small decrease. This disagreement between the models is traced to divergent projections for future CAPE and boundary layer humidity in the Great Plains. This paper also explores the sensitivity of the results to the relative weight given to wind shear in determining how "favorable" a large-scale environment is for the development of severe thunderstorms, and it is found that this weighting is not the dominant source of uncertainty in projections of future thunderstorm severity.

54 ENVIRONMENTAL SCIENCES↗

Quantifying the Occurrence of Record Hot Years Through Normalized Warming Trends

Surface air temperature trends and extreme events are of global concern and they are related. Here we show that the occurrence of record hot years over different latitudes from 1960-2019 are more strongly correlated with the observational annual mean temperature trends normalized by internal variability. Compared with the raw trends showing Arctic amplification, the normalized trends show a tropical amplification over land. Two hot spots with more frequent occurrence of record hot years are identified: northern hemisphere ocean (versus land) and southern hemisphere tropical land (versus mid- and high-latitude lands). Ensemble mean results from 32 Earth system models agree with observations better than individual models, but they do not reproduce observed large differences in correlations across latitudes between normalized trends and record-breaking events over land versus ocean. Furthermore, our results enable the quantification of record hot year occurrence through normalized warming trends and provide new metrics for model evaluation and improvement.

58 GEOSCIENCES↗

Robust and scalable uncertainty estimation with conformal prediction for machine-learned interatomic potentials

Abstract Uncertainty quantification (UQ) is important to machine learning (ML) force fields to assess the level of confidence during prediction, as ML models are not inherently physical and can therefore yield catastrophically incorrect predictions. Established a-posteriori UQ methods, including ensemble methods, the dropout method, the delta method, and various heuristic distance metrics, have limitations such as being computationally challenging for large models due to model re-training. In addition, the uncertainty estimates are often not rigorously calibrated. In this work, we propose combining the distribution-free UQ method, known as conformal prediction (CP), with the distances in the neural network’s latent space to estimate the uncertainty of energies predicted by neural network force fields. We evaluate this method (CP+latent) along with other UQ methods on two essential aspects, calibration, and sharpness, and find this method to be both calibrated and sharp under the assumption of independent and identically-distributed (i.i.d.) data. We show that the method is relatively insensitive to hyperparameters selected, and test the limitations of the method when the i.i.d. assumption is violated. Finally, we demonstrate that this method can be readily applied to trained neural network force fields with traditional and graph neural network architectures to obtain estimates of uncertainty with low computational costs on a training dataset of 1 million images to showcase its scalability and portability. Incorporating the CP method with latent distances offers a calibrated, sharp and efficient strategy to estimate the uncertainty of neural network force fields. In addition, the CP approach can also function as a promising strategy for calibrating uncertainty estimated by other approaches.

74 ATOMIC AND MOLECULAR PHYSICS↗