Search NASA⌕ Search

SEARCH · Search NASA

Results for “Error Mitigation”

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 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Development of a Practical Secondary Control for Hardware Microgrids

Practical, vendor-agnostic interoperability guidelines for the secondary control architecture of microgrids (MGs) with multiple grid-forming (GFM) inverter-based resources (IBRs) have not yet been developed. Therefore, this paper proposes a generic and vendor-agnostic secondary control architecture that operates with all GFM IBRs and synchronous generators. This secondary control does not require the use of additional measurement devices in the MG and utilizes the inherent communication systems of the GFM units, such as Modbus TCP/IP. The practical challenges of Modbus registers, such as packet loss and quantization error, and their detrimental impacts on secondary control actions are investigated. The proposed three-stage modification for any secondary control architecture to mitigate these impacts includes: 1) averaging the data read, 2) situational event-triggering of the controller, and 3) finite iteration of the controlling action. The proposed method is validated using a 3-..phi.., 480 V, 60 Hz, 500 kVA laboratory hardware microgrid with commercial two GFM IBRs and one diesel generator. The experimental results corroborates the fact the proposed modification in the secondary control architecture is advantageous for practical usage under erroneous measurements.

communication systems↗
NASA NTRSDate not supplied

Optical Modem Enabling Broadband Datacom Links for Crewed Cis-Lunar Missions

We report on the design, development, and testing of our high-power broadband optical modem supporting NASA’s crewed Artemis-2 mission. The O2O modem will be mounted in the crewed Orion module and provide a broadband 505,000 km bi-directional optical link back to earth while en route to the moon. The full-duplex modem consists of a high-power optical transmitter and receiver optimized for serially-concatenated pulse-position modulation (SCPPM). The transmitter is a master-oscillator power-amplifier optical architecture using efficient cladding-pumped amplification in erbium-ytterbium co-doped fiber. The transmitter outputs up to 1 W at ≈1550 nm (limited for eye safety) and supports 6 different user-rates ranging from 20.39 Mb/s to 260.95 Mb/s using PPM16 and PPM32 modulation formats. The optical receiver supports two user-rates: 10.19 Mb/s and 20.39 Mb/s with both rates employing PPM32. The narrowband receiver filtering is optimized to simultaneously accept four separate wavelength channels to mitigate atmospherics through spatial diversity. A configurable interleaver provides additional protection against atmospherics-based signal fading and a powerful soft-decision error correction scheme enables highly sensitive detection. The measured sensitivities at the two bit-rates are -73.8 and -71.8dBm, respectively. The architecture was designed for reliable operation in space, featuring automatic hardware interlocks, pump sparing for the amplifiers, and autonomous operation of all internal hardware and software control loops. The protoflight unit (PFU) was put through rigorous environmental testing which included pyroshock, vibration, electromagnetic interference/compatibility, and thermal-vacuum testing. The modem successfully passed all the environmental screening and has been declared at Technology Readiness Level (TRL) 6.

space communications↗
DOE OSTI2025

Mitigating spectral bias in neural operators via high-frequency scaling for physical systems

Neural operators have emerged as powerful surrogates for modeling complex physical problems. However, they suffer from spectral bias making them oblivious to high-frequency modes, which are present in multiscale physical systems. Therefore, they tend to produce over-smoothed solutions, which is particularly problematic in modeling turbulence and for systems with intricate patterns and sharp gradients such as multi-phase flow systems. In this work, we introduce a new approach named high-frequency scaling (HFS) to mitigate spectral bias in convolutional-based neural operators. By integrating HFS with proper variants of UNet, we demonstrate a higher prediction accuracy by mitigating spectral bias in single and two-phase flow problems. Unlike Fourierbased techniques, HFS is directly applied to the latent space, thus eliminating the computational cost associated with the Fourier transform. Additionally, we investigate alternative spectral bias mitigation through a diffusion model conditioned on neural operators. While the diffusion model integrated with the standard neural operator may still suffer from significant errors, these errors are substantially reduced when the diffusion model is integrated with a HFS-enhanced neural operator.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

2x2 Demonstrator High-Coverage Light Readout System in a Novel LArTPC

DUNE is a long-baseline Liquid Argon Time Projection Chamber (LArTPC) neutrino oscillation experiment. Its near detectors, located at Fermi National Laboratory, will constrain systematic errors on far detector measurements. The 2x2 Demonstrator, a novel proof of concept prototype for the ND LArTPC, has a modular TPC design that mitigates event pileup in a high-occupancy nearline beam environment, further aided by a native 3D pixelated charge readout and a 30% coverage, low profile light readout. This will be the first large-scale LArTPC to utilize pixels in place of wire planes

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
NASA NTRS2008

Low-Light-Shift Cesium Fountain without Mechanical Shutters

A new technique for reducing errors in a laser-cooled cesium fountain frequency standard provides for strong suppression of the light shift without need for mechanical shutters. Because mechanical shutters are typically susceptible to failure after operating times of the order of months, the elimination of mechanical shutters could contribute significantly to the reliability of frequency standards that are required to function continuously for longer time intervals. With respect to the operation of an atomic-fountain frequency standard, the term "light shift" denotes an undesired relative shift in the two energy levels of the atoms (in this case, cesium atoms) in the atomic fountain during interrogation by microwaves. The shift in energy levels translates to a frequency shift that reduces the precision and possibly accuracy of the frequency standard. For reasons too complex to describe within the space available for this article, the light shift is caused by any laser light that reaches the atoms during the microwave- interrogation period, but is strongest for near-resonance light. In the absence of any mitigating design feature, the light shift, expressed as a fraction of the standard fs frequency, could be as large as approx. 2 x 10(exp -11), the largest error in the standard. In a typical prior design, to suppress light shift, the intensity of laser light is reduced during the interrogation period by using a single-pass acoustooptic modulator to deflect the majority of light away from the main optical path. Mechanical shutters are used to block the remaining undeflected light to ensure complete attenuation. Without shutters, this remaining undeflected light could cause a light shift of as much as .10.15, which is unacceptably large in some applications. The new technique implemented here involves additionally shifting the laser wavelength off resonance by a relatively large amount (typically of the order of nanometers) during microwave interrogation. In this design, when microwave interrogation is not underway, the atoms are illuminated by a slave laser locked to the lasing frequency of a lower power master laser.

Enzer, Daphna↗
NASA NTRSDate not supplied

Considerations on Electrolytic Conductivity Measurement for Monitoring of Ionic Silver Biocide Dosing

NASA interest in ionic silver (Ag+) as a biocide for spacecraft potable water systems motivates the development of Ag+ concentration sensors to ensure nominal dosing. The electrolytic conductivity change of highly-purified potable water is linearly related to the concentration of chemically-dosed Ag+ and could serve as a useful proxy measurement, while the conductivity change during electrolytic dosing may be less so, depending on the influent water chemistry and electrolytic efficiency. Understanding and mitigating the potentially deleterious effects of Ag+ interaction with conductivity measurement systems requires investigation. Issues associated with traditional conductivity cells, which rely on wetted(typically metal or graphite) electrodes, and capacitively-coupled contactless conductivity detection (C4D) for this application are considered. Traditional conductivity cells may potentially be subject to significant excitation-induced or auto-galvanic reduction of Ag+. These may result in Ag+ depletion or electrode fouling and associated measurement error. Proper selection of electrode material, excitation parameters, and cell geometry may limit such effects. C4D uses electrodes placed outside an inert, insulating material, with dielectric polarization enabling the production of an electric field and resultant current across the analyte solution. This approach could potentially mitigate problems with Ag+ depletion and fouling by eliminating the possibility of auto-galvanic deposition and reducing the Faradaic current density. However, it is necessary to confirm C4D performance at very low conductivity levels and to determine if long-term operation produces conductive deposits, which could result in measurement error. A commercial C4D system with 1/16” (1.59 mm)outer diameter flow tubing and claimed performance in the conductivity range appropriate for this application was identified, and its sensitivity in the upper parts of this range was confirmed in a preliminary experiment. A concept for a C4D detector that could allow for full rate flow-through using planar electrodes and thin-film dielectric layers is discussed.

silver↗
NASA NTRSDate not supplied

Historical Aerospace Software Errors Categorized to Influence Fault Tolerance

- Motivation - Very little literature exists characterizing software errors in real-time avionic systems - How, where, and why is software most likely to fail? - Purpose - Raise awareness of how software fails through historical study - Recommend improvements to software fault tolerant design based on historical study - Outline - Discuss Software Failures - Common Cause, Failure Classes, Mitigation strategies - Review NASA requirements for Software Fault Tolerance - Review Historical Software Failures - Analyze failures and provide statistics - Erroneous vs. fail-Silent - Reboot recoverability likelihood - Code Location - Missing or unknown code?

Flilght↗
NASA NTRS2006

Non Radiation Hardened Microprocessors in Spaced Based Remote Sensing Systems

The CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations) mission is a comprehensive suite of active and passive sensors including a 20Hz 230mj Nd:YAG lidar, a visible wavelength Earth-looking camera and an imaging infrared radiometer. CALIPSO flies in formation with the Earth Observing System Post-Meridian (EOS PM) train, provides continuous, near-simultaneous measurements and is a planned 3 year mission. CALIPSO was launched into a 98 degree sun synchronous Earth orbit in April of 2006 to study clouds and aerosols and acquires over 5 gigabytes of data every 24 hours. The ground track of one CALIPSO orbit as well as high and low intensity South Atlantic Anomaly outlines is shown. CALIPSO passes through the SAA several times each day. Spaced based remote sensing systems that include multiple instruments and/or instruments such as lidar generate large volumes of data and require robust real-time hardware and software mechanisms and high throughput processors. Due to onboard storage restrictions and telemetry downlink limitations these systems must pre-process and reduce the data before sending it to the ground. This onboard processing and realtime requirement load may mean that newer more powerful processors are needed even though acceptable radiation-hardened versions have not yet been released. CALIPSO's single board computer payload controller processor is actually a set of four (4) voting non-radiation hardened COTS Power PC 603r's built on a single width VME card by General Dynamics Advanced Information Systems (GDAIS). Significant radiation concerns for CALIPSO and other Low Earth Orbit (LEO) satellites include the South Atlantic Anomaly (SAA), the north and south poles and strong solar events. Over much of South America and extending into the South Atlantic Ocean the Van Allen radiation belts dip to just 200-800km and spacecraft entering this area are subjected to high energy protons and experience higher than normal Single Event Upset (SEU) and Single Event Latch-up (SEL) rates. Although less significant, spacecraft flying in the area around the poles experience similar upsets. Finally, powerful solar proton events in the range of 10MeV/10pfu to 100MeV/1pfu as are forecasted and tracked by NOAA's Space Environment Center in Colorado can result in Single Event Upset (SEU), Single Event Latch-up (SEL) and permanent failures such as Single Event Gate Rupture (SEGR) in some technologies. (Galactic Cosmic Rays (GCRs) are another source, especially for gate rupture) CALIPSO mitigates common radiation concerns in its data handling through the use of redundant processors, radiation-hardened Application Specific Integrated Circuits (ASIC), hardware-based Error Detection and Correction (EDAC), processor and memory scrubbing, redundant boot code and mirrored files. After presenting a system overview this paper will expand on each of these strategies. Where applicable, related on-orbit data collected since the CALIPSO initial boot on May 4, 2006 will be noted.

Decoursey, Robert J.↗
NASA NTRS2015

Technology Advancement for Active Remote Sensing of Carbon Dioxide from Space Using the ASCENDS CarbonHawk Experiment Simulator: First Results

The ASCENDS CarbonHawk Experiment Simulator (ACES) is a newly developed lidar developed at NASA Langley Research Center and funded by NASA's Earth Science Technology Office (ESTO) Instrument Incubator Program (IIP) that seeks to advance technologies critical to measuring atmospheric column carbon dioxide (CO2) mixing ratios in support of the NASA Active Sensing of CO2 Emissions over Nights, Days, and Seasons (ASCENDS) mission. The technology advancements targeted include: (1) increasing the power-aperture product to approach ASCENDS mission requirements by implementing multi-aperture telescopes and multiple co-aligned laser transmitters; (2) incorporating high-efficiency, high-power Erbium-Doped Fiber Amplifiers (EDFAs); (3) developing and incorporating a high-bandwidth, low-noise HgCdTe detector and transimpedence amplifier (TIA) subsystem capable of long-duration autonomous operation on Global Hawk aircraft, and (4) advancing algorithms for cloud and aerosol discrimination. The ACES instrument architecture is being developed for operation on high-altitude aircraft and will be directly scalable to meet the ASCENDS mission requirements. These technologies are critical towards developing not only spaceborne instruments but also their airborne simulators, with lower platform requirements for size, mass, and power, and with improved instrument performance for the ASCENDS mission. ACES transmits five laser beams: three from commercial EDFAs operating near 1.57 microns, and two from the Exelis oxygen (O2) Raman fiber laser amplifier system operating near 1.26 microns. The three EDFAs are capable of transmitting up to 10 watts average optical output power each and are seeded by compact, low noise, stable, narrow-linewidth laser sources stabilized with respect to a CO2 absorption line using a multi-pass gas absorption cell. The Integrated-Path Differential Absorption (IPDA) lidar approach is used at both wavelengths to independently measure the CO2 and O2 column number densities and retrieve the average column CO2 mixing ratio. The ACES receiver uses three fiber-coupled 17.8-cm diameter athermal telescopes. The transmitter assembly consists of five fiber-coupled laser collimators and an associated Risley prism pair for each laser to co-align the outgoing laser beams and to align them with the telescope field of view. The backscattered return signals collected by the three telescopes are combined in a fiber bundle and sent to a single low noise detector. The detector/TIA development has improved the existing detector subsystem by increasing its bandwidth to 4.7 MHz from 500 kHz and increasing the duration of autonomous, service-free operation periods from 4 hours to >24 hours. The new detector subsystem enables the utilization of higher laser modulation rates, which provides greater flexibility for implementing advanced thin-cloud discrimination algorithms as well as improving range-determination resolution and error reduction. The cloud/aerosol discrimination algorithm development by Langley and Exelis features a new suite of algorithms for the minimization/elimination of bias errors in the return signal induced by the presence of intervening thin clouds. Multiple laser modulation schemes are being tested in an effort to significantly mitigate the effects of thin clouds on the retrieved CO2 column amounts. Full instrument development concluded in the spring of 2014. After ground range tests of the instrument, ACES successfully completed six test flights on the Langley Hu-25 aircraft in July, 2014, and recorded data at multiple altitudes over land and ocean surfaces with and without intervening clouds. Preliminary results from these test flights will be presented in this paper.

Obland, Michael D.↗
NASA NTRSDate not supplied

Monitoring of Ionic Silver Biocide with Capacitively-Coupled Contactless Conductivity Detection (C4D)

Monitoring of biocide dosing in spacecraft potable water systems is desirable to ensure dosing system operation and resultant microbial control. Electrolytic conductivity is linearly related to the concentration ionic silver (Ag+, the baseline biocide for future NASA missions) chemically dosed into highly purified water, and could serve as a proxy measurement. Traditional conductivity cells rely on immersed metal or graphite electrodes, which may potentially be subject to significant current-induced or auto-galvanic deposition of Ag+. Furthermore, oxide fouling may potentially induce measurement errors. Capacitively-coupled contactless conductivity detection (C4D) is a promising alternative. C4D uses electrodes placed outside an inert, non-conductive flow path, potentially mitigating such problems. We identified and purchased a commercial C4D system with claimed performance (conductivity range: 20 nS/cm to 4 µS/cm) appropriate for this application. We characterized the instrument’s performance with conductivity standards and Ag+ solutions. We began investigations considering practical and long-term detector operation, including potential current-induced deposition of conductive deposits at the tubing wall. In addition, we designed a side-stream flow system to provide liquid flow through the narrow tubing (1/16” outer diameter) required by the detector and characterized the performance of the combined C4D detector/flow system. Finally, we completed theoretical modeling of expected conductivity change with multiple dosing techniques and water chemistries.

Lance Dean Delzeit↗
DOE OSTI2025

Suppressing the sample variance of DESI-like galaxy clustering with fast simulations

Ongoing and upcoming galaxy redshift surveys, such as the Dark Energy Spectroscopic Instrument (DESI) survey, will observe vast regions of sky and a wide range of redshifts. In order to model the observations and address various systematic uncertainties, N-body simulations are routinely adopted, however, the number of large simulations with sufficiently high mass resolution is usually limited by available computing time. Therefore, achieving a simulation volume with the effective statistical errors significantly smaller than those of the observations becomes prohibitively expensive. In this study, we apply the Convergence Acceleration by Regression and Pooling (CARPool) method to mitigate the sample variance of the DESI-like galaxy clustering in the AbacusSummit simulations, with the assistance of the quasi-N-body simulations FastPM. Based on the halo occupation distribution (HOD) models, we construct different FastPM galaxy catalogs, including the luminous red galaxies (LRGs), emission line galaxies (ELGs), and quasars, with their number densities and two-point clustering statistics well matched to those of AbacusSummit. We also employ the same initial conditions between AbacusSummit and FastPM to achieve high cross-correlation, as it is useful in effectively suppressing the variance. Our method of reducing noise in clustering is equivalent to performing a simulation with volume larger by a factor of 5 and 4 for LRGs and ELGs, respectively. We also mitigate the standard deviation of the LRG bispectrum with the triangular configurations k 2 = 2k 1 = 0.2 h Mpc -1 by a factor of 1.6. With smaller sample variance on galaxy clustering, we are able to constrain the baryon acoustic oscillations (BAO) scale parameters to higher precision. The CARPool method will be beneficial to better constrain the theoretical systematics of BAO, redshift space distortions (RSD) and primordial non-Gaussianity (NG).

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS2015

Tactically Extensible and Modular Communications - X-Band TEMCOM-X

This paper will discuss a proposed CubeSat size (3U) telemetry system concept being developed at Marshall Space Flight Center (MSFC) in cooperation with the U.S. Department of the Army and Dynetics Corporation. This telemetry system incorporates efficient, high-bandwidth communications by developing flight-ready, low-cost, Protoflight software defined radio (SDR) and Electronically Steerable Patch Array (ESPA) antenna subsystems for use on platforms as small as CubeSats and unmanned aircraft systems (UASs). The current telemetry system is slightly larger in dimension of footprint than required to fit within a 0.5U CubeSat volume. Extensible and modular communications for CubeSat technologies will partially mitigate current capability gaps between traditional strategic space platforms and lower-cost small satellite solutions. Higher bandwidth capacity will enable high-volume, low error-rate data transfer to and from tactical forces or sensors operating in austere locations (e.g., direct imagery download, unattended ground sensor data exfiltration, interlink communications), while also providing additional bandwidth and error correction margin to accommodate more complex encryption algorithms and higher user volume.

Sims, William Herbert↗
NASA NTRS2021

Unsteady Pressure-Sensitive-Paint Shot Noise Reduction

In the study of unsteady Pressure-Sensitive Paint (uPSP) image data sets, pixel intensity values collected by Phantom cameras from multiple perspectives are used to calculate the surface pressure of an experimental model. This paper focuses on mitigation of shot noise which is introduced into the image data set by the cameras. Shot noise impacts the quality of data collected, increasing the system error in its pressure value calculations. This paper characterises statistical methods to reduce shot noise within experimental data by taking an average or weighted average across an area of pixels of their respective counts. Each combined-pixel has a higher total effective intensity value and therefore exhibits reduced noise throughout its time history. This is demonstrated by calculating the standard deviations of the combined pixel values through time, which decrease as the combined pixel area increases. Finally, these methods are compared across experimental wind tunnel conditions to examine their effectiveness at different Mach numbers. Ultimately, the results are used to recommend a blur filter type and size which will be applied to the raw image data prior to processing, as well as a minimum camera resolution for future experiments. Downstream of the filter application, the software to convert the 2D image data sets into surface pressure readings on a 3D grid use a ratio of decimation that assigns multiple pixels to each 3D grid node. This decimation ratio will also be aligned with the size of the blur filter, resulting in a more manageable data set size and compromised spatial resolution. Combined with the effect of the blur filter, the resulting pixel intensity readings will exhibit significantly reduced shot noise, reducing the magnitude of uncertainty and error in the final calculated pressure. For future tests, the camera spatial resolution must be sufficient to capture small flow and model features even after the recommended blur filter is applied. The image data set is taken from a uPSP experiment of the Space Launch System Block 1 cargo vehicle model in September 2019.

Lucy Tang↗
NASA NTRS2021

Unsteady Pressure-Sensitive Paint (UPSP) Shot Noise Reduction

In the study of unsteady Pressure-Sensitive Paint (uPSP) image data sets, pixel intensity values collected by Phantom cameras from multiple perspectives are used to calculate the surface pressure of an experimental model. This paper focuses on mitigation of shot noise which is introduced into the image data set by the cameras. Shot noise impacts the quality of data collected, increasing the system error in its pressure value calculations. This paper characterises statistical methods to reduce shot noise within experimental data by taking an average or weighted average across an area of pixels of their respective counts. Each combined-pixel has a higher total effective intensity value and therefore exhibits reduced noise throughout its time history. This is demonstrated by calculating the standard deviations of the combined pixel values through time, which decrease as the combined pixel area increases. Finally, these methods are compared across experimental wind tunnel conditions to examine their effectiveness at different Mach numbers. Ultimately, the results are used to recommend a blur filter type and size which will be applied to the raw image data prior to processing, as well as a minimum camera resolution for future experiments. Downstream of the filter application, the software to convert the 2D image data sets into surface pressure readings on a 3D grid use a ratio of decimation that assigns multiple pixels to each 3D grid node. This decimation ratio will also be aligned with the size of the blur filter, resulting in a more manageable data set size and compromised spatial resolution. Combined with the effect of the blur filter, the resulting pixel intensity readings will exhibit significantly reduced shot noise, reducing the magnitude of uncertainty and error in the final calculated pressure. For future tests, the camera spatial resolution must be sufficient to capture small flow and model features even after the recommended blur filter is applied. The image data set is taken from a uPSP experiment of the Space Launch System Block 1 cargo vehicle model in September 2019.

Lucy Tang↗
NASA NTRS2022

Polarization performance simulation for the GeoXO atmospheric composition instrument: NO2 retrieval impacts

NOAA’s Geostationary Extended Observations (GeoXO) constellation will continue and expand on the capabilities of the current generation of geostationary satellite systems to support US weather, ocean, atmosphere, and climate operations. It is planned to consist of a dedicated atmospheric composition instrument (ACX) to support air quality forecasting and monitoring by providing similar capabilities to missions such as TEMPO (Tropospheric Emission: Monitoring Pollution), currently planned to launch in 2023, and Ozone Monitoring Instrument (OMI), TROPOMI (TROPOspheric Monitoring Instrument), and GEMS (Geostationary Environment Monitoring Spectrometer) currently in operation. As the early phases of ACX development are progressing, design trade-offs are being considered to understand the relationship between instrument design choices and trace gas retrieval impacts. Some of these choices will affect the instrument polarization sensitivity (PS), which can have radiometric impacts on environmental satellite observations. We conducted a study to investigate how such radiometric impacts can affect NO2 retrievals by exploring their sensitivities to time of day, location, and scene type with an ACX instrument model that incorporates PS. The study addresses the basic steps of operational NO2 retrievals: the spectral fitting step and the conversion of slant column to vertical column via the air mass factor (AMF). The spectral fitting step was performed by generating at-sensor radiance from a clear sky scene with a known NO2 amount, the application of an instrument model including both instrument PS and noise, and a physical retrieval. The spectral fitting step was found to mitigate the impacts of instrument PS. The AMF-related step was considered for clear sky and partially cloudy scenes, where instrument PS can lead to errors in interpreting the cloud content, propagating to AMF errors and finally to NO2 retrieval errors. For this step, the NO2 retrieval impacts were small but non-negligible for high NO2 amounts; we estimated that a typical high NO2 amount can cause maximum of 0.25 x 10^15 molecules/cm^2 for a PS of 5%. These simulation capabilities were designed to aid in the development of a GeoXO atmospheric composition instrument that will improve our ability to monitor and understand the Earth’s atmosphere.

trace gases↗
DOE OSTI2025

Stable Machine‐Learning Parameterization of Subgrid Processes in a Comprehensive Atmospheric Model Learned From Embedded Convection‐Permitting Simulations

Modern climate projections often suffer from inadequate spatial and temporal resolution due to computational limitations, resulting in inaccurate representations of sub-grid processes. A promising technique to address this is the multiscale modeling framework (MMF), which embeds a kilometer-resolution cloud-resolving model (CRM) within each atmospheric column of a host climate model to replace traditional convection and cloud parameterizations. Machine learning offers a unique opportunity to make MMF more accessible by emulating the embedded CRM and reducing its substantial computational cost. Although many studies have demonstrated proof-of-concept success of achieving stable hybrid simulations, it remains a challenge to achieve near operational-level success with real geography and comprehensive variable emulation that includes, for example, explicit cloud condensate coupling. In this study, we present a stable hybrid model capable of integrating for at least 5 years with near operational-level complexity, including coarse-grid geography, seasonality, explicit cloud condensate and wind predictions, and land coupling. Our model demonstrates skillful online performance, achieving a 5-year zonal mean tropospheric temperature bias within 2 K, water vapor bias within 1 g/kg, and a precipitation root mean square error of 0.96 mm/day. Key factors contributing to our online performance include an expressive U-Net architecture and physical thermodynamic constraints for microphysics. With microphysical constraints mitigating unrealistic cloud formation, our work is the first to demonstrate realistic multi-year cloud condensate climatology under the MMF framework. Despite these advances, online diagnostics reveal persistent biases in certain regions, highlighting the need for innovative strategies to further optimize online performance.

Hu, Zeyuan [NVIDIA Corporation, Santa Clara, CA (U↗
DOE OSTI2024

A robust approach to Gaussian process implementation

Abstract. Gaussian process (GP) regression is a flexible modeling technique used to predict outputs and to capture uncertainty in the predictions. However, the GP regression process becomes computationally intensive when the training spatial dataset has a large number of observations. To address this challenge, we introduce a scalable GP algorithm, termed MuyGPs, which incorporates nearest-neighbor and leave-one-out cross-validation during training. This approach enables the evaluation of large spatial datasets with state-of-the-art accuracy and speed in certain spatial problems. Despite these advantages, conventional quadratic loss functions used in the MuyGPs optimization, such as root mean squared error (RMSE), are highly influenced by outliers. We explore the behavior of MuyGPs in cases involving outlying observations and, subsequently, develop a robust approach to handle and mitigate their impact. Specifically, we introduce a novel leave-one-out loss function based on the pseudo-Huber function (LOOPH) that effectively accounts for outliers in large spatial datasets within the MuyGPs framework. Our simulation study shows that the LOOPH loss method maintains accuracy despite outlying observations, establishing MuyGPs as a powerful tool for mitigating unusual observation impacts in the large data regime. In the analysis of US ozone data, MuyGPs provides accurate predictions and uncertainty quantification, demonstrating its utility in managing data anomalies. Through these efforts, we advance the understanding of GP regression in spatial contexts.

Mukangango, Juliette↗