Space Weather and Heliophysics Modelling Workshop

7 Apr 2026, 08:45 → 9 Apr 2026, 18:00 Europe/Amsterdam

ESOC Press Centre

Jorge Amaya (ESA/ESOC)

About the workshop

This workshop will bring together Europe’s leading developers working at the crossroads of numerical modelling and space physics. It’s an opportunity for experts to showcase breakthrough research and innovative applications that advance our understanding of the Sun–Earth system.

Beyond scientific insights, the event will spotlight the evolution of next-generation forecasting tools, transforming research into operational capabilities for real-world space weather applications.

Whether you’re driving technological innovation or shaping the future of predictive modelling, this is the platform to connect, collaborate, and lead.

Fee

Participation to the workshop is free of charge, but registration is mandatory and subject to approval by the organiser.

Objectives

This event intends to create an environment of open dialog, bringing European experts that can help shape the future of this domain. To achieve this goal we have identified a concrete list of objectives:

• Map the existing modelling capabilities of different institutions in Europe.
• Present the latest results and development plans for the near future.
• Provide critical inputs for the ESA Space Weather Modelling Roadmap.
• Discuss critical issues blocking the transition towards reliable forecasting tools.
• Define ambitious objectives for the European space physics modelling community.
• Share experiences, ideas, and tools.
• Monitor the progress of emerging technology that can radically change our approach to modelling, including computing architectures, AI/ML, quantum computing, data assimilation, new scientific and operational missions.
• Exchange ideas on how to access the necessary resources to achieve the objectives of the community.

 

Contributions

The organisers are inviting recognised experts in the area of numerical modelling of space weather, heliophysics, and space physics, to contribute with their expertise. All oral presentations will be held in-person. Each talk is expected to describe the development status of a single numerical model or modelling framework.

⚠️ The number of available places is limited. Presentation slots will be assigned with priority on the order of submission of abstracts.

📺 While the presentations will be in-person, they will be also streamed online to registered participants.

ESA PRIVACY NOTICE and CONSENT FORM Space Weather Modelling Workshop

Tuesday 7 April

08:45 → 09:30 Badge Collection

09:15 → 09:45 Welcome coffee

09:45 → 10:00 Workshop

09:45 Welcome

Speaker: Juha-Pekka Luntama (ESA)

2026-SWAHM-Intro.pptx

S2P SWE Prospects.pptx

10:00 → 11:15 The Active Heliosphere

10:00 The ESA Vigil Mission at L5:Heliospheric Modelling Opportunities from a New Perspective

The ESA Vigil mission will be the first operational space weather mission positioned at the Sun-Earth L5 Lagrange point. From this unique vantage point, Vigil will continuously observe solar activity and monitor regions near the Sun-Earth line, enhancing both near real-time space weather nowcasting and forecasting, as well as long-term scientific studies. This presentation will provide an overview of the mission's objectives, payload, and planned data products.

Vigil’s observational payload includes the Compact Coronagraph (CCOR) and the Heliospheric Imager (HI) for tracking the evolution and propagation of coronal mass ejections (CMEs), the Photospheric Magnetic Imager (PMI) for magnetic field mapping and solar wind modeling and forecasting, and the EUV Imager (JEDI), a NASA-contributed instrument, which will capture full-disk and extended coronal observations in multiple EUV passbands out to 6 solar radii. Alongside these are in-situ measurements from a magnetometer (MAG) and plasma analyser (PLA), enabling Vigil to deliver a consistent, calibrated, and openly accessible dataset to both operational and scientific communities.

The mission is designed to provide continuous, 24/7 monitoring—including during severe space weather events—with rapid delivery of low-latency data for operational users. In parallel, high-quality science data will be made available to the research community, adhering to community standards for open science. By bridging operational and scientific domains, Vigil will not only enhance real-time space weather capabilities but also deliver new insights into solar and heliospheric dynamics from a novel, off-Sun–Earth line vantage point.

In the spirit of the workshop, the author would like to encourage the audience to discuss their modelling needs: metadata, telemetry, data products, as well as lessons learned from previous missions.

Speaker: Matthew West (European Space Agency)

West_Space_Weather_and_Heliophysics_Modelling_Workshop_2026.pptx

10:15 Space weather forecasting with sub-L1 solar wind monitors

I will present the pipelines at the Austrian Space Weather Office for solar wind forecasting, using a combination of empirical-, physics- and data based models for modeling the propagation, automatic detection and flux rope characterization of CMEs (ELEvo(HI), ARCANE, 3DCORE). A particular emphasis is given on the usage of sub-L1 and far upstream data, which has just recently become available with STEREO-A and Solar Orbiter. We were now able to predict with empirical- and physics based modeling applied to Solar Orbiter and STEREO-A magnetometer data the geomagnetic effects of CMEs for a few events. Solar Orbiter MAG has observed the 2026 January 18-19 coronal mass ejection at 0.74 au and 8° away from Earth, leading to the strongest interplanetary field at L1 since systematic observations began in the mid-1990s. This event could have caused the strongest geomagnetic storm since 1989 or even 1921, if its flux rope field would have been mainly southward instead of northward. In a highly fortunate scenario, Solar Orbiter MAG allows us to test sub-L1 capabilities way before the arrival of the ESA HENON and SHIELD missions on distant retrograde orbits, even for cases of extreme solar eruptive events.

Speaker: Christian Möstl (Austrian Space Weather Office, GeoSphere Austria, Graz, Austria)

Moestl_esa_modeling_Apr_2026_final.pptx

Moestl_esa_modeling_Apr_2026_final_small.pdf

10:30 Operational Machine-Learning-Enhanced CME Arrival-Time Forecasting and In-Situ ICME Detection

Accurate and timely forecasts of coronal mass ejection (CME) arrival times at Earth are are essential for operational space weather services aimed at mitigating impacts on spacecraft, ground-based infrastructure, and radiation-sensitive systems. We present two complementary machine-learning-based tools that address both CME arrival-time prediction and the identification of interplanetary CMEs (ICMEs) in near-Earth in-situ data, forming a coherent Sun-Earth operational forecasting chain.
PhysNetCME is a physics-driven machine-learning model that combines CME initial parameters derived from coronagraph observations with real-time upstream solar-wind measurements to estimate transit times to 1 AU. A drag-based propagation model is embedded within the neural-network training process, ensuring physical consistency while retaining adaptability to data. An ensemble approach provides deterministic forecasts together with uncertainty estimates, supporting risk-informed operational decision-making.
ICMEAlert complements this capability by continuously monitoring L1 solar-wind plasma and magnetic-field data to detect ICME signatures in near-real time. Using unsupervised anomaly detection based on the Isolation Forest algorithm, the system assigns confidence levels to anomalous structures without relying on labeled event catalogs.
Together, these tools enhance forecast lead time, reliability, and uncertainty awareness, contributing to the transition of advanced modelling approaches toward sustained operational space weather services.

Speaker: Sabrina Guastavino (sabrina.guastavino@unige.it)

slides_2026-04-05_CME.pptx

10:45 Variational Data Assimilation for Improved 3D Global MHD Forecasting of the Ambient Solar Wind

Over the past several decades, there has been a growing recognition of the adverse impacts that space weather can pose to human infrastructure, activity, and health. Consequently, as modern society becomes increasingly reliant on vulnerable technological systems, there is a corresponding demand for accurate and reliable space weather forecasting capabilities. Successful space weather forecasting requires precise knowledge of the ambient, or quasi-steady background, solar wind as this is the medium through which solar disturbances propagate towards the Earth. For example, the ambient solar wind strongly influences the transit time of interplanetary coronal mass ejections and contributes to significant near-Earth space weather effects through the formation of stream interaction regions. State-of-the-art predictions of the ambient solar wind are today based on three-dimensional (3D), global, magnetohydrodynamic (MHD) models. Despite their sophistication and physical fidelity, the predictive potential of these models is not yet fully realized---solar wind forecasts often fail to accurately match observations. While the global MHD models are data-driven and estimate boundary data at the inner inflow heliospheric boundary for the simulations by combining observations of the solar magnetic field and empirical estimates of the plasma properties, these models are typically also "free-running" and unconstrained by in-situ. measurements of the solar wind outflows. Thus, any uncertainties or errors arising from driving inflow boundary conditions can grow unbounded and pollute or contaminate the predictions. To overcome these limitations, this presentation introduces a novel variational data assimilation framework---the first of its kind---that optimally combines in-situ. solar wind observations with the predictive solutions of a global MHD model. A Bayesian inverse MHD problem is solved to obtain the most probable solar wind inflow boundary data for the given set of observational data. An overview of the variational data assimilation approach and several results from observing system simulation experiments conducted using both ACE and STEREO spacecraft data will be presented, with the latter demonstrating the substantially improved predictions of near-Earth solar wind conditions afforded by the proposed data assimilation approach.

Speaker: Jose Arnal (University of Toronto)

SWHMW2026-Arnal.pdf

11:00 Architectural and Performance Optimization of the Energetic Particle Radiation Environment Model (EPREM-CPP)

We have performed a comprehensive redesign of the Energetic Particle Radiation Environment Model (EPREM) to address certain limitations of the original implementation. This new implementation, written in C++, introduces enhancements to address grid resampling artifacts at the inner boundary, as well as time-variable, pitch-anisotropic, and spatially-distributed seed functions to simulate realistic source population variability over the inner boundary of the model. To ensure accurate initial conditions, a dedicated relaxation phase was implemented throughout the domain, based on relative local truncation error to reach steady-state equilibrium before time-dependent flux injection. Computational performance is optimized by transitioning from process-based MPI to a shared-memory thread-pool using work-stealing scheduling, which mitigates synchronization bottlenecks. Finally, the redesigned EPREM-CPP model has been successfully deployed on HPC resources, and an accompanying Python package for configuration preparation and high-level data visualization has been developed.

Speaker: Mr Artem Epifanov (Institute of Astronomy and National Astronomical Observatory, Bulgarian Academy of Sciences)

SWHM2026-EPREM-CPP.pptx

11:15 → 11:45 Coffee break

11:45 → 13:15 Solar Corona & Solar Wind

11:45 Solar and heliospheric modeling with MPI-AMRVAC

The open-source MPI-AMRVAC software [https://amrvac.org and Keppens et al, A&A 673, A66, 2023] is widely used for numerical plasma-astrophysical research, and has a fair number of offspin codes (BHAC, Gmunu, GR-AMRVAC) that routinely perform up to general relativistic magnetohydrodynamic (MHD) simulations. Its modular design is a key feature of the framework, along with the automated grid-adaptivity (block-based AMR) and its high flexibility for user-defined prescriptions. Here, I will give an overview of the latest applications and code developments that target solar and heliospheric modeling, where Newtonian physics is at play, sometimes enriched with non-thermal particle processes. These include the first realistic models of solar prominence formation, studies of particle acceleration in erupting flux rope settings, self-consistent flare models with energetic particle beams included, global solar coronal models, and heliospheric forecasting using the Icarus user module. Multiple ways to numerically handle the governing MHD equations - including many non-ideal effects - are available and can serve to benchmark the optimal method selection for robust operational purposes. I will end with a preview on the AGILE development, where the same code flexibility will efficiently exploit the most modern hybrid CPU-GPU platforms.

Speaker: Rony Keppens (CmPA, KU Leuven)

SWAHM2026.pdf

12:00 Uniturbulence and Alfvén wave solar model (UAWSoM): the new physics-based coronal model

The coronal heating problem remains a central challenge in solar physics, which requires a detailed understanding of wave-based heating mechanisms. While the Alfvén Wave Solar Model (AWSoM) has shown great success in physics-based modelling of the solar corona, its focus on a single heating mechanism leaves room for improvement. This is why we developed the Uniturbulence and Alfvén Wave Solar Model (UAWSoM), which additionally includes heating due to uniturbulent kink wave dissipation. We constructed the theoretical framework necessary to derive the governing kink wave energy evolution equations. This was followed by implementing a 1D version of the model in MPI-AMRVAC, where we showed that kink waves are able to sustain a stable coronal atmosphere without requiring an artificial background heating term. We found that kink wave dissipation is more efficient at heating the lower corona than Alfvén wave dissipation and its inclusion improves model stability. Further, we demonstrate the usage of UAWSoM in 1D coronal loop modelling and compare the results with other analytical and numerical methods. Finally, we present the ongoing efforts, focusing on the development of the 3D version of the model. We envision that the resulting self-consistent 3D coronal model will offer a lot of potential for space weather modelling.

Speaker: Luka Banovic

UAWSoM_Darmstadt_2026.pptx

12:15 Computationally efficient, dynamical solar wind forecasting using in-situ solar wind observations

Accurate prediction of the ambient solar wind at Earth is a key requirement for space-weather forecasting but is limited by uncertainties in coronal boundary conditions and, to a lesser extent, heliospheric transport. We present a solar wind forecasting approach that uses near-Earth in situ solar wind observations to estimate the inner boundary condition for the Heliospheric Upwind eXtrapolation with Time dependence (HUXt) model. In terms of the forecast at Earth, this method is similar to 27-day recurrence forecasting. However, unlike recurrence, it produces a solution throughout the heliosphere which can potentially be used to improve CME propagation estimates and forecasting at a range of in-ecliptic heliospheric locations. To test this forecast method, daily hindcasts are generated over a 15-year interval spanning 2008–2022. Performance is evaluated and directly compared with forecasts generated by the standard operational forecast approach; namely magnetogram-based estimates using the Wang–Sheeley–Arge (WSA) coronal model. The InSitu-HUXt hindcasts achieve median solar wind speed errors around 65 km s$^{-1}$ for lead times up to 27 days. WSA-HUXt hindcasts are most accurate around 4-days lead time, with an MAE of around 85 km s$^{-1}$. More diagnostic metrics show that this is partly because InSitu-HUXt naturally produces a bias-free prediction, whereas WSA-HUXt currently over-predicts the occurrence of high-speed streams in the solar wind by around 30%. Thus InSitu-HUXt provides a valuable independent method to forecast near-Earth solar wind conditions and can be used even when magnetogram data is unavailable, providing forecast resilience.

Speaker: Mr Nathaniel Edward-Inatimi (University of Reading)

HUXt_ESA_2026.pptx

12:30 Improving CME Arrival-time assessment with realistic solar wind backgrounds

RIMAP (Reverse In situ data and MHD APproach) is a data-driven model designed to reconstruct the ambient solar-wind conditions in the solar equatorial plane starting from in situ measurements, coupling analytical backmapping with numerical simulations performed with the PLUTO MHD code. Unlike models primarily driven by remote-sensing observations, RIMAP is built to preserve longitudinal variability and small-scale structures observed in the solar wind (thin streams, sharp gradients, and localized inhomogeneities) that are often smoothed out.
This capability makes RIMAP an ideal test bench to investigate the propagation and interaction of solar transients such as Coronal Mass Ejections within a realistic, structured interplanetary background, enabling controlled assessments of distortion, compression, and deflection effects induced by the ambient wind. This has measurable effects on the predicted arrival times of solar disturbances, depending on the local properties of the solar wind streamlines.
RIMAP has also been operationally integrated into the space weather portal at INAF-Turin Astrophysical Observatory, SWELTO, where a low-resolution version runs in quasi-real time as a product supporting space-weather activities and the validation of new diagnostic routines.

Speaker: Ruggero Biondo (INAF - Turin Astrophysical Observatory)

Biondo_ESAWokshop_CMEArrivalTimes.pptx

12:45 The COCONUT Model for Data-Driven Solar Corona & Heliospheric Simulations

COolfluid COrona uNstrUcTured (COCONUT) [1-10] is a data-driven physics-based model for plasma simulations implemented within the open source COOLFluiD platform (https://github.com/andrealani/COOLFluiD/wiki). The core C++ solver implements a second-order accurate Finite Volume (FV) discretization for arbitrary unstructured grids, is fully parallel through the Message Passing Interface (MPI) also supporting parallel I/O capabilities on 1000s of CPU-cores, and provides efficient implicit time integration schemes already allowing for >20X faster-than-real-time simulations. To this end, the Portable Extensible Toolkit for Scientific Computation (PETSc) library is used for solving the resulting linear system of discretized equations via available GMREs algorithms and parallel preconditioners.
While COCONUT’s baseline version [5][9] solves single-fluid magnetohydrodynamics (MHD) equations, a more advanced multi-fluid/Maxwell model [4] is also under development, in particular to better tackle the collisional and radiative processes in the chromosphere. COCONUT features both a polytropic and a full MHD model [2] including heating source term, thermal conduction and radiation cooling.
All simulations are data-driven, i.e. accepting real (pre-processed) magnetic maps as input at the low solar corona or solar surface from various sources (e.g. ADAPT, HMI, GONG) [8][10], providing the radial component of the magnetic field which is used by a Potential-Field Source Surface (PFSS) solver, based upon COOLFluiD’s implicit FV code, to initialize steady MHD simulations. In time-dependent simulations [11], COCONUT accepts a time series of magnetograms as input with arbitrary cadence and can also propagate flux ropes [6] which are injected directly in its initial field.
COCONUT’s computational domain, which is typically meshed with prismatic cells using a Blender-based tool [3], can extend up to 0.1AU, feeding inputs to inhouse-developed open source heliospheric codes such as EUHFORIA [7] and Icarus [1], or all the way up to 2.5AU, as in the most recent simulation efforts.
This presentation will provide an overview of all the above mentioned features, highlighting COCONUT’s strengths and the best results to date (e.g. for solar minimum and maximum cases, solar eclipse [2], CME propagation [7], Sun-to-Earth simulations through couplings to EUHFORIA [7] or Icarus [1] or using COCONUT standalone), also including some performance considerations and insights into future work.

REFERENCES

[1] Baratashvili, T., Brchnelova, M., Linan, L., Lani, A., Poedts, S. (2024). The operational-ready full 3D MHD model from Sun to Earth: COCONUT + Icarus. Astronomy & Astrophysics, 690, Art.No. A184. doi: 10.1051/0004-6361/202449389.
[2] Baratashvili, T., Wang, H., Sorokina, D., Lani, A., Poedts, S. (2026). Modelling the total solar eclipse in 2024 with COCONUT. Astronomy & Astrophysics, 705, Art.No. A145. doi: 10.1051/0004-6361/202556300.
[3] Brchnelova, M., Zhang, F., Leitner, P., Perri, B., Lani, A., Poedts, S. (2022). Effects of Mesh Topology on MHD Solution Features in Coronal Simulations. Journal Of Plasma Physics, 88 (2), Art.No. 905880205, 1-29. doi: 10.1017/S0022377822000241.
[4] Brchnelova, M., Kuźma, B., Zhang, F., Lani, A., Poedts, S. (2023). COCONUT-MF: Two-fluid ion-neutral global coronal modelling. Astronomy & Astrophysics, 678, Art.No. A117. doi: 10.1051/0004-6361/202346525.
[5] Kuźma, B., Brchnelova, M., Perri, B., Baratashvili, T., Zhang, F., Lani, A., Poedts, S. (2023). COCONUT, a novel fast-converging MHD model for solar corona simulations: III. Impact of the pre-processing of the magnetic map on the modeling of the solar cycle activity and comparison with observations. Astrophysical Journal, 942, Art.No. 31. doi: 10.3847/1538-4357/aca483.
[6] Linan, L., Regnault, F., Perri, B., Brchnelova, M., Kuźma, B., Lani, A., Poedts, S., Schmieder, B. (2023). Self-consistent propagation of flux ropes in realistic coronal simulations. Astronomy & Astrophysics, 675, Art.No. A101. doi: 10.1051/0004-6361/202346235.
[7] Linan, L., Baratashvili, T., Lani, A., Schmieder, B., Brchnelova, M., Guo, J.H., Poedts, S. (2025). CME propagation in the dynamically coupled space weather tool: COCONUT + EUHFORIA. Astronomy & Astrophysics, 693, Art.No. A229. doi: 10.1051/0004-6361/202451854.
[8] Murteira, J., Brchnelova, M., Lani, A., Poedts, S. (2025). Magnetogram filtering techniques for global coronal modelling. RAS techniques and instruments, 4, Art.No. rzaf030, 1-15. doi: 10.1093/rasti/rzaf030.
[9] Perri, B., Leitner, P., Brchnelova, M., Baratashvili, T., Kuźma, B., Zhang, F., Lani, A., Poedts, S. with Perri, B. (joint first author), Leitner, P. (joint first author) (2022). COCONUT, a novel fast-converging MHD model for solar corona simulations: I. Benchmarking and optimization of polytropic solutions. Astrophysical Journal, 936, Art.No. 19. doi: 10.3847/1538-4357/ac7237.
[10] Perri, B., Kuźma, B., Brchnelova, M., Baratashvili, T., Zhang, F., Leitner, P., Lani, A., Poedts, S. (2023). COCONUT, a novel fast-converging MHD model for solar corona simulations: II. Assessing the impact of the input magnetic map on space-weather forecasting at minimum of activity. Astrophysical Journal, 943, Art.No. 124. doi: 10.3847/1538-4357/ac9799.
[11] Wang, H., Poedts, S., Lani, A., Linan, L., Baratashvili, T., Zhang, F., Sorokina, D., Jeong, H-J., Li, Y.C., Najafi-Ziyazi, M., Schmieder, B. (2025). Time-evolving coronal modelling of solar maximum around the May 2024 storm by COCONUT. Astronomy & Astrophysics, 702, Art.No. A37. doi: 10.1051/0004-6361/202555760.

Speaker: Andrea Lani (KU Leuven)

AndreaLani_ESOC_7_4_2026 3.pptx

13:00 Towards novel data-driven coronal and heliospheric space weather modelling

In this presentation, novel time-dependent modelling assets recently developed at the University of Helsinki are presented. We focus especially on highlighting two recent developments: 1) physics-based modelling of the large-scale dynamics caused by space-weather relevant magnetized Coronal Mass Ejections (CMEs) starting at 5 solar radii and extending out to the heliosphere and 2) self-consistent data-driven modelling of the formation and eruption of active region magnetic fields. Combining these methods provides a unique capability to efficiently perform data-driven dynamic modelling of CMEs from the low corona to the heliosphere, with a potential to significantly improve upon current space weather modelling paradigms.

Speaker: Jens Pomoell (University of Helsinki)

13:15 → 14:30 Lunch

14:30 → 15:45 Tracking Solar Activity

14:30 Opportunities for utilization of Vigil-like data and deep learning approach

In this contribution, we will present two machine learning models that employ measurements from already-operated space instruments analogous to those Vigil will have on board (Majirsky et al., 2025a). The first model predicts the occurrence of geomagnetic storms by combining coronagraph images and in situ solar wind and IMF data (Majirsky et al., 2025b). The second model classifies in situ data to provide early detection of CIRs at L5. The details of these models will be presented, and opportunities for follow-up activities to develop operational solutions for the ESA mission Vigil will be discussed.

Speaker: Simon Mackovjak (Institute of Experimental Physics, Slovak Academy of Sciences)

2026_Darmstadt_ESA_workshop_Mackovjak.pdf

14:45 Prevision of phenomena through coronal hole outline recognition with neural network

The properties and the spatial distribution of the large-scale structures of the solar corona determine the observed solar wind structure at 1~au. Coronal holes are a major source of fast solar wind, an important geo-effective component, and appear as large dark patches in extreme ultraviolet images. The solar observatories provide images of the solar corona at different wavelengths, enabling the identification of coronal hole morphology and other large-scale structures along a given line of sight. It is crucial to understand the properties of coronal holes for effective space weather forecasting. The main goal of this work is to develop a threshold-based coronal hole detection tool across solar cycles 23, 24, and 25, using artificial intelligence. Here, we allow the user to input a specific time within solar cycles 23, 24, and 25, enabling the retrieval of the threshold value used to detect coronal hole contours in line-of-sight extreme ultraviolet images from SDO/AIA and SoHO/EIT. We retrieve data from the heliophysics events knowledge database for the large-scale features such as active regions, solar flares, coronal mass ejections, and filaments, and then we engineer them, which allows us to train the neural network model (Prevision Of Phenomena through Coronal-hole Outline Recognition with Neural-network; POP-CORN). The input of the model is the categorical features of the large-scale structures of the solar corona based on their spatial distribution and additional properties, such as solar flare class based on intensity. A neural network model (POP-CORN) was then trained to achieve higher accuracy. The model determines the threshold needed to detect coronal holes, allowing their boundaries to be identified automatically and consistently in extreme ultraviolet images from solar cycles 23, 24, and 25. To interpret the performance of our neural network model (POP-CORN), we divided the predicted coronal hole results into different phases across the solar cycles 23, 24, and 25. Later, we compare them qualitatively and quantitatively with other coronal hole detection tools. We conclude that the properties of large-scale structures affect the determination in coronal hole regions, and incorporating these properties manually into the training improves coronal hole detection. We find that POP-CORN performs well at detecting coronal hole contours, even when many bright features, such as active regions and solar flares, are present, which makes it hard for threshold-based methods to detect dark regions like coronal holes. In the future, we plan to integrate the coronal hole detection tool into a solar wind model validation pipeline, creating a fully automated validation tool that provides a quantitative score for predictions.

Speaker: Kalpa Harindra Perera Henadhira Arachchige (Dept. d’Astrophysique/AIM, CEA/IRFU, CNRS/INSU, Université Paris et Paris-Saclay, 91191 Gif-sur-Yvette Cedex, France)

POP-CORN_ESA_final.pptx

15:00 Operational Active Region Detection, Classification, and Flare Forecasting Using Deep Learning

Early identification of flare-productive solar active regions is essential for operational space-weather forecasting. We present an integrated deep-learning framework for automated active-region detection, localization, magnetic classification, and short-term flare forecasting, designed for continuous monitoring and near-real-time deployment.

The system combines three complementary components. SUN-FD operates directly on full-disk magnetograms, performing simultaneous active-region detection, localization, and magnetic classification via deep learning–based object detection. It outputs bounding boxes together with probabilistic class scores, enabling automated full-disk surveillance with uncertainty-aware estimates.

SUN-ARC refines magnetic characterization by classifying extracted active-region patches into Mount Wilson classes using supervised deep learning. The model captures polarity configurations and morphological complexity and provides calibrated probabilistic outputs suitable for downstream decision-making.

Building on these products, DFF (DeepFlareForecast) predicts the likelihood of flares within the next 24 hours from multi-wavelength, time-ordered sequences of active-region observations (SDO/HMI magnetograms jointly with SDO/AIA channels). The forecasting model follows a spatio-spectro-temporal design: a shared convolutional encoder extracts compact spatial representations per timestep, channel-aware fusion emphasizes the most informative wavelengths, and a lightweight temporal transformer captures the evolution of magnetic and coronal signatures leading to flares. The model produces probabilistic forecasts to reflect flare hierarchy and support threshold-based alerting.

Together, these tools provide a coherent operational pipeline from full-disk magnetic monitoring to active-region characterization and flare alert generation, supporting timely mitigation strategies for space-based and ground-based technological systems.

Speaker: Edoardo Legnaro (University of Genova)

Operational Active Region Detection, Classification, and Flare Forecasting Using Deep Learning

15:15 The Virtual Space Weather Modelling Centre

The ESA Virtual Space Weather Modelling Centre (VSWMC) project was defined as a long-term initiative comprising successive parts. Parts 1 and 2 were completed in the first 4-5 years, and a system was designed and developed that enables models and other components to be installed locally or geographically distributed and to be coupled and run remotely from the central system. A first, limited version went operational in May 2019 under the H-ESC umbrella on the ESA SSA SWE Portal. It is similar to CCMC but interactive (no runs on demand), with models geographically distributed and coupled over the internet. In the ESA project "Virtual Space Weather Modelling Centre - Part 3" (2019-2022), building on the Part 2 prototype system and focusing on the interaction with the ESA SSA SWE system, 11 new models and many new model couplings have been integrated, including daily automated end-to-end (Sun to Earth) simulations, and the front-end GUI was upgraded, making the operational system more robust and user-friendly. In the current VSWPC-P4 project, the capabilities of the coupling framework are being expanded, additional solar and heliospheric models are being introduced, some of the already integrated models are being improved, and steps are being taken towards integrating the framework and its derived products into the ESA space weather portal. Are being completed.
The current state of the art of the VSWMC will be presented, and future plans will be discussed.

Speaker: Stefaan Poedts (KU Leuven)

2026-04 The VSWMC-Poedts.pptx

15:30 S2WARM (St Andrews Space Weather Active Region Monitor)

S2WARM (St Andrews Space Weather Active Region Monitor) recognises eruptive solar active regions by assimilating magnetogram data into 3D NLFFF simulations and projecting their evolution. It computes first a theoretical and then a specific metrics to issue green, amber, or red alerts, factoring magnetic flux changes and Lorentz force evolution. Tested on a full rotation with 23 cases, S2WARM correctly classified most of them and provides some interesting insights.
S2WARM is a model based on the 3D reconstruction of the magnetic field above and active region and its physical properties, hence providing an interesting platform where space weather prediction and physics of the solar corona meet.

Speaker: Paolo Pagano (Università degli Studi di Palermo)

Paolo_Pagano_SWHM_ESA.pptx

15:45 → 16:15 Workshop

15:45 ESA funding opportunities, participation statistics, and team building activities

16:15 → 17:00 Business Meetings with ESA

These are sessions available for the participants to have one-on-one conversations with the ESA Space Weather Office.

Wednesday 8 April

09:00 → 09:30 Badge Collection

09:30 → 10:00 Welcome coffee

10:00 → 11:15 Geomagnetic Storms and Indices

10:00 Space Weather forecasting models for onboard operations

Solar Energetic Particles (SEPs) and geomagnetic storms are major Space Weather events which require efficient forecasting for ensuring mitigation against their harmful effects for technological systems and humans. We have implemented a machine learning model (Stumpo et al., 2024), based on the Random Forest Regressor algorithm, to forecast SEP events at the Earth by using in situ observations of only energetic electrons. The model can provide a reliable prediction of the >10 MeV proton flux expected at the Earth with an advance of 1 hour (i.e., before an increase of the proton flux is directly measured). For forecasting geomagnetic storms, we have developed a method based on an Artificial Neural Network (ANN) for making a real-time regression of SYM-H index. We adapted the EDDA (Empirical Dst Data Algorithm, Pallocchia et al. 2016) algorithm, using only magnetic field data, to predict the Sym-H index 1 hour ahead every 20 minutes. We have evaluated both models and have obtained a good performance in both cases.
These models are especially suited for operations onboard next generation spacecraft aimed at Space Weather and\or human exploration. For instance, they will be operating onboard the HEliospheric pioNeer for sOlar and interplanetary threats defeNce (HENON) mission (Provinciali et. al., 2024) to provide timely and reliable alerts for potential harmful Space Weather events. In particular, SEPs will be predicted onboard for the first time, offering an unprecedented test for an advanced alert system for future constellations (e.g., SHIELD) or manned missions. Moreover, geomagnetic storms will be predicted with a 10 time improvement in the lead times with respect to current predictions.

Speaker: Monica Laurenza

Laurenza_Heliophysics_2026.pptx

10:15 Multi-Horizon Operational Forecasting of Geomagnetic Storms Using L1 Solar-Wind Observations

Providing actionable forecasts of geomagnetic storm occurrence on timescales of several hours is essential for space weather services aimed at protecting critical infrastructure and supporting operational decision-making. We present GeoStormAlert, a machine-learning-based forecasting system that leverages real-time in-situ solar-wind measurements at L1 to predict geomagnetic storm conditions with lead times up to four hours.
GeoStormAlert ingests high-cadence solar-wind plasma and interplanetary magnetic-field observations, together with derived physically meaningful parameters such as magnetic energy and helicity. These upstream quantities describe the drivers of magnetospheric disturbances. The inputs are processed by a Long Short-Term Memory (LSTM) neural network designed to capture temporal dependencies and the evolving solar-wind–magnetosphere coupling. To enhance robustness and interpretability, feature-ranking techniques are applied to assess the relative importance of input variables and to identify the most predictive subset across multiple forecast horizons.
The system generates warning alerts when geomagnetic indices are forecast to cross operational thresholds associated with moderate to intense storms. By delivering multi-horizon alerts, GeoStormAlert enables timely, risk-informed decisions for satellite operators and infrastructure managers.

Speaker: Michele Piana (University of Genova - Department of Mathematics)

slides_2026-04-05_geo.pptx

10:30 Data Driven Advances in SERENADE: Larger Datasets, Stable Forecasts

Machine‑learning techniques, whether supervised, self‑supervised, or unsupervised, have become indispensable tools in the modelling of space weather in recent years. By tapping into the vast, heterogeneous archives collected over decades, they produce prediction models that are both swift and highly accurate, often rivalling or even surpassing traditional physics‑based approaches, though sometimes at the cost of interpretability.
The presentation will centre on SERENADE, a deep‑learning pipeline that predicts the daily maximum Kp index several days in advance from SDO/AIA EUV images at 193 Å. Initial prototypes already achieved performance comparable to the state of the art for fast‑solar‑wind‑driven events, yet several limitations were identified. Recent work has replaced the generic GoogLeNet latent‑vector extractor with a variational auto‑encoder specifically trained on solar images, yielding a more physically‑meaningful latent space and more stable forecasts.
A major thrust of the current effort is the enrichment of the training database. Beyond the original SDOML collection (2010‑2020), we have assembled a new ML‑prepared dataset that spans 2010‑mid‑2025 (including the rising and maximum phases of the present, more active cycle) and a complementary SOHO/EIT dataset extending back to 1996, thereby providing nearly 30 years of observations. By analysing the impact of this temporal extension, we explore cross‑instrument generalisation (zero‑shot learning) by training on one instrument’s data and testing on the other, a capability that will become crucial once the SDO and SOHO missions are retired.
This work highlights critical aspects of data preparation and model assessment, and demonstrate how expanding and harmonising large data archives can improve the reliability and operational readiness of space‑weather forecasts.

Speaker: Guillerme Bernoux (ONERA)

BERNOUX_ESOC_Workshop.pptx

10:45 Days ahead forecast of geomagnetic activity

The geomagnetic index Kp has widespread use in space weather due to the apparent simple interpretation and due to the close relation to the upstream solar wind. Clearly, Kp also has limitations for space weather but we will not discuss that here.

From an L1 monitor, Kp can be forecast with high accuracy with a lead time of a couple of hours, under the assumption that high resolution (minutes) uncorrupted magnetic field and plasma data exist. To extend the forecast lead time, solar observations and heliospheric models are required with for example their L1 forecasts feeding into L1-Kp-models. By evaluating 3-day Kp forecasts from different providers we see that it is extremely difficult to make detailed accurate forecasts in the general case. We discuss whether it is possible to reformulate forecasts to more accurately capture current capabilities in order to provide more informative forecasts. We describe one approach for probabilistic forecasts for G4 and G5 events relevant for power grid operators.

Speaker: Peter Wintoft (Swedish Institute of Space Physics)

wintoft20260407esoc.pptx

11:00 Disturbance Storm Time Index Prediction with Interpretable Machine Learning

The Disturbance Storm Time (Dst) index quantifies geomagnetic storm intensity by measuring global magnetic field variations. In this study, we apply interpretable machine-learning (ML) techniques to derive data-driven models describing the temporal evolution of the Dst index. We use historical data from the NASA OMNIWeb database, including solar wind density, bulk velocity, convective electric field, dynamic pressure, and magnetic pressure. We employ KAN networks and the symbolic regression framework PyOperon, based on an evolutionary algorithm, to identify closed-form expressions linking dDst/dt to key solar wind parameters. The equations obtained via symbolic regression form a hierarchy of complexity levels and capture nonlinear dependencies and threshold effects in Dst evolution. In addition, we use a conventional MLP network as a reference black-box model. We benchmark all ML models against observed Dst data and compare their performance with empirical formulations such as the Burton-McPherron--Russell and O'Brien-McPherron models. The performance evaluation on historical storm events includes the 2003 Halloween storm, the 2015 St. Patrick's Day storm, a moderate storm in 2017, and the extreme storm of May 2024. The data-driven models, particularly the MLP, demonstrate superior accuracy in most cases. While the symbolic regression expressions provide insight into the underlying physics, the results highlight an intrinsic trade-off between model interpretability and predictive accuracy.

Speaker: Jonah Ekelund (KTH Royal Institute of Technology)

2026-04-06_SpaceWeather_ESA.pdf

11:15 → 11:45 Coffee break

11:45 → 12:45 Magnetosphere-Thermosphere-Plasmasphere

11:45 THERION Retrieve Thermospheric parameters at mid and equatorial latitudes

Monitoring the thermospheric state is a highly relevant problem due to its impact on ionospheric dynamics and satellite drag. A novel approach based on routine ionosonde observations, originally developed for mid-latitudes (Perrone and Mikhailov, 2018), has proven effective in retrieving a self-consistent set of key aeronomic parameters and is now widely used in our analyses.
The method, known as THERION (THERmospheric parameters from IONosonde observations), uses as input: (1) plasma frequency at 180 km derived at selected local times (10–14 LT) from automatically scaled bottom-side electron density profiles Ne(h) obtained with ionogram inversion codes such as AUTOSCALA; (2) the noon-time foF2 value; and (3) solar (F10.7) and geomagnetic (Ap) indices. An advanced version of the method additionally includes neutral density from satellite observations as fitted parameters.
The retrieved parameters include the neutral composition (O, O₂, N₂ concentrations), exospheric temperature (Tₑₓ), the total solar EUV flux at wavelengths λ ≤ 1050 Å, and the vertical plasma drift W, which at mid-latitudes is mainly related to the meridional thermospheric wind component.
The method has been recently extended to low-latitude and equatorial regions (Perrone and Mikhailov, 2025), where it is applicable around local noon at the geomagnetic equator. In this case, the retrieved vertical plasma drift is primarily controlled by the zonal electric field component (Eᵧ). This unified approach provides a potential valuable tool for monitoring thermospheric conditions using long-term, routinely available observations.
Refernces:
Perrone, L., Mikhailov, A. V. (2018). A New Method to Retrieve Thermospheric Parameters From Daytime Bottom-Side Ne(h) Observations. J. Geophys. Res., 123,10,200–10,212. https://doi.org/10.1029/2018JA025762.

Perrone, L., & Mikhailov, A. V. (2025). A method to retrieve aeronomic parameters from noontime bottom‐side equatorial Ne(h) and satellite neutral density observations. Space Weather, 23, e2025SW004398. https://doi.org/10.102/2025SW004398

Speaker: Loredana Perrone (Istituto Nazionale di Geofisica e Vulcanologia)

THERION_Perrone1.pdf

12:00 Physics–Informed Machine Learning Modeling of Plasmaspheric Cold Electron Density

Abstract
Machine learning (ML) methods are increasingly utilized in space weather research; however, their performance is often limited by sparse observations during extreme events and a lack of physical constraints. Physics-based models, on the other hand, rely on empirical parameterizations and simplifying assumptions that can limit their
performance. Bridging these two approaches offers a pathway toward more reliable and generalizable models of the near-Earth plasma environment.
In this study, we investigate plasmaspheric electron density modeling using a physics-based model (VERB-CS) [1], a purely data-driven neural network model (PINE) [4], and three hybrid physics–machine learning configurations. The hybrid frameworks are driven by geomagnetic indices and their time history, together with spacecraft location and in situ electron density measurements. In each configuration, density outputs from the physics-based model are incorporated into the neural network at different stages of training, enabling varying degrees of physical guidance.
Model performance is evaluated against independent in situ observations [2] and global plasmaspheric imaging [3]. The hybrid approaches consistently outperform both the purely data-driven and purely physics-based models, achieving lower root-mean-square errors, reduced bias, and improved generalization across quiet and disturbed geomagnetic conditions. The results demonstrate that embedding physical information within machine learning frameworks enhances predictive capability, particularly during periods of elevated geomagnetic activity.

References
[1] Aseev, N., Shprits, Y., 2019. Reanalysis of ring current electron phase space densities using Van Allen Probe observations, convection model, and log-normal Kalman filter. Space weather 17, 619–638.
[2] Kletzing, C., Kurth, W., Acuna, M., MacDowall, R., Torbert, R., Averkamp, T., Bodet, D., Bounds, S., Chutter, M., Connerney, J., et al., 2013. The electric and magnetic field instrument suite and integrated science (EMFISIS) on RBSP. Space Science Reviews 179, 127–181.
[3] Sandel, B., Goldstein, J., Gallagher, D., Spasojevic, M., 2003. Extreme ultraviolet imager observations of the structure and dynamics of the plasmasphere. Magnetospheric imaging—The image prime mission, 25–46.
[4] Zhelavskaya, I.S., Shprits, Y.Y., Spasojević, M., 2017. Empirical modeling of the plasmasphere dynamics using neural networks. Journal of Geophysical Research: Space Physics 122, 11–227.

Speaker: Sadaf Shahsavani (GFZ Helmholtz Centre for Geosciences)

ESA_workshop_26.pptx

12:15 Mapping the Magnetosphere with Generative AI

Spacecraft do not sample Earth’s magnetosphere uniformly. Measurements cluster along orbital tracks and mission-targeted regions, making it hard to build global maps of plasma environments directly from data. In this talk, I will describe a fully data-driven approach for constructing a magnetospheric atlas from a decade of NASA Magnetospheric Multiscale (MMS) observations. We first compress the continuous time series into short states (2-minute windows) that capture both typical conditions and variability using summary statistics of key plasma moments and magnetic-field features. To counter spatial sampling imbalance, we discretize space with an adaptive 3D octree that refines only where observations are dense, and we use octree-informed weighting to reduce orbit-driven bias during learning.

On top of this representation, we train a discrete generative model (a vector-quantized variational autoencoder) to discover a small set of recurring plasma regimes without expert thresholds or hand labels. The resulting atlas maps where each regime occurs and, crucially, provides per-regime feature distributions. I will show how the learned regimes align with familiar structures (e.g., sheath, boundary layers, plasma sheet, lobes) and how their occupancy shifts between quiet and storm-time conditions, illustrating the map as both an interpretive tool and a foundation for automated labeling, anomaly detection, and generative expected conditions baselines

Speaker: Stefano Markidis (KTH Royal Institute of Technology)

ESAmarkidis_presentation.pdf

12:30 Perspective for Magnetospheric Modelling with Multi-Moment Fluid Codes

Operational space weather forecasting requires magnetospheric models that balance physical accuracy with computational feasibility. Magnetohydrodynamic (MHD) simulations are affordable in terms of computational cost but miss critical kinetic effects, while hybrid/kinetic models provide accuracy at prohibitive costs.
We intend to pursue a novel approach based on the existing muphy2 framework. muphy2 is a multi-physics framework including Vlasov, ten-moment and five-moment two-fluid models. Simulations can be run either stand-alone or in coupled mode with threshold criteria determining model switching.
In this talk we present five- and ten-moment simulations of processes of relevance for terrestrial magnetospheres benchmarked against their Vlasov counterpart: Magnetic reconection, turbulence, and Kelvin-Helmholtz instability. We comment on the role of the heat flux closure in ten-moment models, as a way of introducing reduced kinetic effects in large-scale fluid simulations. Our ambition is to develop a coupled five-moment/ten-moment model for the global magnetosphere.

Speaker: Simon Lautenbach (Ruhr University Bochum)

Talk_Magnetospheric_Modellig-3.pdf

12:45 → 13:00 Workshop

12:45 Introducing the ESA Space HPC

The ESA Space HPC is the European Space Agency’s new high performance computing platform designed to accelerate research, development, and innovation in the space sector. With a flexible architecture including three different partitions, the system enables complex simulations, large scale data analysis, and advanced modelling with high speed and precision. It offers scalable computational resources and straightforward access through a lightweight and transparent allocation process.

Speaker: Dr Neva Besker (ESA -ESRIN)

IntroducingSpaceHPC-ESOC2026-NB2.pptx

13:00 → 14:00 Lunch

14:00 → 15:30 Community Capabilities & Radiation Environment

14:00 Italian Space Weather Modelling Capabilities: A Comprehensive Overview from the SWICo Community

We present a comprehensive overview of Italy’s national capabilities in Space Weather modelling, compiled by the Space Weather Italian Community (SWICo) that consolidates contributions from universities, research institutes, agencies, and private partners across the country. It spans the full Sun–Earth chain, covering modelling efforts from solar activity and flare forecasting to heliospheric propagation, magnetospheric dynamics, ionospheric specification, and geomagnetic and technological impact assessment.

The compilation includes more than fifty models, tools, and operational services, ranging from physics‑based simulations and data‑driven frameworks to machine‑learning systems and real‑time monitoring pipelines. These contributions describe methodologies, maturity levels, domains of applicability, and operational readiness, illustrating the breadth and depth of Italian expertise in both research and operational space‑weather services.

The document provides an authoritative snapshot of Italy’s current modelling landscape and its potential for collaboration within future European Space Weather and Heliophysics initiatives.

Speaker: Prof. Mirko Piersanti (University of L'Aquila, Italy)

SWICo_ESA_ESOC.pptx

14:15 Space Weather at University of L'Aquila

The Space Physics Group at the University of L’Aquila conducts advanced research on Sun–Earth coupling processes within the magnetosphere–ionosphere–thermosphere (MIT) system, with particular attention to phenomena that directly affect technological infrastructures and space-based services.

A core research line focuses on the formation and evolution of high-latitude ionospheric irregularities—structured plasma disturbances generated by solar-wind–magnetosphere interactions that significantly degrade GNSS signal propagation and positioning accuracy. In parallel, the group has developed the MA.I.GIC. model, a quantitative framework for assessing geomagnetically induced currents (GICs) during geomagnetic storms. A distinctive feature of this model is its capability to separate magnetospheric and ionospheric current contributions, thereby enhancing risk evaluation and mitigation strategies for power-grid systems.

The group also advances physics-informed neural-network methodologies to model thermospheric mass-density variability under disturbed geomagnetic conditions. This approach addresses a critical operational challenge: improving satellite drag forecasting during periods of elevated solar activity.

In addition, magnetospheric plasma dynamics are investigated through remote sensing of field-line resonances using ground-based magnetometer arrays. This technique enables reconstruction of plasmaspheric mass-density variations and supports the assessment of their impact on radiation-belt dynamics and GNSS system performance.

Collectively, these research activities provide an integrated and physically consistent framework for understanding and modelling geospace variability, contributing both to fundamental scientific progress and to the development of robust space-weather operational capabilities.

Speaker: Prof. Giulia D'Angelo (University of L'Aquila, Italy)

UnivAQ_ESA_ESOC_GD.pptx

14:30 Solar Cast: a suite of forecasting tool for the solar activity and the inner heliosphere

We will present our project Solar Cast@CEA that regroups several forecasting tools developed over many years by our team at CEA under CNES, ERC, french funding agency (ANR) and ESA fundings to be able to anticipate the solar activity and its influence of the inner heliosphere and our technological society.
Solar Cast is at present including three tools: Solar Predict, Wind Predict and Flare Predict, the first two being already deployed and maintained on ESA SWE web site and Flare Predict is anticipated to do so in the next 3 years. Solar predict forecasts using advanced 4-D var data assimilation methods aand a solar mean field dynamo, the solar cycle phase and amplitude with a 3-yr horizon window and is part of S-ESC, Wind-Predict is forecasting the state of the solar corona and wind via data-driven 3-D mHD simulation and is part of the VSWMC and H-ESC and finally Flare-Predict is based on a self-critical sandpile model to forecast the intense (> C8) flares. We will discuss their advantages and how we foresee their next development.

Speaker: Allan Sacha BRUN (Dept of Astrophysics, CEA-Paris Saclay)

SolarCast_ASBrun_ESAWorkshop_April2026.pptx

14:45 Modelling of Cosmic Radiation at the Institute of Aerospace Medicine of the German Aerospace Center

The Radiation Biology Department at the Institute of Aerospace Medicine of the German Aerospace Center has extensive experience in modelling cosmic radiation, its transport trough space craft shielding and planetary atmospheres and the conversion to quantities applicable to experimental validation and radiation protection. The presentation comprises an overview over past and current activities, including modelling the radiation field from galactic and solar cosmic radiation on Moon and Mars, on the ISS, in interplanetary space and at aviation altitudes, and validations of the results.

Speaker: Dr Daniel Matthiä (Institute of Aerospace Medicine, German Aerospace Center (DLR), Linder Höhe, Köln, Germany)

2026-04-08-Matthiä_Modelling of Cosmic Radiation at the Institute of Aerospace Medicine of the German Aerospace Center_ESA - Space Weather and Heliophysics Modelling Workshop.pptx

15:00 FLAG: Operational Orchestration of Integrated Sun–Earth Modelling

Building on the Horizon 2020 projects PROGRESS (PRediction of Geospace Radiation Environment and Solar wind parameterS) and PAGER (Prediction of Adverse effects of Geomagnetic Storms and Energetic Radiation), the ESA-funded FLAG (Forecasts and Long-term probabilistic data Assimilative prediction of the effects of Geomagnetic storms) project advances the integration and operational deployment of space weather forecasting capabilities. While earlier initiatives established ensemble-based predictions of the radiation environment and geomagnetic disturbances, FLAG focuses on reorganizing and coupling these modelling components into a unified and operationally robust framework.
At the core of FLAG lies the orchestration architecture, which coordinates numerical models describing solar wind conditions, geomagnetic indices, plasmasphere dynamics, radiation belts, ring current processes, ionospheric parameters, and spacecraft charging risks. The system integrates heterogeneous modelling approaches, including machine learning, data assimilation, and physics-based simulations, through standardized data interfaces and controlled execution logic. Each component retains operational independence while remaining interoperable within the system architecture.
The orchestration layer manages automated data ingestion, ensemble handling, execution sequencing, and structured output exchange, ensuring continuity of service even under degraded upstream data conditions. By enabling modular yet coherent integration of multiple forecasting components, FLAG strengthens the transition from advanced research models to coordinated operational space weather services within the European landscape.

Speaker: Kuan-Yu Ke

Space Weather and Heliophysics Modelling Workshop.pptx

15:15 RECENT DEVELOPMENTS IN THE VERSATILE ELECTRON RADIATION BELT (VERB) CODE AND DATA ASSIMILATION APPLICATIONS

Accurate specification and forecasting of Earth’s radiation belt electron environment remain critical for understanding radiation belt dynamics and mitigating space weather hazards to spacecraft. The Versatile Electron Radiation Belt (VERB) code has become a widely used physics-based model for simulating the evolution of relativistic electrons through radial diffusion and wave–particle interactions. In recent years, substantial advances have been made to enhance the physical realism, numerical performance, and data assimilation capability of the VERB framework.

In this work, we present an overview of recent developments in the VERB code, including improvements to diffusion coefficient parameterizations and optimization of the numerical solver for increased computational efficiency. Particular emphasis is placed on the integration of modern data assimilation techniques within VERB. We describe the implementation of ensemble- based and Kalman filter–type approaches to incorporate in situ observations from missions such as Van Allen Probes, GOES, ARASE and other relevant datasets. These upgrades enable more accurate reconstruction of phase space density and improved nowcasting and forecasting capability.
Case and long-term studies are presented demonstrating the impact of assimilation on reproducing observed radiation belt dynamics during geomagnetically active periods. Quantitative validation against independent measurements shows that the updated VERB framework signif icantly reduces model–data discrepancies and improves predictive skill relative to stand-alone simulations. Ongoing work focuses on further coupling with global magnetospheric models, machine-learning-assisted parameter estimation, and real-time applications for space weather operations.

Speaker: Yuri Yevgenyevich Shprits (GFZ, Potsdam, UCLA)

15:30 → 16:45 Workshop

15:30 Wrap up of 2nd day and team building activities

16:00 Visit of the ESOC site

16:45 → 17:30 Business Meetings with ESA

These are sessions available for the participants to have one-on-one conversations with the ESA Space Weather Office.

19:00 → 22:00 Group diner

Thursday 9 April

09:00 → 09:30 Badge Collection

09:30 → 10:00 Welcome coffee

10:00 → 11:15 The Ionosphere

10:00 NN modelling of the topside ionosphere based on four solar cycles of measurements

The Earth’s ionosphere affects the propagation of signals from the Global Navigation Satellite Systems (GNSS). The part of the ionosphere above the F2-layer peak, known as the topside ionosphere, contains a major portion of the total electron content and is therefore crucial for both scientific and practical applications. One of the major challenges for modeling the topside ionosphere has been the sparse data coverage, primarily limited to in-situ observations along satellite orbits that are relatively fixed in time and do not provide comprehensive three-dimensional coverage. Over the past two decades, a large number of radio occultation (RO) electron density profiles have become available, offering a valuable data source for enhancing our modeling capabilities. In this study, we use RO, in-situ, and digisonde data, along with historical observations from topside sounder missions, to develop an updated version of the Neural Network model of Electron density in the Topside ionosphere (NET). The updated model is based on the data set spanning approximately 4 solar cycles and therefore covers a wide range of solar activity and geomagnetic conditions. We validate the model using data from several prominent ionospheric missions and demonstrate that it generalizes well to unseen observations, with approximately 90% of the predictions falling within a factor of two of the measurements. We analyze the scientific capabilities of the model during geomagnetic storms, by investigating the behavior of the topside ionosphere, particularly the storm-time enhancements of the F2-peak height (hmF2) at high latitudes and their equatorward propagation as depicted by the model. Additionally, we discuss the operational capabilities of the NET model, including its real-time operation within the European FLAG project.

Speaker: Artem Smirnov (GFZ Helmholtz Centre for Geosciences, Potsdam, Germany)

SWAHM.pptx

10:15 Forecasting of foF2 over Europe

EUROMAP (Mikhailov and Perrone, Radio Science, 2014) is an an empirical forecasting model designed to predict foF2 over the European region. The system is based on local prediction models developed for individual European ionospheric stations, enabling continuous monitoring of foF2 across the entire continent.
The model is driven by geomagnetic and solar activity parameters, including the 3-hour ap index (converted into ap(τ)), the effective ionospheric T index, and available hourly foF2 observations, both historical and real-time. For disturbed conditions (ap(τ) > 30), local regression storm models are applied to describe strong negative ionospheric disturbances, while for quieter conditions (ap(τ) ≤ 30), the system relies on 28-day training models to reproduce foF2 variability. The T index is used to define the background ionospheric level.
The modelling domain extends from 40.0°W to 100.0°E in longitude and from 20.0°N to 80.0°N in latitude, covering the European region. The system generates foF2 maps on a 1° × 5° spatial grid. The primary controlling parameters are the solar F10.7 flux and the geomagnetic 3-hour ap index; notably, the model can operate even in the absence of current foF2 observations.
EUROMAP provides forecasts from 1 to 24 hours ahead, producing 24 maps every hour. The product is fully developed and ready for operational implementation

Speaker: Loredana Perrone (Istituto Nazionale di Geofisica e Vulcanologia)

foF2_Forecasting_Perrone2.pdf

10:30 Regional Adaptive and Assimilative 3D Ionospheric Modeling over Italy

A regional three-dimensional (3D) ionospheric model developed by the INGV Upper Atmosphere Physics and Radiopropagation Unit, has been developed for real-time monitoring over the Italian territory. The model combines a climatological background with real-time data ingestion from multiple ionosondes to reconstruct high-fidelity electron density profiles. Based on the Advanced Ionospheric Profiler (AIP) technique, the system adaptively corrects empirical parameters by minimizing the root-mean-square deviation between modeled and observed plasma frequency profiles. Testing conducted during both geomagnetically quiet and disturbed periods using the data of Rome, Gibilmanna, and San Vito dei Normanni stations has demonstrated high adaptability and significant accuracy improvements over standard climatological models. The modeling domain covers the Italian region, spanning approximately 6° to 19° longitude and 36° to 47.5° . Key outputs include real-time maps of critical frequencies foF2, 3D electron density distributions, and simulated ionogram traces suitable for nowcasting and HF propagation applications. Validations during moderate geomagnetic storms show that the model achieves better accuracy than the Solar Wind driven (SWIF) model, providing a reliable tool for estimating ionospheric conditions for technological and scientific purposes. The system's low processing latency and dynamic assimilation capabilities make it possible asset for regional space weather services

Speaker: Dr Carlo Scotto (INGV)

Regional-Ad-As-Mod-Ita-CS_DS2.pdf

Regional-Ad-As-Mod-Ita-CS_DS2.pptx

10:45 UCTOMO: Multi-Scale Estimation and Parameterization of the Ionospheric Radiofrequency Propagation Environment

Understanding and preparing for space weather events is critical for our technology innovations that operate in the near-earth space environment and/or rely on trans-ionospheric and skywave signal propagation. Safety-critical radio-based applications include satellite-based positioning, navigation, and timing (PNT), HF radar, and non-terrestrial communications networks. For example, Global Navigation Satellite System (GNSS) chipsets now deliver sub-decimetre-level precise positioning for billions of PNT devices worldwide, with rapidly growing downstream markets for autonomous location-based services driving ever more rigorous safety and integrity standards. With such technologies highly vulnerable to natural and human-made interference, there is urgent need for assured space domains which identify, classify and geo-reference system threats in real-time. Here we present multi-scale adaptive models for ionospheric estimation and parameterization via data-driven models.

Fundamental to this work is an ionospheric data assimilation framework that specifies three-dimensional temporally evolving electron density distributions over regional and global spatial scales. Input observations include integrated total electron content (TEC) for multi-frequency GNSS signal combinations from ground- and space-based receivers, electron density profiles (e.g. radio occultations and ionosondes), and in situ (e.g. Swarm plasma) observations. Unique to our model is the implementation of adaptive data driven bottomside (85-200 km altitude) precipitation profiles, HF hyperspectral sensing, and the capability to ingest thousands of low-earth orbiter PNT signals for improved vertical and horizontal resolution. Features such as the auroral oval, polar patches, and storm-enhanced density are resolved in a global unifying context with space weather benchmarks estimated to inform monitoring and mitigation approaches. Adapting methods to address NATO capabilities for integrity assurance, which includes capturing threats in under-sampled regions, we propagate ionosphere states and uncertainties into user domains. To assess model performance, predictions of key metrics were generated for a three-year period and compared with more than 200,000 truth observations. The framework was validated across all real-world performance requirements.

We provide examples of space weather events from our vast network of 120+ remote sensors, a national investment in over 40 new sophisticated optical and radio instruments across Alaska, Canada and Greenland. Combined with our modelling tools, this complement forms a national scale testbed for model development and verification, as well as basic research into the space environment, remote sensing techniques, space situational awareness and impacts on technology. Targeted investigations provide insight into the most impactful and meaningful observations from existing and emerging space- and ground-based observing systems, and identify gaps and opportunities in ionospheric modelling for space domain resilience.

Speaker: Susan Skone

ESA_slides_Skone_9Apr26.pptx

11:00 Constraining Ionospheric State from Ground-Based Auroral Observations: The TREx-ATM Framework for Space Weather Applications

The University of Calgary operates a suite of ground-based instruments that together provide a uniquely integrated view of the high-latitude ionosphere during space weather events. Multi-spectral auroral imagers and meridian spectrographs spanning 59°–71° geomagnetic latitude deliver high-cadence optical diagnostics of particle precipitation across multiple emission lines [Gillies et al., 2019; Liang et al., 2024]. Hyperspectral imaging riometers provide spectrally resolved cosmic noise absorption, offering independent, quantitative constraints on energetic particle deposition into the D-region. GNSS scintillation receivers distributed across the network characterize ionospheric plasma structuring and total electron content variability on timescales directly relevant to space weather impacts. Operating in concert, these instruments span precipitation, absorption, and plasma irregularities — delivering the multi-diagnostic observational context needed to constrain and validate physics-based space weather models.

The TREx Auroral Transport Model (TREx-ATM) [Liang et al., 2016, 2021, 2022; 2026] was purpose-built to exploit this multi-wavelength architecture. Compared to other existing electron transport codes ( (e.g., B3C and GLOW) that do not fully self-consistently evolve the ionospheric plasma state and/or have limitations in precipitation electron energies (typically no more than a few tens keV), TREx-ATM is a time-dependent model that self-consistently computes electron density, plasma temperature, and ionospheric conductivity profiles from first principles [Liang et al., 2022], and is extended to the relativistic energy range up to 10 MeV (Liang et al., 2026). The use of multiple emission lines simultaneously enables not only the derivation of precipitation mean energy and energy flux, but also independent constraint of the neutral O/N₂ ratio — a key source of error in competing inversion schemes [Liang et al., 2021 AGU]. Additionally, TREx-ATM incorporates coupled proton-hydrogen-electron transport, extending its inversion capacity to mixed precipitation regimes and enabling self-consistent conductance calculation for proton aurora [Liang et al., 2024, JGR]. A kinetic treatment of interhemispheric secondary electron transport provides further physical fidelity beyond two-stream approximation models.

Applied to the University of Calgary observations, this framework produces time-evolving, 2D maps of precipitation parameters and ionospheric conductances — quantities central to space weather specification and forecasting. These model-derived products are complemented by D-region absorption maps from the hyperspectral riometer network and TEC-based ionospheric products including tomographic reconstructions (e.g., UCTOMO), together providing a multi-layer characterization of ionospheric state from the D-region through the F-region. We discuss how this integrated observational and modeling capability can contribute to operational space weather efforts, including event-driven ionospheric specification and the potential for systematic precipitation characterization during geomagnetic activity.

Speaker: Emma Spanswick (University of Calgary)

ESOC_SW_ATM_April2026.pptx

11:15 → 12:45 Workshop

11:15 Doing Business with ESA (in a nutshell) - Understanding ESA Procurement

This short lecture will provide the most basic concepts of the ESA procurement process:

• To understand ESA's main operating principles and basic institutional
  set up
• To learn about the Agency's procurement principles and procedures
• To get to know the Agency's Procurement tools
• To learn how to sign up as a tenderer and access Invitations to
  Tender
• To be introduced to the Agency's SME Initiative activities.
• To write a good ITT proposal

12:15 Workshop Wrap-up