C
Poster Session at the April 11–14, 2022, Workshop
The workshop included a poster session to provide the community with an opportunity to discuss topics relevant to the workshop (research needs, capabilities, data infrastructure/analytics, and machine learning). Posters were introduced in a pre-recorded, virtual “Lightning Round” (3 minutes per poster provided by the presenter). Links to the posters and to the video presentations were available to attendees throughout the workshop and are also included online.1 Thirteen posters were submitted. The title, author, links, and summary information about each of the posters are provided below, grouped into four, sometimes overlapping, categories.
- Solar and solar wind (3)
- Magnetosphere, ionosphere, and thermosphere (3)
- Data science, analytics, and ensemble modeling (3)
- Machine learning (4)
SOLAR AND SOLAR WIND
Moon to Mars (M2M) Space Weather Analysis Office
Yaireska (Yari) Collado-Vega, NASA GSFC
https://vimeo.com/showcase/9407816/video/699034448
The presentation provided a description of the new NASA Moon to Mars (M2M) Space Weather Analysis Office mission and goals. The Office was established to test novel capabilities in collaboration with CCMC and the Space Radiation Analysis Group at Johnson Space Center, serving as the proving ground of real-time analysis to characterize the radiation environment (including both lunar and Mars missions). M2M also supports NASA robotic missions and, by serving as a proving ground, its capabilities
___________________
1 Links to the posters and videos can be found at National Academies of Sciences, Engineering, and Medicine, “Space Weather Operations and Research Infrastructure: Proceedings of the Phase II - (Workshop),” https://www.nationalacademies.org/event/04-11-2022/space-weather-operations-and-research-infrastructure-workshop-phase-ii-workshop.
can be used for commercial purposes if transitioned to operational agencies. M2M analysts also perform real-time validation after solar events and characterize the limitations of the models used in forecasting.
Space Weather at Mars: Research (and Forecasting) Needs
Christina Lee, Space Sciences Laboratory, UC Berkeley
https://vimeo.com/showcase/9407816/video/697514529
Space weather research and forecast needs at Mars were discussed. MARSIS (Mars Advanced Radar for Subsurface and Ionosphere Sounding), for example, experienced a 10-day blackout from SEPs in September 2017. There is a history of space weather observations, where MAVEN (Mars Atmosphere and Volatile Evolution mission) has the most complete coverage, but data gaps occur frequently and sometimes for long durations. Continuous upstream measurements, like an ACE at Mars L1, are needed. Continuation of new instrument payloads on future missions would be valuable. Currently, forecasting at Mars requires real-time beacons from Earth. This is a problem during conjunction, so a dedicated space weather monitor will be needed for Mars inhabitants.
New Developments in Space Weather Modeling
Gabor Toth, University of Michigan
https://vimeo.com/showcase/9407816/video/697513568
The poster described an open-source model of background solar wind with uncertainty quantification. Work continues on CME models to produce the best results in terms of arrival time and other characteristics. Current activity includes incorporating data assimilation for geospace modeling to improve CME prediction. Solar wind monitors would improve solar wind and CME forecasts, including L5 in situ observations and multi-point real-time white light images.
MAGNETOSPHERE, IONOSPHERE, AND THERMOSPHERE
3-D Regional Ionosphere Imaging and SED Reconstruction with a New TEC-Based Ionospheric Data Assimilation System (TIDAS)
Ercha Aa, MIT Haystack Observatory
https://app.smartsheet.com/b/publish?EQBCT=32396a6d7b4d4d85b61d72979abfc257
A new total electron content-based ionospheric data assimilation system (TIDAS) over the Continental U.S. is developed using a hybrid Ensemble-Variational scheme. This data assimilation system can provide accurate and reliable three-dimensional time-evolving electron density maps with high spatial-temporal resolution (1° × 1° × 20 km × 5 min). This high-fidelity regional data assimilation system is a powerful space weather nowcasting tool to reconstruct localized storm-time ionospheric morphology with unprecedented and fine-scale details. Results can help advance current understanding of the fine structures and underlying mechanisms of the midlatitude ionospheric density gradients.
Exploring Coverage of Accelerometer Satellites in the Thermosphere: An Observing System Simulation Experiment
Eric Sutton, CU-Boulder/SWx TREC
https://vimeo.com/showcase/9407816/video/697513206
The poster described an Observing System Simulation Experiment assessing optimal satellite distribution in the thermosphere (8 satellites) for neutral density measurements. It was used to investigate the impact of satellite coverage. While it focused on a specific instrument, it can be expanded to other areas of exploration.
Measuring Upper Atmospheric Winds with the Zephyr Meteor Radar Network and a Look to the Future
Ryan Volz, MIT Haystack Observatory
https://app.smartsheet.com/b/publish?EQBCT=32396a6d7b4d4d85b61d72979abfc257
This project addresses a significant gap in measuring mesospheric and lower thermospheric neutral winds. The presentation describes the NSF DASI (Distributed Array of Small Instruments) project to develop and build a meteor radar network near Boulder, CO, for estimating upper atmosphere neutral winds. A signal reflected from ionized meteor trails is used to estimate zonal and meridional wind with mesoscale horizontal resolution (10-50 km) as a function of altitude (80-100 km) and time. The results are strongly data-driven with minimal assumptions. Rigorous output uncertainties are derived from prior confidence, measurement errors, and sampling density. Covering the United States would require on the order of mid-scale funding. Such a system would provide insight into the lower atmosphere forcing of the upper atmosphere, using technology already proven in Germany.
DATA SCIENCE, ANALYTICS, AND ENSEMBLE MODELING
Geospace Data Systems Infrastructure: Current Needs and Status
Tai Yin Huang, National Science Foundation (NSF)
https://vimeo.com/showcase/9407816/video/694079474
There are many unmet data infrastructure needs. NSF has many solicitations to support data infrastructure. Challenges at NSF in this area were also highlighted. Training courses to how to use data resources and to improve open access is recommended. Collaboration with federal agencies, a community workshop, and other preliminary findings are recommended to address unmet infrastructure needs.
The “Silent” Technology We Need: The Earth and Space Science Knowledge Commons
Ryan McGranaghan, Orion Space Solutions LLC; NASA Goddard Space Flight Center
https://vimeo.com/showcase/9407816/video/694080757
What emerging topic(s) are we missing? This poster describes the idea of the “Knowledge Commons,” a combination of intelligent information representation and the openness, governance, and trust required to create a participatory ecosystem whereby the whole community maintains and evolves this shared information space. In describing the history of knowledge graphs and commons across contexts and nascent work to create them for heliophysics, we identify the technologies that are prerequisite to more robust, responsive, and responsible data assimilation, machine learning, and ensemble modeling. In organizing thoughts to guide the committee and community around these underlying technologies, we arrive at very concrete recommendations for the future of space weather operations and research and for our increasingly wide-reaching community.
Improving Predictive Skill with Ensemble Modeling
Steven K. Morley, Los Alamos National Laboratory
https://vimeo.com/showcase/9407816/video/694080162
This presentation noted that it is critical to capture uncertainty and to propagate it through simulations. The SWMF and Minimal Substorm Model (MSM) were related to geometrical hazards. Ensemble modeling allows greater fidelity than single-forecast models where physical models can be outperformed. Ensemble modeling is tractable and can be achieved at modest computational cost, and can improve the skill of predictions
MACHINE LEARNING
Convection Patterns Based on Machine Learning
William Bristow, Penn State University
https://app.smartsheet.com/b/publish?EQBCT=32396a6d7b4d4d85b61d72979abfc257
The poster describes a study that combines data from the Super Dual Auroral Radar Network (SuperDARN) with a machine-learning (ML) model of convection based on five years of SuperDARN observations. The model is driven with a set of parameters: measures of the magnetospheric state (indices), solar wind/IMF parameters and drift vectors resolved into N-S and E-W, in each latitude-MLT grid cell. Three regressor algorithms provided by the Scikit-Learn software package were tested for forming the model. The Random Forest Regressor produced the lowest root mean square error. The ML model seems to develop a “memory” that responds to expansion and contraction of the polar cap in latitude in response to changing Al and Au indices.
Magnetoseismology for Space Weather Operations and Research
Peter Chi, University of California, Los Angeles
https://vimeo.com/showcase/9407816/video/697514066
Seismology is a well-established method. Ground-based networks can make these measurements. Plasma density profiles can be measured where travel-time magneto-seismology can give information about the magnetotail. Machine learning is needed to facilitate big data analysis.
Prediction of Near–Bow Shock Solar Wind Conditions
Terry Liu, University of California, Los Angeles
https://vimeo.com/showcase/9407816/video/694076888
Multipoint observations are needed to capture near–bow shock localized solar wind conditions. New concentric spacecraft constellations with combination of physics-based prediction models, assimilative reconstruction, and machine language including data from Magnetospheric Multiscale (MMS) mission and THEMIS can be used to train predictive models. This can lead to ten minute to one hour forecast warnings. Predicting near-bow shock solar wind is needed for this observation.
Machine Learning for Thermosphere Operations and Science
Piyush M. Mehta, West Virginia University
https://vimeo.com/showcase/9407816/video/694084508
The thermosphere is the largest source of uncertainty in LEO operations. Performance comparisons including during geomagnetic storms show that data-driven machine learning methodologies can help improve the overall fidelity, resolution, and accuracy of both empirical and physics-based models. They can help improve scientific understanding of processes and provide a capability for robust and reliable uncertainty quantification.