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.
The Allan variance for an averaging time tau can be estimated either from all available phase samples or from a subgrid of samples with spacing tau. This paper gives a set of computational recipes that yield the variance of both estimators, with less than 2 percent error, for the five power-law components of the classical continuous-time clock noise model.
The overachieving goal of the whole CMDV-MCS project is to improve understanding of warm season continental convection and to develop treatments of convection and microphysics capable of representing mesoscale convective systems (MCSs) features in large-scale models. Our tasks for this project contributing to the overachieving goal include: (1) Improve the ice nucleation formulation for MG2 and P3 cloud microphysics schemes; (2) Improve the treatment of subgrid dynamics and thermodynamics driving the ice nucleation in E3SM; and (3) Test the performance of improved ice microphysics in E3SM with observation data. In this project, we have (1) Improved the ice nucleation parameterization for MG2 and P3 in E3SM by implementing two advanced empirical parameterizations with connection to aerosols. The two deterministic heterogeneous ice nucleation parameterizations (i.e., DeMott et al., 2015; Niemand et al., 2012) were merged with the MG2 and P3 cloud microphysics schemes in E3SM. Long-term simulations were conducted to examine the impacts of these new parameterizations on simulated cloud properties; (2) Improved the treatment of subgrid dynamics and thermodynamics driving the ice nucleation in E3SM. We evaluated the double Gaussian PDF of vertical velocity simulated by the Cloud Layers Unified By Binormals (CLUBB) and the sub-column vertical velocity sampled from the Subgrid Importance Latin Hypercube Sampler (SILHS) in E3SM. We introduced the vertical velocity variance induced by topographic gravity waves for ice nucleation and droplet activation; and (3) Tested the performance of improved ice microphysics in E3SM with observation data. We tested the new treatments of ice nucleation in the single column model (SCM) mode for the stratiform mixed-phase clouds observed during 9-10 October 2004 in the DOE ARM Mixed-Phase Arctic Cloud Experiment (M-PACE) and for the convective clouds observed on 20 May 2011 in the Midlatitude Continental Convective Clouds Experiment (MC3E). Modeled ice nucleating particles (INPs) concentrations were compared against observations collected around the globe.
Foam materials are starting to find application in laser-heated hohlraums used to drive inertial confinement fusion (ICF) implosions. The disparate scales of the solid foam elements (nm) compared to the typical problem scale (mm) make it expensive to resolve the individual foam elements in hydrodynamic calculations. In this project we developed two new models for treating foam materials – a 3D hydrodynamic model that treats the foam as bits of pre-expanded solid material separated by voids, and a subgrid model that captures the essential physics of laser-heated foams without requiring resolution of the individual solid elements. We also explored the microphysics of the transition from interpenetrating kinetic behavior to fully-collisional fluid-like behavior during the early time heating and expansion of foam elements through a combination of single-fluid and multi-fluid hydrodynamics calculations and particle-in-cell calculations. The model was benchmarked against these calculations and against experimental data on laser-heated chemical and additive-manufactured (AM) foams. Since there was no existing data on laser-heated AM foams, we performed the first experiments on such foams at the Jupiter Laser Facility. Samples of three different types of printed AM foams were heated using a single 0.5 micron laser. The foam densities ranged from 10 to 100 mg/cc (supercritical for 0.5 micron light). We measured the backscattered light (power and spectrum), the transmitted light, and the time-integrated temperature of the foam samples. This benchmarked subgrid foam model is a new capability that should allow routine calculation of ICF designs using foam elements, potentially opening up a new design space.
In atmospheric models, stochastic generation of subgrid-scale profiles or “subcolumns” has been used for a variety of purposes. Such subcolumns can be generated from subgrid probability density functions (PDFs) at different vertical levels, when such PDFs are available. To do so, the generator needs to decide how strongly points should be correlated in the vertical, that is, how much the values should be overlapped. This is sometimes called “PDF overlap.” To assess vertical correlation in a simplified, observable setting, here the vertical correlation of vertical velocity in subcloud layers is examined. Doppler lidar is used to evaluate the vertical profiles of vertical velocity produced by a large-eddy simulation (LES) model and the Subgrid Importance Latin Hypercube Sampler (SILHS) subcolumn generator. In order to diagnose unrealistic features in subcolumn profiles, various statistical diagnostics are examined here, including the bivariate PDF of vertical velocity at two separated points (i.e., altitudes), the two-point velocity correlation, the integral correlation length, the PDF of two-point velocity differences, and the skewness and kurtosis of two-point velocity differences. The profiles produced by LES match lidar well, except that they are too smooth at small scales. The profiles produced by SILHS exhibit sharp jumps from updraft to downdraft that are not observed in the lidar data. To reduce the generation of these unrealistically sharp jumps, the SILHS sampling method is revised. The diagnostics confirm that the revised sampling method reduces the overprediction of sharp jumps.
As part of the Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC), the HELiX uncrewed aircraft system (UAS) was deployed over the sea ice in the central Arctic Ocean during summer 2020. Albedo measurements were obtained with stabilized pyranometers, and melt pond fraction was calculated from orthomosaic imagery from a surface-imaging multispectral camera. This study analyzed HELiX flight data to provide insights on the temporal and spatial evolution of albedo and melt pond fraction of the MOSAiC floe during the melt season as it drifted south through Fram Strait. The surface albedo distributions showed peak values changing from high albedo (0.55–0.6) to lower values (0.3) as the season advanced. Inspired by methods developed for satellite data, an algorithm was established to retrieve melt pond fraction from the orthomosaic images. We demonstrate that the near-surface observations of melt pond fraction were highly dependent on sample area, offering insight into the influence of subgrid scale features and spatial heterogeneity in satellite observations. Vertical observations conducted with the HELiX were used to quantify the influence of melt pond scales on observed surface albedo as a function of sensor footprint. These scaling results were used to link surface-based measurements collected during MOSAiC to broader-scale satellite data to investigate the influence of surface features on observed albedo. Albedo values blend underlying features within the sensor footprint, as determined by the melt pond size and concentration. This study framed the downscaling (upscaling) problem related to the airborne (surface) observations of surface albedo across a variety of spatial scales.
One of the most exciting recent developments in the field of large eddy simulation (LES) is the dynamic subgrid-scale model. The dynamic model concept is a general procedure for evaluating model constants by sampling a band of the smallest scales actually resolved in the simulation. To date, the procedure has been used primarily in conjunction with the Smagorinsky model. The dynamic procedure has the advantage that the value of the model constant need not be specified a priori, but rather is calculated as a function of space and time as the simulation progresses. This feature makes the dynamic model especially attractive for flows in complex geometries where it is difficult or impossible to calibrate model constants. The dynamic model was highly successful in benchmark tests involving homogeneous and channel flows. Having demonstrated the potential of the dynamic model in these simple flows, the overall direction of the LES effort at CTR shifted toward an evaluation of the model in more complex situations. The current test cases are basic engineering-type flows for which Reynolds averaged approaches were unable to model the turbulence to within engineering accuracy. Flows currently under investigation include a backward-facing step, wake behind a circular cylinder, airfoil at high angles of attack, separated flow in a diffuser, and boundary layer over a concave surface. Preliminary results from the backward-facing step and cylinder wake simulations are encouraging. Progress on the LES of a boundary layer on a concave surface is discussed. Although the geometry of a concave wall is not very complex, the boundary layer that develops on its surface is difficult to model due to the presence of streamwise Taylor-Gortler vortices. These vortices arise as a result of a centrifugal instability associated with the convex curvature.
Subgrid horizontal variations of land surface heat fluxes to the atmosphere resulting from subgrid land cover heterogeneity are important in land-atmosphere interaction in global climate models (GCMs). To incorporate it, a parameterization using stochastic sampling based on truncated normal distributions diagnosed from the land model and internal ensemble mean of multiple calls to the planetary boundary layer and deep convection schemes is developed. After its implementation in the National Center for Atmospheric Research (NCAR) Community Earth System Model version 1.2 (CESM1.2), large changes of simulated land precipitation are found to occur in regions where large subgrid variations of surface heat fluxes exist. The simulated precipitation is improved in boreal summer, especially over eastern China and the coastal areas of the Bay of Bengal. The improved precipitation is mainly a result of improved large-scale moisture convergence and advection by altered vertical diffusion and convection.
The intensification of extreme precipitation under anthropogenic forcing is robustly projected by global climate models, but highly challenging to detect in the observational record. Large internal variability distorts this anthropogenic signal. Models produce diverse magnitudes of precipitation response to anthropogenic forcing, largely due to differing schemes for parameterizing subgrid-scale processes. Meanwhile, multiple global observational datasets of daily precipitation exist, developed using varying techniques and inhomogeneously sampled data in space and time. Previous attempts to detect human influence on extreme precipitation have not incorporated model uncertainty, and have been limited to specific regions and observational datasets. Using machine learning methods that can account for these uncertainties and capable of identifying the time evolution of the spatial patterns, we find a physically interpretable anthropogenic signal that is detectable in all global observational datasets. Machine learning efficiently generates multiple lines of evidence supporting detection of an anthropogenic signal in global extreme precipitation.
Arctic low clouds influence the evolution of the Arctic system through their effects on radiative fluxes, boundary layer mixing, stability, turbulence, humidity, and precipitation. Unfortunately, atmospheric models and retrospective analysis (reanalysis) products struggle to accurately simulate the occurrence and properties of low clouds in the Arctic. To better understand the magnitude and possible causes of this problem, we analyze the disagreements between the Arctic System Reanalysis version 2 (ASR) and data taken during the September 2014 Arctic Radiation-IceBridge Sea and Ice Experiment (ARISE) airborne campaign conducted over the Beaufort Sea. Previous analysis of grid scale-averaged meteorological variables revealed a lack of cloud water produced in ASR, related to warm/dry biases, a failure to produce sufficient cloud water in conditions that are observed to be favorable for large cloud water values, and sampling issues with the ARISE flight paths. However, ARISE collected meteorological observations at one second intervals, offering spatial data resolution comparable to the subgrid scale of ASR. Subgrid scale properties and processes must be parameterized in conventional atmospheric models and reanalyses such as ASR, and often rely on assumed distributions of meteorological variables about the grid box mean state. To enable an examination of the realism of ASR’s subgrid scale properties, we use high resolution data from ARISE determine the “subgrid scale” distributions of cloud water and thermodynamic variables in the real atmosphere. Preliminary results show that the subgrid scale distribution of thermodynamic variables about the grid scale mean resembles the triangular distribution assumed in many bulk microphysical parameterization schemes, but also with tails resembling Gaussian distributions. The main goal is to understand the partitioning of the total available atmospheric water into water vapor and cloud water. Models typically covert some percentage of total water in a grid box to cloud water based on the estimated amount of supersaturation from the assumed vapor distribution. The ARISE measurements can reveal the relationship between subgrid scale supersaturation and the partition between cloud water and water vapor.
Predicting realistic cloud subgrid variability remains a challenge for Global Climate Models (GCMs) even though it can play an important role for proper representation of processes pertaining to cloud microphysics, precipitation, and radiation, but also for comparisons with satellite observations which are of much higher resolution than model grids. The diagnostic approach of subgrid cloud variability is commonly handled by subcolumn cloud generators. For a specific GCM, one ideally wants the subgrid variability used for comparisons with satellite observations to be created by the same generator and with the same rules as the one used for model integration. With this in mind, we have embarked in an effort to test and improve cloud subcolumn generators appropriate for GCMs. For this purpose, we use cloud (hydrometeor) products from active observations by the CloudSat radar (CPR) and the CALIPSO lidar (CALIOP). Cloud products from active sensors while suffering significant sampling and coverage drawbacks have the advantage of resolving both horizontal and vertical variability. The main question is: given a profile of cloud condensate mean and variance, can we create a subgrid cloud field that is statistically similar to the observed subgrid cloud field? By “statistically”, we suggest that we do not aspire to reproduce “well” each individual subgrid cloud field of a GCM-scale region, but that our generator performs well for a large ensemble of cases. We simulate radar, passive imager, and radiation flux fields from the observed 2D cloud fields and create one-point statistics; we use the profiles of cloud condensate mean and variance as input to the generator to create subgrid cloud fields; we compare the statistics; we adjust the rules of the generator to create the best possible agreement between the radar, imager and radiation field statistics. In this process, the active observations have actually a dual role: they provide the actual subgrid cloud field which can be reduced to a profile of mean and variance used by the subcolumn generator, but they also provide the rules needed by the generator, such as measures (e.g, decorrelation length) of the vertical overlap of cloud fraction and of the condensate horizontal variability. This presentation will show our progress using Cloudsat and CALIPSO products in this dual fashion.
Predicting realistic cloud subgrid variability remains a challenge for Global Climate Models (GCMs) even though it can play an important role for proper representation of processes pertaining to cloud microphysics, precipitation, and radiation, but also for comparisons with satellite observations which are of much higher resolution than model grids. The diagnostic approach of subgrid cloud variability is commonly handled by subcolumn cloud generators. For a specific GCM, one ideally wants the subgrid variability used for comparisons with satellite observations to be created by the same generator and with the same rules as the one used for model integration. With this in mind, we have embarked in an effort to test and improve cloud subcolumn generators appropriate for GCMs. For this purpose, we use cloud (hydrometeor) products from active observations by the CloudSat radar (CPR) and the CALIPSO lidar (CALIOP). Cloud products from active sensors while suffering signicant sampling and coverage drawbacks have the advantage of resolving both horizontal and vertical variability. The main question is: given a profile of cloud condensate mean and variance, can we create a subgrid cloud field that is statistically similar to the observed subgrid cloud field? By “statistically”, we suggest that we do not aspire to reproduce “well” each individual subgrid cloud field of a GCM-scale region, but that our generator performs well for a large ensemble of cases. We simulate radar, passive imager, and radiation flux fields from the observed 2D cloud fields and create one-point statistics; we use the profiles of cloud condensate mean and variance as input to the generator to create subgrid cloud fields; we compare the statistics; we adjust the rules of the generator to create the best possible agreement between the radar, imager and radiation field statistics. In this process, the active observations have actually a dual role: they provide the actual subgrid cloud field which can be reduced to a profile of mean and variance used by the subcolumn generator, but they also provide the rules needed by the generator, such as measures (e.g, decorrelation length) of the vertical overlap of cloud fraction and of the condensate horizontal variability. This presentation will show our progress using Cloudsat and CALIPSO products in this dual fashion.
Shaped emitters are of interest to a broad range of applications in vacuum electronic devices. In particular, thermionic energy converters (TECs) take advantage of shaped emitters to increase the local surface field, thereby extracting more current for a given cathode temperature and applied voltage. However, modeling these devices is challenging; Warp [J.-L. Vay, D. P. Grote, R. H. Cohen, and A. Friedman, Comput. Sci. Discov. 5, 014019 (2012)] is a fully 3D particle-in-cell code capable of handling a wide range of physics problems and is well suited to modeling TECs. Additionally, recent improvements to Warp have enabled the accurate modeling of emitters with arbitrary curved surfaces. Specifically, the inclusion of subgrid resolution for computing the electrostatic potential and the ability to apply mesh refinement for specific areas of interest allow for a more accurate solution to the fields on these surfaces. These improvements coupled with Warp’s ability to handle variable particle weights make it an ideal candidate for simulating these complex devices. In this paper, the authors study the applicability of different subgrid configurations for simulating shaped emission surfaces and field convergence for different mesh-refinement techniques. They then implement a custom weighting algorithm that allows for uniform sampling of emission surfaces with a large variation in the surface electric field. They then use this algorithm to study emission for curved emitters in both the field-enhancement regime and the space-charge regime.
Abstract Machine learning models have been employed to perform either physics-free data-driven or hybrid dynamical downscaling of climate data. Most of these implementations operate over relatively small downscaling factors because of the challenge of recovering fine-scale information from coarse data. This limits their compatibility with many global climate model outputs, often available between ∼50- and 100-km resolution, to scales of interest such as cloud resolving or urban scales. This study systematically examines the capability of a type of superresolving convolutional neural network (SR-CNNs) to downscale surface wind speed data over land from different coarse resolutions (25-, 48-, and 100-km resolution) to 3 km. For each downscaling factor, we consider three convolutional neural network (CNN) configurations that generate superresolved predictions of fine-scale wind speed, which take between one and three input fields: coarse wind speed, fine-scale topography, and diurnal cycle. In addition to fine-scale wind speeds, probability density function parameters are generated through which sample wind speeds can be generated, accounting for the intrinsic stochasticity of wind speed. For assessing generalization to new data, CNN models are tested on regions with different topography and climate that are unseen during training. The evaluation of superresolved predictions focuses on subgrid-scale variability and the recovery of extremes. Models with coarse wind and fine topography as inputs exhibit the best performance when compared with other model configurations, operating across the same downscaling factor. Our diurnal cycle encoding results in lower out-of-sample generalizability when compared with other input configurations.
We report that the weights of a deep neural-network model are optimized in conjunction with the governing flow equations to provide a model for subgrid-scale stresses in a temporally developing plane turbulent jet at Reynolds number Re 0 = 6000 . The objective function for training is first based on the instantaneous filtered velocity fields from a corresponding direct numerical simulation, and the training is by a stochastic gradient descent method, which uses the adjoint Navier-Stokes equations to provide the end-to-end sensitivities of the model weights to the velocity fields. In-sample and out-of-sample testing on multiple dual-jet configurations show that its required mesh density in each coordinate direction for prediction of mean flow, Reynolds stresses, and spectra is half that needed by the dynamic Smagorinsky model for comparable accuracy. The same neural-network model trained directly to match filtered subgrid-scale stresses, without the constraint of being embedded within the flow equations during the training, fails to provide a qualitatively correct prediction. The coupled formulation is generalized to train based only on mean-flow and Reynolds stresses, which are more readily available in experiments. The mean-flow training provides a robust model, which is important, though a somewhat less accurate prediction for the same coarse meshes, as might be anticipated due to the reduced information available for training in this case. The anticipated advantage of the formulation is that the inclusion of resolved physics in the training increases its capacity to extrapolate. This is assessed for the case of passive scalar transport, for which it outperforms established models due to improved mixing predictions.
Column CO2 observations from current and future remote sensing missions represent a major advancement in our understanding of the carbon cycle and are expected to help constrain source and sink distributions. However, data assimilation and inversion methods are challenged by the difference in scale of models and observations. OCO-2 footprints represent an area of several square kilometers while NASA s future ASCENDS lidar mission is likely to have an even smaller footprint. In contrast, the resolution of models used in global inversions are typically hundreds of kilometers wide and often cover areas that include combinations of land, ocean and coastal areas and areas of significant topographic, land cover, and population density variations. To improve understanding of scales of atmospheric CO2 variability and representativeness of satellite observations, we will present results from a global, 10-km simulation of meteorology and atmospheric CO2 distributions performed using NASA s GEOS-5 general circulation model. This resolution, typical of mesoscale atmospheric models, represents an order of magnitude increase in resolution over typical global simulations of atmospheric composition allowing new insight into small scale CO2 variations across a wide range of surface flux and meteorological conditions. The simulation includes high resolution flux datasets provided by NASA s Carbon Monitoring System Flux Pilot Project at half degree resolution that have been down-scaled to 10-km using remote sensing datasets. Probability distribution functions are calculated over larger areas more typical of global models (100-400 km) to characterize subgrid-scale variability in these models. Particular emphasis is placed on coastal regions and regions containing megacities and fires to evaluate the ability of coarse resolution models to represent these small scale features. Additionally, model output are sampled using averaging kernels characteristic of OCO-2 and ASCENDS measurement concepts to create realistic pseudo-datasets. Pseudo-data are averaged over coarse model grid cell areas to better understand the ability of measurements to characterize CO2 distributions and spatial gradients on both short (daily to weekly) and long (monthly to seasonal) time scales
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.
A large-eddy simulation (LES) code was developed at the NASA Glenn Research Center to provide more accurate and detailed computational analyses of propulsion flow fields. The accuracy of current computational fluid dynamics (CFD) methods is limited primarily by their inability to properly account for the turbulent motion present in virtually all propulsion flows. Because the efficiency and performance of a propulsion system are highly dependent on the details of this turbulent motion, it is critical for CFD to accurately model it. The LES code promises to give new CFD simulations an advantage over older methods by directly computing the large turbulent eddies, to correctly predict their effect on a propulsion system. Turbulent motion is a random, unsteady process whose behavior is difficult to predict through computer simulations. Current methods are based on Reynolds-Averaged Navier- Stokes (RANS) analyses that rely on models to represent the effect of turbulence within a flow field. The quality of the results depends on the quality of the model and its applicability to the type of flow field being studied. LES promises to be more accurate because it drastically reduces the amount of modeling necessary. It is the logical step toward improving turbulent flow predictions. In LES, the large-scale dominant turbulent motion is computed directly, leaving only the less significant small turbulent scales to be modeled. As part of the prediction, the LES method generates detailed information on the turbulence itself, providing important information for other applications, such as aeroacoustics. The LES code developed at Glenn for propulsion flow fields is being used to both analyze propulsion system components and test improved LES algorithms (subgrid-scale models, filters, and numerical schemes). The code solves the compressible Favre-filtered Navier- Stokes equations using an explicit fourth-order accurate numerical scheme, it incorporates a compressible form of Smagorinsky s model for the subgrid-scale turbulence, and it uses generalized curvilinear coordinates to allow analysis of a wide range of geometries. The code runs in parallel on shared memory multiprocessor computers and is written in Fortran 90 with dynamic memory allocation. A sample result for a Mach-1.4 round jet is presented in the figure. Instantaneous Mach number contours in several cross-planes downstream of the nozzle exit are shown, illustrating how an LES captures the large unsteady three-dimensional turbulent structures present in the jet.
Abstract Neural networks (NNs) are increasingly used for data‐driven subgrid‐scale parameterizations in weather and climate models. While NNs are powerful tools for learning complex non‐linear relationships from data, there are several challenges in using them for parameterizations. Three of these challenges are (a) data imbalance related to learning rare, often large‐amplitude, samples; (b) uncertainty quantification (UQ) of the predictions to provide an accuracy indicator; and (c) generalization to other climates, for example, those with different radiative forcings. Here, we examine the performance of methods for addressing these challenges using NN‐based emulators of the Whole Atmosphere Community Climate Model (WACCM) physics‐based gravity wave (GW) parameterizations as a test case. WACCM has complex, state‐of‐the‐art parameterizations for orography‐, convection‐, and front‐driven GWs. Convection‐ and orography‐driven GWs have significant data imbalance due to the absence of convection or orography in most grid points. We address data imbalance using resampling and/or weighted loss functions, enabling the successful emulation of parameterizations for all three sources. We demonstrate that three UQ methods (Bayesian NNs, variational auto‐encoders, and dropouts) provide ensemble spreads that correspond to accuracy during testing, offering criteria for identifying when an NN gives inaccurate predictions. Finally, we show that the accuracy of these NNs decreases for a warmer climate (4 × CO 2 ). However, their performance is significantly improved by applying transfer learning, for example, re‐training only one layer using ∼1% new data from the warmer climate. The findings of this study offer insights for developing reliable and generalizable data‐driven parameterizations for various processes, including (but not limited to) GWs.