Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “two-part”

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 37 records · Page 2

2022 Central California Travel Study

# 2022 Central California Travel Study The 2022 Central California Travel Study collected demographic and travel pattern information to better understand household travel behavior in the state’s San Joaquin Valley region. ## Data Collection Agency The Fresno Council of Governments conducted the household travel study in collaboration with metropolitan planning organizations in Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties in California. ## Survey Methodology The two-part study consisted of a survey and a travel diary. The survey gathered data on household demographic composition and typical travel behaviors. The travel diary gathered individual travel data during a specified travel period for all members of a given household. ## Travel Diary Methods Households with smartphones were encouraged to complete their travel diaries using the rMove smartphone app for up to seven consecutive days. Households without smartphones or those unwilling to participate via smartphone completed their travel diaries online (using the web-version of rMove) or by calling the survey call center. These households reported travel for one day (Tuesday, Wednesday, or Thursday). ## Recruitment Sampling Methods The majority of the recruitment was conducted via address-based sampling, a type of probability sampling, with a focus on reaching county-level targets in collaboration with metropolitan planning organizations in the region. Supplemental sampling methods, primarily non-probability, were employed during all waves of data collection to improve survey representation. The supplemental sample included targeted outreach to hard-to-survey populations via transit rider email lists; local housing authorities; Nichols Research, a California-based market research firm; and the Ipsos™ KnowledgePanel, an online random probability panel. See the documentation for more information about sampling and weighting. ## Survey Records, Data, and Documentation In total, 7,406 households completed 19,084 surveys representing 150,012 trips across 42,567 person-days.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

2022 Central California Travel Study

# 2022 Central California Travel Study The 2022 Central California Travel Study collected demographic and travel pattern information to better understand household travel behavior in the state’s San Joaquin Valley region. ## Data Collection Agency The Fresno Council of Governments conducted the household travel study in collaboration with metropolitan planning organizations in Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties in California. ## Survey Methodology The two-part study consisted of a survey and a travel diary. The survey gathered data on household demographic composition and typical travel behaviors. The travel diary gathered individual travel data during a specified travel period for all members of a given household. ## Travel Diary Methods Households with smartphones were encouraged to complete their travel diaries using the rMove smartphone app for up to seven consecutive days. Households without smartphones or those unwilling to participate via smartphone completed their travel diaries online (using the web-version of rMove) or by calling the survey call center. These households reported travel for one day (Tuesday, Wednesday, or Thursday). ## Recruitment Sampling Methods The majority of the recruitment was conducted via address-based sampling, a type of probability sampling, with a focus on reaching county-level targets in collaboration with metropolitan planning organizations in the region. Supplemental sampling methods, primarily non-probability, were employed during all waves of data collection to improve survey representation. The supplemental sample included targeted outreach to hard-to-survey populations via transit rider email lists; local housing authorities; Nichols Research, a California-based market research firm; and the Ipsos™ KnowledgePanel, an online random probability panel. See the documentation for more information about sampling and weighting. ## Survey Records, Data, and Documentation In total, 7,406 households completed 19,084 surveys representing 150,012 trips across 42,567 person-days.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

2022 Central California Travel Study

# 2022 Central California Travel Study The 2022 Central California Travel Study collected demographic and travel pattern information to better understand household travel behavior in the state’s San Joaquin Valley region. ## Data Collection Agency The Fresno Council of Governments conducted the household travel study in collaboration with metropolitan planning organizations in Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties in California. ## Survey Methodology The two-part study consisted of a survey and a travel diary. The survey gathered data on household demographic composition and typical travel behaviors. The travel diary gathered individual travel data during a specified travel period for all members of a given household. ## Travel Diary Methods Households with smartphones were encouraged to complete their travel diaries using the rMove smartphone app for up to seven consecutive days. Households without smartphones or those unwilling to participate via smartphone completed their travel diaries online (using the web-version of rMove) or by calling the survey call center. These households reported travel for one day (Tuesday, Wednesday, or Thursday). ## Recruitment Sampling Methods The majority of the recruitment was conducted via address-based sampling, a type of probability sampling, with a focus on reaching county-level targets in collaboration with metropolitan planning organizations in the region. Supplemental sampling methods, primarily non-probability, were employed during all waves of data collection to improve survey representation. The supplemental sample included targeted outreach to hard-to-survey populations via transit rider email lists; local housing authorities; Nichols Research, a California-based market research firm; and the Ipsos™ KnowledgePanel, an online random probability panel. See the documentation for more information about sampling and weighting. ## Survey Records, Data, and Documentation In total, 7,406 households completed 19,084 surveys representing 150,012 trips across 42,567 person-days.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

2022 Central California Travel Study

# 2022 Central California Travel Study The 2022 Central California Travel Study collected demographic and travel pattern information to better understand household travel behavior in the state’s San Joaquin Valley region. ## Data Collection Agency The Fresno Council of Governments conducted the household travel study in collaboration with metropolitan planning organizations in Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties in California. ## Survey Methodology The two-part study consisted of a survey and a travel diary. The survey gathered data on household demographic composition and typical travel behaviors. The travel diary gathered individual travel data during a specified travel period for all members of a given household. ## Travel Diary Methods Households with smartphones were encouraged to complete their travel diaries using the rMove smartphone app for up to seven consecutive days. Households without smartphones or those unwilling to participate via smartphone completed their travel diaries online (using the web-version of rMove) or by calling the survey call center. These households reported travel for one day (Tuesday, Wednesday, or Thursday). ## Recruitment Sampling Methods The majority of the recruitment was conducted via address-based sampling, a type of probability sampling, with a focus on reaching county-level targets in collaboration with metropolitan planning organizations in the region. Supplemental sampling methods, primarily non-probability, were employed during all waves of data collection to improve survey representation. The supplemental sample included targeted outreach to hard-to-survey populations via transit rider email lists; local housing authorities; Nichols Research, a California-based market research firm; and the Ipsos™ KnowledgePanel, an online random probability panel. See the documentation for more information about sampling and weighting. ## Survey Records, Data, and Documentation In total, 7,406 households completed 19,084 surveys representing 150,012 trips across 42,567 person-days.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

2022 Central California Travel Study

# 2022 Central California Travel Study The 2022 Central California Travel Study collected demographic and travel pattern information to better understand household travel behavior in the state’s San Joaquin Valley region. ## Data Collection Agency The Fresno Council of Governments conducted the household travel study in collaboration with metropolitan planning organizations in Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties in California. ## Survey Methodology The two-part study consisted of a survey and a travel diary. The survey gathered data on household demographic composition and typical travel behaviors. The travel diary gathered individual travel data during a specified travel period for all members of a given household. ## Travel Diary Methods Households with smartphones were encouraged to complete their travel diaries using the rMove smartphone app for up to seven consecutive days. Households without smartphones or those unwilling to participate via smartphone completed their travel diaries online (using the web-version of rMove) or by calling the survey call center. These households reported travel for one day (Tuesday, Wednesday, or Thursday). ## Recruitment Sampling Methods The majority of the recruitment was conducted via address-based sampling, a type of probability sampling, with a focus on reaching county-level targets in collaboration with metropolitan planning organizations in the region. Supplemental sampling methods, primarily non-probability, were employed during all waves of data collection to improve survey representation. The supplemental sample included targeted outreach to hard-to-survey populations via transit rider email lists; local housing authorities; Nichols Research, a California-based market research firm; and the Ipsos™ KnowledgePanel, an online random probability panel. See the documentation for more information about sampling and weighting. ## Survey Records, Data, and Documentation In total, 7,406 households completed 19,084 surveys representing 150,012 trips across 42,567 person-days.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

2022 Central California Travel Study

# 2022 Central California Travel Study The 2022 Central California Travel Study collected demographic and travel pattern information to better understand household travel behavior in the state’s San Joaquin Valley region. ## Data Collection Agency The Fresno Council of Governments conducted the household travel study in collaboration with metropolitan planning organizations in Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties in California. ## Survey Methodology The two-part study consisted of a survey and a travel diary. The survey gathered data on household demographic composition and typical travel behaviors. The travel diary gathered individual travel data during a specified travel period for all members of a given household. ## Travel Diary Methods Households with smartphones were encouraged to complete their travel diaries using the rMove smartphone app for up to seven consecutive days. Households without smartphones or those unwilling to participate via smartphone completed their travel diaries online (using the web-version of rMove) or by calling the survey call center. These households reported travel for one day (Tuesday, Wednesday, or Thursday). ## Recruitment Sampling Methods The majority of the recruitment was conducted via address-based sampling, a type of probability sampling, with a focus on reaching county-level targets in collaboration with metropolitan planning organizations in the region. Supplemental sampling methods, primarily non-probability, were employed during all waves of data collection to improve survey representation. The supplemental sample included targeted outreach to hard-to-survey populations via transit rider email lists; local housing authorities; Nichols Research, a California-based market research firm; and the Ipsos™ KnowledgePanel, an online random probability panel. See the documentation for more information about sampling and weighting. ## Survey Records, Data, and Documentation In total, 7,406 households completed 19,084 surveys representing 150,012 trips across 42,567 person-days.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

2022 Central California Travel Study

# 2022 Central California Travel Study The 2022 Central California Travel Study collected demographic and travel pattern information to better understand household travel behavior in the state’s San Joaquin Valley region. ## Data Collection Agency The Fresno Council of Governments conducted the household travel study in collaboration with metropolitan planning organizations in Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties in California. ## Survey Methodology The two-part study consisted of a survey and a travel diary. The survey gathered data on household demographic composition and typical travel behaviors. The travel diary gathered individual travel data during a specified travel period for all members of a given household. ## Travel Diary Methods Households with smartphones were encouraged to complete their travel diaries using the rMove smartphone app for up to seven consecutive days. Households without smartphones or those unwilling to participate via smartphone completed their travel diaries online (using the web-version of rMove) or by calling the survey call center. These households reported travel for one day (Tuesday, Wednesday, or Thursday). ## Recruitment Sampling Methods The majority of the recruitment was conducted via address-based sampling, a type of probability sampling, with a focus on reaching county-level targets in collaboration with metropolitan planning organizations in the region. Supplemental sampling methods, primarily non-probability, were employed during all waves of data collection to improve survey representation. The supplemental sample included targeted outreach to hard-to-survey populations via transit rider email lists; local housing authorities; Nichols Research, a California-based market research firm; and the Ipsos™ KnowledgePanel, an online random probability panel. See the documentation for more information about sampling and weighting. ## Survey Records, Data, and Documentation In total, 7,406 households completed 19,084 surveys representing 150,012 trips across 42,567 person-days.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Design of a shipping fixture for a compact cryomodule hermetic assembly

In support of the development of a conduction-cooled 915MHz superconducting radio frequency (SRF) cryomodule, this study highlights the design of a shipping fixture for transporting the hermetic assembly 4500 km from Jefferson Lab to General Atomics in San Diego, California. The hermetic assembly consists of a 2-cell 915 MHz SRF cavity, a coaxial fundamental power coupler and warm-to-cold transition beam tubes. The two-part shipping assembly consists of an inner frame, providing direct mounting of the components, and an outer frame mounted to the ground transport vehicle. The inner frame is then connected to the outer frame by way of wire-rope isolators. Accelerometer data from ground transportation of previous projects at Jefferson Lab provides the baseline for the expected frequency and magnitude of vibrational and shock events during transit. Modal analyses were carried out in ANSYS on the inner frame assembly and critical components to identify an appropriate wire-rope isolator configuration such that peak loads are mitigated and the incurred frequencies do not correspond with the fundamental modes of the structures.

Accelerator Physics↗

Development of Radiation and Fire Resistant Polyurethane Foam in Support of Deactivation and Decommissioning [Poster]

FoamBag™ is a two-part polyurethane foam that has been down selected as a possible engineering solution for the decommissioning of radiologically contaminated pipes. Our main objective consists of establishing an internal barrier (plug) to mitigate the release of residual contamination prior to cutting operations and required pipework for deactivation and decommissioning projects. To ensure worker safety, we will also be analyzing hazards associated with the off gasses present during curing.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Proton Improvement Plan II (PIP-II) Clock and Timing

The PIP-II timing system is planned to be a two-part system consisting of a global timing system (referred to as ACLK) that provides high level, event-based timing for the whole Fermilab accelerator complex while the second part is a RF synchronized clock system unique to the PIP-II Linac itself (referred to as LCLK). The ACLK System will make use of an external 10 MHz GPS based signal source as a reference for its 650 MHz phase lock as it is the reference frequency for the TCLK output of the system (needed to support legacy hardware around complex). The LCLK System will use a PIP-II Linac RF reference (162.5 MHz) from the Linac LLRF system to allow beam synchronized event placement. Both the ACLK and LCLK systems will have a clock output with a data frame of 16 event bits + 32 data bits with frames broadcast at 650 MHz, phase locked to the 10MHz reference in the case of ACLK and the PIP-II LLRF sourced 162.5 MHz reference in the case of LCLK.

43 PARTICLE ACCELERATORS↗

Puck Loop Seal (Final Project Technical Report)

Sandia National Laboratories (SNL) is advancing technical capabilities used in passive loop seals. The “Puck” seal used a set of International Atomic Energy Agency (IAEA) requirements for new passive loop seals published in 2020 as a design guide. The seal is based on an oxygen-sensitive inner mixture encased in an oxygen-impermeable shell, is monolithic rather than two-part, incorporates self-capturing wire features, contains colored water beads and bubbles formed during processing as unique identifiers (UIDs), and visually indicates tamper (whether from seal body penetration or from seal wire removal) by irreversibly changing the seal body from multi-colored to black. This paper will provide details on the design, development, and testing of Puck seal prototypes.

42 ENGINEERING↗

A Regional Approach to Offshore Wind Energy Manufacturing in the Central Atlantic: Workforce

Meeting the additional job requirements spurred by potential growth in the offshore wind industry is likely to involve intentional and effective recruitment practices and training program development across all occupations (e.g., skilled trades, engineers, professionals) at state, regional, and local levels. The workforce development ecosystem is complex, especially for an industry like offshore wind, so assessing the readiness of an area to support the workforce needs of new manufacturing facilities involves evaluating numerous economic and training factors at an occupational level. Funded by the National Offshore Wind Research and Development Consortium and led by the National Renewable Energy Laboratory, this report represents one part of a two-part study that explores the challenges and opportunities for Delaware, Maryland, Virginia, and North Carolina in fostering regional collaboration to build a domestic supply chain for the U.S. offshore wind energy sector.

17 WIND ENERGY↗

Leveraging ARM Data to Improve Models for Predictive Understanding of Energy and Security Challenges

Extreme weather and natural hazards can disrupt the energy sector, affecting demand, generation, transmission, distribution, consumption and operational planning at regional and national scales. These disruptions stem from a broad range of atmospheric phenomena, including winter storms, freezing rain, wet snow loading, severe convection, flooding and landslides, wildfires, prolonged heat, and drought. Many of these same phenomena can also affect national security through impacts to transportation and infrastructure. To support the U.S. Department of Energy (DOE) focus on energy resilience and national security, the Atmospheric Radiation Measurement (ARM) User Facility is uniquely positioned to contribute measurement data, analyses, and modeling frameworks that can significantly improve predictive understanding of these hazards to mitigate their effects. To explore this opportunity, ARM convened a two-part virtual workshop in November 2025. The workshop engaged interdisciplinary experts in atmospheric science, energy systems, modeling, and operations. The goal of the meeting was to engage with these interdisciplinary experts to address three questions: • What are examples of atmospheric processes that represent significant risks to energy security or national security and where are those risks greatest? • What measurements or measurement strategies would improve ARM’s capacity to address these issues? • How can ARM and users of the ARM facility better work with the Energy Exascale Earth System Model (E3SM) and multi-sector modeling communities to apply ARM data to improving E3SM simulations of these phenomena? Participants were asked to submit white papers ahead of the meeting to initiate thinking on these themes and to help organize discussions. Workshop sessions were then organized around themes identified in the white papers. First from the white papers and then through subsequent discussions, workshop participants identified many examples that address the three questions listed above. Participants called out energy system vulnerabilities to weather phenomena such as the impact of freezing rain, strong winds, and excessive heat on power grids. They also noted the effects that weather phenomena could have on energy demand or supply (e.g., through effects of extreme temperatures). They called out security vulnerabilities such as impacts to crops from aerosol-borne pathogens and risks to industry due to melting permafrost in the Arctic. In all, over a dozen meteorological phenomena were linked to energy or security vulnerabilities. For many of the identified phenomena, participants pointed out where ARM was well poised to address issues (e.g., through measurements of cloud microphysics to inform studies of freezing rain) but also noted needs for additional measurements or modified measurement strategies. For example, adaptive scanning of severe weather would be valuable for probing winter storms or severe convection. Participants pointed out the value in integrating external observations with ARM measurements and with applying artificial intelligence (AI) to ARM observation analysis and they advocated for using model simulations to help optimize measurement strategies through Observing System Simulation Experiments (OSSEs). It was clear from the workshop that there are many ways that ARM observations can be used to mitigate energy and security concerns, but meeting participants were also asked to identify what they considered to be the greatest opportunities by ranking issues pertaining to the three workshop questions. This was accomplished through a survey administered to participants between the two virtual sessions. The highest-priority phenomena identified were winter storms, severe convection, and arctic processes. Discussion in the second session, therefore, focused primarily on these three areas, which were most fully developed in exploring ARM opportunities. Nevertheless, it was also clear that ARM has opportunities to contribute to all the identified topics. This report describes the workshop, including input from discussion and white papers (Sections 2 and 3) and a list of priority recommendations (section 4). Many other ideas for ARM contributions are discussed in individual white papers (Appendix D).

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Lahaina Energy Partnership: Technical Assistance Task Updates and Discussion Part 2

The Lahaina Energy Partnership (LEP) is an initiative funded by the U.S. Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy (EERE) to support energy planning and rebuilding efforts in Hawaii's historic town of Lahaina on Maui as the community recovers from a devastating fire on August 8, 2023, with technical assistance provided by the National Laboratory of the Rockies. NLR has partnered with Hawaii-based community organizations to engage with Lahaina the community on energy priorities and inform the technical assistance scope. This virtual workshop presentation is the second in a two-part series to provide progress updates and request community input to guide next steps. Workshop 1 on November 18 focused on hydropower resource potential, building energy modeling, and workforce development. Workshop 2 on December 11 (this presentation) will focus on microgrids, electric grid hardening, policy and regulation.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Lahaina Energy Partnership: Technical Assistance Task Updates and Discussion Part 1 [Slides]

The Lahaina Energy Partnership (LEP) is an initiative funded by the U.S. Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy (EERE) to support energy planning and rebuilding efforts in Hawaii's historic town of Lahaina on Maui as the community recovers from a devastating fire on August 8, 2023, with technical assistance provided by the National Laboratory of the Rockies. NLR has partnered with Hawaii-based community organizations to engage with Lahaina the community on energy priorities and inform the technical assistance scope. This virtual workshop presentation is the first in a two-part series to provide progress updates and request community input to guide next steps. Workshop 1 on November 18 (this presentation) will focus on Hydropower resource potential, building energy modeling, workforce development. Workshop 2 on December 11 (presentation forthcoming) will focus on microgrids, electric grid hardening, policy and regulation.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Urban Impacts on Convective Squall Lines over Chicago in the Warm Season—Part I: Observations of Multi-Scale Convective Evolution

In this study, our aim is to diagnose how two quasi-linear convective systems (QLCS) are organized so one can determine the possible role of the city of Chicago, IL, USA, in modifying convective precipitation systems. In this Part I of a two-part study, we employ large-scale analyses, radiosonde soundings, surface observations, and Doppler radar data to diagnose the precursor atmospheric circulations that organize the evolution of two mesoscale convective systems and compare those circulations to radar and precipitation. Several multi-scale processes are found that organize and modify convection over the Chicago metroplex. Two sequential quasi-linear convective systems (QLCS #1 and #2) were organized that propagated over Chicago, IL, USA, during an eight-hour period on 5–6 July 2018. The first squall line (QLCS #1) built from the southwest to the northeast while strengthening as it propagated over the city, and the second (QLCS #2) propagated southeastwards and weakened as it passed over the city in association with a polar cold front. The weak upper-level divergence associated with a diffluent flow poleward of an expansive ridge built over and strengthened a low-level trough and confluence zone, triggering QLCS #1. Convective downdrafts from QLCS #1 produced a cold pool that interacted with multiple confluent low-level jets surrounding and focused on the metroplex urban heat island, thus advecting the convection poleward over the metroplex. The heaviest precipitation occurred just south-southeast of Midway Airport, Chicago. Subsequently, a polar cold front propagated into the metroplex, which triggered QLCS #2. However, the descending air above it under the polar jet and residual cold pool from QLCS #1 rapidly dissipated the cold frontal convection. This represents a case study where very active convection built over the metroplex and was likely modified by it, as evidenced in numerical simulations to be described in Part II.

Kaplan, Michael L.↗

Impact of Irradiation on Corrosion Performance of Hybrid Organic/Inorganic Coatings on Austenitic Stainless Steel

The effects of gamma radiation on the performance of two corrosion-resistant coatings applied to stainless-steel 304L (SS304L) surfaces are presented. Specifically, the ability of the coatings to mitigate corrosion of SS304L surfaces as a function of the dose received (0–1300 Mrad) and dose rate (176 compared to 1054 rad/s) is evaluated using electrochemical methods, spectroscopy, and microscopy. Coating A, an organic/inorganic hybrid coating consisting of a two-part silica ceramic component and a polymer linker was evaluated in comparison to Coating B, which utilized Coating A as a topcoat for a commercial, off-the-shelf, Zn-rich primer. Post irradiation, Coating A demonstrated some corrosion protection following exposure to low levels of gamma radiation, but coating degradation occurred with an increased exposure dose and resulted in isolated regions of corrosion initiation. For Coating B, greater corrosion resistance was observed compared to Coating A due to the sacrificial nature of the Zn at elevated doses of gamma radiation. No effect of the dose rate (for the single dose examined) was observed for either coating. It is proposed for Coating B that as the polymer coating thermally degrades above 250 °C (bond scission of the polymer occurs), the remaining Zinc layer adhered to the SS304L post-irradiation enables enhanced corrosion resistance as compared to Coating A, which displays solely polymer degradation. The results presented herein establish an understanding of coating behavior with radiation exposure, specifically the relationship between corrosion coating performance and radiation dose, and can inform ageing and lifetime management for various applications.

degradation↗

Huge ensembles – Part 1: Design of ensemble weather forecasts using spherical Fourier neural operators

Abstract. Simulating low-likelihood high-impact extreme weather events in a warming world is a significant and challenging task for current ensemble forecasting systems. While these systems presently use up to 100 members, larger ensembles could enrich the sampling of internal variability. They may capture the long tails associated with climate hazards better than traditional ensemble sizes. Due to computational constraints, it is infeasible to generate huge ensembles (comprised of 1000–10 000 members) with traditional, physics-based numerical models. In this two-part paper, we replace traditional numerical simulations with machine learning (ML) to generate hindcasts of huge ensembles. In Part 1, we construct an ensemble weather forecasting system based on spherical Fourier neural operators (SFNOs), and we discuss important design decisions for constructing such an ensemble. The ensemble represents model uncertainty through perturbed-parameter techniques, and it represents initial condition uncertainty through bred vectors, which sample the fastest-growing modes of the forecast. Using the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System (IFS) as a baseline, we develop an evaluation pipeline composed of mean, spectral, and extreme diagnostics. With large-scale, distributed SFNOs with 1.1 billion learned parameters, we achieve calibrated probabilistic forecasts. As the trajectories of the individual members diverge, the ML ensemble mean spectra degrade with lead time, consistent with physical expectations. However, the individual ensemble members' spectra stay constant with lead time. Therefore, these members simulate realistic weather states during the rollout, and the ML ensemble passes a crucial spectral test in the literature. The IFS and ML ensembles have similar extreme forecast indices, and we show that the ML extreme weather forecasts are reliable and discriminating. These diagnostics ensure that the ensemble can reliably simulate the time evolution of the atmosphere, including low-likelihood high-impact extremes. In Part 2, we generate a huge ensemble initialized each day in summer 2023, and we characterize the simulations of extremes.

Mahesh, Ankur↗