CD6 – Flare research, forecasts and industry: connecting the pieces

Talks

CD6 Tue 5/11 09:00-10:15, room C2D-Almedina

Author(s): Manolis K. Georgoulis

Johns Hopkins Applied Physics Laboratory

Abstract: A recent COSPAR International Space Weather Action Teams (ISWAT) review and roadmap paper pertained to the prediction of solar energetic events, namely solar flares, coronal mass ejections (CMEs) and solar energetic particle (SEP) events. We briefly review the main findings of this paper when solar flares are concerned, but go a step further: besides the fact that flare forecasting was historically humanity’s first attempt to forecast solar weather, we explain why flare prediction is still important and relevant. We further argue that flare forecasting may, in fact, be more difficult than CME forecasting, given stochasticity and the categorization of flares into eruptive (i.e., CME-associated) and confined, or non-eruptive, events. Therefore, we stress the importance of a test yet to be performed: that of the potential gains between forecasting of confined vs. eruptive flares as opposed to forecasting flares indiscriminately.

As explained in the recent COSPAR/ISWAT review, AI methodologies (machine [ML] and deep [DL] learning) have taken over more conventional flare forecasting efforts. We briefly discuss the main categories of these methodologies and speculate on why they do not seem to work as envisioned. Deep learning might have been holding better promise, but it does not come without a price. Indeed, DL does not necessarily rely on any physical interpretation, that may be useful to the industry if successful, but is a showstopper for the research-to-operation-to-research (R2O2R) loop that is actively pursued as a strategic priority around the world. A key to circumventing this stalemate is explainable ML / DL, in the form of solvers of partial differential equations, which brings physics back to the table. This does not come without challenges, however, at least for flares, with the main question being which are the most pertinent physics equations.

One clear path forward despite the above challenges, if not controversies, is a proposed way of evaluation of given prediction methods, from input data to performance verification. Similar ways have been proposed and we present one of them, that has been tested in flare prediction in the recent past and has afforded us the only strict comparison between many different operational flare prediction methods, including some ML and DL ones.

Author(s): KANYA KUSANO

Nagoya University

Abstract: The prediction of solar flares is essential for developing space weather forecasts. However, although various models have been developed so far, the onset mechanism is not sufficiently elucidated, and the prediction of solar flares still needs to be advanced. We have recently developed a new prediction scheme for large flares, i.e., the kappa scheme, based on the MHD instability theory (Kusano et al. 2020). The kappa scheme can accurately predict large flares’ onset and is capable of pinning down the precise location of flares for most events. The successful prediction suggests that the distribution of the twist flux density on the polarity inversion line determines when and where a flare occurs and that a small-scale reconnection may trigger a large solar flare. However, the kappa scheme is imperfect and fails to predict specific flares in a few cases. In this talk, I report on the recent development of physics-based flare prediction and propose the mechanism of flare onset based on the predictive analyses.
References[1] Kusano K, Iju T, Bamba Y, Inoue S. A physics-based method that can predict imminent large solar flares. Science. 369, 587–591 (2020).

Author(s): Yang Chen

University of Michigan

Abstract: We present novel statistical methods for early forecasting of solar flare events and compare them with machine learning approaches we adopted in our previous work. The results that will be presented include (1) strong and weak flare classification with spatial statistics features, together with physics and topological parameters; (2) active region-based solar flare intensity prediction with a mixed Long-Short Term Memory (LSTM) regression; and (3) Tensor Gaussian Process with Contraction model for solar flare forecasting combining data of various types and sources (PIL, HMI and AIA images). We will also compare the advantages and disadvantages of off-the-shelf and newly-developed machine learning methods and discuss future directions for research.

09:45 Flare Forecasting Challenges and Bayesian Approach For Uncertainty Quantification, Grégoire Francisco
09:55 Assessment of Ionospheric Responses to Solar and Meteorological Drivers in near-equatorial region, Ozuem Edward
10:00 Ionospheric and radio propagation effects related to solar flares, Biagio Forte
10:05 Probing the Magnetic Connectivity between Solar Flares and Energetic Particle Events with Solar Orbiter Measurements, Nils Janitzek
10:10 Ground-Based Radio Observations Of Space Weather Events With Optimized Callisto Stations, Malte Bröse

Posters

Posters I  Display Tue 5/11 – Wed 6/11, room C1A – Aeminium

Authors in attendance: Tue 5/11 10:15–11:30, 15:15-16:15; Wed 6/11 10:15–11:30

Author(s): Ozuem Victor Edward, Victor U. J Nwankwo, Omodara Obisesan, Oluwaseun Fatoye, Olufemi OYANAMEH, Timothy Akinsola, Samuel Okoro, Ndifreke Ebong, Vikto Fedun, William Denig

Centre for Space Research, Anchor University Lagos; Centre for Space Research, Anchor University Lagos; Centre for Space Research, Anchor University Lagos; Centre for Space Research, Anchor University Lagos; Centre for Space Research, Anchor University Lagos; Centre for Space Research, Anchor University Lagos; Centre for Space Research, Anchor University Lagos; Centre for Space Research, Anchor University Lagos; University of Sheffield, UK; Department of Sciences, St Joseph’s college of Maine, Standish, ME, 04084, USA

Abstract: The ionosphere is a crucial layer of Earth’s atmosphere and is highly dynamic and sensitive to various perturbations from both solar and terrestrial sources. These perturbations can significantly impact radio communications across various frequency ranges thereby causing issues such as signal absorption, fading, and delay. Understanding these effects could give insight to appropriate strategies to mitigating potential threats to technological systems in the face of ionospheric variability. Our ground-based receivers at Anchor University Centre for Space Research (CESPAR) records solar-terrestrial data for observation and study of atmospheric and ionospheric irregularities. In this study, we investigate the impact of these drivers on the near-equatorial ionosphere by combine analysis of the propagation characteristics of very low Frequency (VLF) radio waves (received at CESPAR) with proxies of solar (and geomagnetic) and meteorological drivers. We thus present results that reflects the prompt impact of sudden ionospheric disturbances (SID) and delayed geomagnetic perturbations in the lower ionosphere, as well as irregularities that appear to be connected to meteorological dynamics. This multi-parametric integrated approach enhances our understanding of ionospheric perturbations, contributing to better space weather models and communication system reliability.

Author(s): Marianna Korsos

University of Sheffield

Abstract: The integration of long-term and short-term solar flare predictions is a crucial component of space weather forecasting, given their potential impacts on Earth’s technological infrastructure and astronaut safety. This presentation examines the importance of long-term and short-term solar flare prediction methods and the potential benefits of combining them to enhance the reliability and precision of forecasts.
Long-term predictions (weeks) provide a broad understanding of solar activity, facilitating better preparedness for heightened periods of solar activity. In contrast, short-term predictions are based on studying the evolution of individual solar active regions in the solar atmosphere, offering warnings within hours or a daily timeframe.
By combining long-term and short-term insights, a more robust and effective solar flare prediction framework can be established. This comprehensive approach enhances the accuracy of specific flare event predictions and significantly advances our understanding of solar dynamics.

Author(s): Larisza Krista

University of Colorado/CIRES, NOAA/NCEI

Abstract: The Detection and EUV Flare Tracking (DEFT) tool automatically identifies flare precursors in EUV observations in a fast and consistent manner, with minimal computational overhead. DEFT currently uses GOES/SUVI 304 A observations to detect and locate high intensity pre-flare signatures. The detected signals are grouped and tracked in consecutive observations in order to isolate and flag sudden impulses that could be precursors to flares. In this study we analyzed precursors before 351 flares (150 C, 150 M and 51 X class flares) that occurred between 2017- to date. Across these magnitudes, precursor signatures were detected for 93% of the flares when using a 6 hour window before the flare start times.
Using superposed epoch analysis, we found that elevated precursor activity tends to occur across all magnitude flares in the last two hours before the flares. The frequency of precursors gradually increases before M class flares, but decreases for C class flares. We also found that in the last 20 minutes before the flare there is a significantly higher precursor frequency, pixel count and power associated with M class flares than C class flares.
We suggest that the observed EUV precursors are the observable signatures of small-scale magnetic reconnection events, and the consistently increasing frequency of precursor activity could indicate that the region is becoming increasingly unstable and reaching a critical stage that could results in flare initiation. Continuing research on EUV precursors is essential to better understand pre-flare processes that build and reduce magnetic instability prior to main-stage flares. The consistent and reliable detection and differentiation of EUV precursors could also complement and significantly improve current flare forecasting efforts.

Author(s): Grégoire Francisco, Dario Del Moro, Teresa Barata, João Fernandes

Department of Physics, University of Rome Tor Vergata; Department of Physics, University of Rome Tor Vergata; IA, Instituto De Astrofisica E Ciências Do Espaço, University of Coimbra; CITEUC, Geophysical and Astronomical Observatory, University of Coimbra

Abstract: Solar flares present significant risks to human technology and motivate the need for forecasting models. However, current models lack precision and reliability. In this talk we will discuss some challenges to be addressed to improve the current state of the art, as well as an alternative approach to traditional binary classifiers.
We first examine classical performance evaluation metrics and identify key weaknesses of current flare forecasting models. In particular, we show that current models are not able to forecast a change in activity significantly better than  random guessing. We compare models using point-in-time magnetograms, coronal EUV observations, as well as time-series modeling of the flare activity and find this challenge to be consistent across all of these approaches.
Next, we propose regressing the SXR maximum peak flux and  the total SXR-fluence emitted by flares within 24 hours time windows. Using Bayesian deep learning methods, we subsequently generate reliable probability distributions of the regressed targets. The resulting models not only achieve performance comparable to traditional binary classifiers but provide more detailed insights with quantification of prediction uncertainties. We also observe that the total flare-emitted fluence is slightly more explainable compared to the traditional maximum peak flux. The SXR-fluence thus appears to better characterize the solar activity over a given time window while offering valuable information for astronauts, directly relating to the radiation amount to which they are exposed.

Author(s): Biagio Forte, Paul Kinsler, Tianchu Lu, Mario M. Bisi, Richard A. Fallows, Pawel Flisek, Kacper Kotulak, Andrzej Krankowski, Bartosz Dabrowski, Pietro Zucca

(1) Department of Electronic and Electrical Engineering, University of Bath, UK; (1) Department of Electronic and Electrical Engineering, University of Bath, UK; (1) Department of Electronic and Electrical Engineering, University of Bath, UK; (2) RAL Space, United Kingdom Research and Innovation, Science & Technology Facilities Council, Harwell Campus, Oxfordshire, OX11 0QX, UK; (2) RAL Space, United Kingdom Research and Innovation, Science & Technology Facilities Council, Harwell Campus, Oxfordshire, OX11 0QX, UK; (3) Space Radio-Diagnostics Research Centre, University of Warmia and Mazury in Olsztyn, Poland; (3) Space Radio-Diagnostics Research Centre, University of Warmia and Mazury in Olsztyn, Poland; (3) Space Radio-Diagnostics Research Centre, University of Warmia and Mazury in Olsztyn, Poland; (3) Space Radio-Diagnostics Research Centre, University of Warmia and Mazury in Olsztyn, Poland; (4) ASTRON, Oude Hoogeveensedijk 4, 7991 PD Dwingeloo, The Netherlands

Abstract: Adverse space weather conditions originated by solar events, induce modifications of the state of the Earth’s ionosphere, as a consequence of the combination of solar and geomagnetic activities within the framework of the neutral atmosphere.
Radio signals propagating through the ionosphere can experience effects due to the dispersive nature of the ionosphere together with the presence of inhomogeneities (or irregularities) in the plasma density distribution. These radio propagation effects are related to ionisation and plasma instability mechanisms typically occurring in the D, E, and F regions of the ionosphere.
In the case of satellite radio signals, radio propagation effects can translate in disturbances and outages to applications and services enabled by those satellites. An example is provided by radio signals transmitted from the Global Navigation Satellite Systems (GNSS).
Solar flares induce a variety of effects in the ionosphere: these are a direct manifestation of the presence of adverse space weather conditions originating with solar events. These effects include an increase in ionisation and the presence of solar radio bursts.
This contribution discusses effects of solar flares on radio signals utilised, for example, in satellite applications near Earth (such as GNSS), the implication on applications reliant on GNSS, and the challenges related to the forecast of an impact. Within the framework of the RISER project (funded by the UKRI Natural Environment Research Council – NERC), this contribution also illustrates opportunities provided by modern instruments (e.g. LOFAR) in the observation of the effect of solar flares and how these could be utilised in a forecasting context.

Author(s): Mario M. Bisi, Biagio Forte, Richard A. Fallows, Hanna Rothkaehl, Pietro Zucca, Barbara Matyjasiak, Mariusz Pożoga, Oyuki Chang, David Jackson, Edmund Henley, Siegfried Gonzi, Bernard V. Jackson, Matthew Bracamontes, Dusan Odstrcil, David Barnes

UKRI STFC RAL Space, UK; University of Bath, UK; UKRI STFC RAL Space, UK; CBK-PAN, Poland; ASTRON, The Netherlands; CBK-PAN, Poland; CBK-PAN, Poland; UKRI STFC RAL Space, UK; Met Office, UK; Met Office, UK; Met Office, UK; UCSD, USA; UCSD, USA; GMU/NASA, USA; UKRI STFC RAL Space, UK

Abstract: The use of radio observations for space-weather studies has been increasing for many decades; from studying the ionosphere and solar emissions through to looking at distant, compact radio signals and pulsars to study the corona and the inner heliosphere.  Various radio-astronomy techniques and sophisticated modelling techniques have also been employed to make the most out of current radio-observing capabilities and also to enhancing blue-skies capabilities in the ongoing need to better understand the many elements of space-weather impacts as well as their drivers (solar wind, coronal mass ejections – CMES, flares, and suchlike) emanated from our local star, The Sun.  Over the years radio has featured more in the monitoring and possible prediction of various space-weather phenomena.  In this presentation, concentrating on the VHF radio-frequency range, we will provide a brief historical overview of recent radio observations including current capabilities, and we will outline avenues of where and how radio is going towards being used in operational space-weather forecasting.  We will endeavour to provide some examples and describe the possible pathways of using radio observations operationally outlining the constraints, requirements, drawbacks, and advantages of doing so – particularly from the ground.

Author(s): Nils Janitzek, Fabian Kistler, Louise Harra, Krzysztof Barczynski, Mario Roco, Raul Gomez Herrero, Alexis Rouillard, Muriel Stiefel, Samuel Krucker, Robert Wimmer-Schweingruber, Javier Rodriguez-Pacheco Martin

PMOD / ETH Zürich; ETH Zürich; PMOD / ETH Zürich; PMOD / ETH Zürich; Universidad de Alcala; Universidad de Alcala; Institut de Recherche en Astrophysique et Planetologie; FHNW / ETH Zürich; FHNW / UC Berkeley; CAU Kiel; Universidad de Alcala

Abstract: Flare-associated eruptions in the solar atmosphere are a well-known cause of solar energetic particle (SEP) events throughout the solar system. To probe the connection between these eruptions and SEP events in the inner heliosphere, we combine data from two instruments on Solar Orbiter with a recent model of magnetic connectivity: The Spectrometer/Telescope for Imaging X-rays (STIX) measures hard X-rays associated with solar flares and allows the identification of the flare location on the Sun. The Magnetic Connectivity Tool (developed at the Institut de Recherche en Astrophysique et Planetologie) models the solar magnetic field in the solar atmosphere and in interplanetary space to predict the magnetic connectivity of the flaring region to several spacecraft in the inner heliosphere. This modeled connectivity can be then compared to the onset of SEP events measured in-situ with the Energetic Particle Detector (EPD) onboard Solar Orbiter – as such events are expected to be measured if Solar Orbiter is actually connected to the eruption site. Based on a SEP onset algorithm, that is optimized for the automatic detection of energetic electron events with EPD, we derived a list of more than 150 SEP events that are estimated to be connected to solar flares between February 2021 and April 2023. We analyze these events statistically with respect to the size and location of the flares on the solar disk as well as their associated SEP electron intensity measured at Solar Orbiter. We further compare the list to a catalogue of manually detected flare-associated SEP events. Overall, the study shows that an automatized linkage between flaring activity on the Sun and a given spacecraft in the inner heliosphere is feasible and provides valuable additional information on the event characteristics. Yet, the precise prediction of magnetic connectivity for future flare-associated SEP events remains challenging.

Author(s): Malte Bröse, Dr. Daniela Banyś

German Aerospace Center (DLR); German Aerospace Center (DLR)

Abstract: Ground-based observations of accelerated electrons in the solar corona are a central part of space weather monitoring and research. Harmful impacts on technological systems are costly and should therefore be prevented in times of rising activity in space. The DLR contributes to this international task by providing continous and reliable radio observations recorded with its own CALLISTO (Compound Astronomical Low frequency Low cost Instrument for Spectroscopy and Transportable Observatory). A small network is built out of four combined CALLISTO stations. Each of the stations will be equiped with receivers for 10 – 80 MHz, 100 – 800 MHz, and 1000 – 1600 MHz. The recorded data from single stations is already used for both space weather services and scientific analysis of solar flares or CMEs. Various types of radio bursts can be studied across a broad frequency range. The original receivers are optimized in hardware and software to increase the signal-to-noise ratio, ease maintenance work, and provide improved comparability. We highlight the scientific aspect by presenting selected events from this year. It is shown how the evolution of space weather processes can be investigated with these radio instruments.

Author(s): KD Leka, Eric Wagner, Kiran Jain, Amr Hamada, Lisa Upton, Bibhuti Kumar Jha, Kathryn Whitman

NWRA and Nagoya University / ISEE; NWRA; NSO; NSO; SwRI; SwRI; NASA/SRAG

Abstract: Forecasting solar energetic events often starts with forecasting solar flares.  Yet solar flare forecasting is presently limited to essentially Earth-facing active regions or just slightly beyond the West limb (by invoking longer-range validity or latency periods).  In Solar Cycle 25, we have witnessed a number of beyond-limb flares that were related to geo-effective impacts including communication disruption and major SEPs.  As we become more concerned about the impact of space weather throughout the heliosphere due to increasing presence of missions and eventually humans off of the Sun-Earth line, it is imperative to bring forecasting to the full-Sun.
We present the first results from a 4-pi solar flare forecasting system developed under the NASA “Research 2-Operations” program.  We incorporate far-side helioseismic results (mapping the solar seismic signals to magnetic flux concentrations and their characteristics) as input to magnetic flux transport maps which model the evolution of magnetic concentrations such as Active Regions in the areas where data are not available for assimilation otherwise, e.g. the “unseen” (or un-seeable with today’s instruments) far-side hemisphere. Specifically, for this proof-of-concept demonstration, we utilized an operationally-running improved helioseismic data pipeline from the Global Oscillations Network Group (GONG) and the Advective Flux Transport model [Upton & Hathaway 2014], augmented with a new Active-Region identification and tracking system.  Significant infrastructure was needed to develop the ability to first evaluate and then ingest the far-side data, including addressing such mundane issues of active-region numbering (to which we suggest the need for a new system based not on assigned incremental numbering, but rather by Carrington-coordinates and time-of-detection).  Flare forecasts are then produced using the characteristics of the magnetic field, plus prior flare history, of the identified active regions.
We utilize quantitative performance metrics from the NWRA Classification Infrastructure (NCI), the research arm of the Discriminant Analysis Flare Forecasting System [DAFFS, Leka+2018], to gauge improvement gained by including this additional information.  Given the data available for validation, we concentrate on forecasts for limb- and beyond-limb regions that were nonetheless predicted to produce Earth-visible events.  This effort also demonstrates the need for 4pi event lists and 4pi infrastructure coordination for further validation.
Industry participation from NASA Johnson’s Space Radiation Analysis Group is guiding the effort.  NASA ROSES 21-SWR2O2R grant #80NSSC22K0273 funds this work.