21 October 2022 in Journal of Geophysical Research: Space Physics
Many models of the near‐Earth's space environment (radiation belts, ionosphere, upper atmosphere, etc.) are driven by geomagnetic indices, representing the state of disturbance of the Earth's magnetosphere. Over the past decade, machine learning‐based methods for forecasting geomagnetic indices from near‐Earth solar wind parameters have become popular in the space weather community. These methods often prove to be very accurate and skilled. However, these approaches have the notable drawback of being effective in an operational context only for limited forecasting horizons (often up to a couple of hours ahead at best). In order to increase this prediction horizon, we introduce SERENADE, a novel deep learning‐based proof‐of‐concept model using images delivered by the Atmospheric Imaging Assembly instrument onboard the Solar Dynamics Observatory spacecraft to directly provide probabilistic forecasts of the daily maximum of the geomagnetic index Kp up to a few days ahead. We show in particular that SERENADE is able to capture information on the geomagnetic dynamics from solar imaging alone. In addition, despite it being a prototypical model, our model is more accurate in most situations than three empirical baseline models. However, the model still shows some strong limitations inherent to its structure and the used data set, which could be the focus of future works. This opens the way to a better mid‐to‐long term data‐driven magnetospheric modeling within space weather and geophysical pipelines.
22 June 2020 in Space Weather
One of the main goals when studying Space Weather is to characterize extreme events occurrences and related characteristics. To do so, dedicated statistical methods from the so‐called extreme value analysis (EVA) field have been developed. In this study we used Ca index, derived from aa, in order to characterize geoeffectiveness from the radiation belts point of view with a 150‐year‐long data set. The analysis performed in this study thus focuses on this newsworthy index to provide clues on the reliability of EVA methods. The first main result we present here is that the 1‐in‐10‐, 1‐in‐50‐, and 1‐in‐100‐year events, respectively, match Ca values of 100.39, 131.39, and 142.84 nT. Consequently, the only 1‐in‐100 event observed during the Space Era would be the “Halloween Storm” in 2003 that reached a Ca value of 147.6 nT. The second main result highlighted in this work is that performing the same analysis with shorter subsets (20 years long) can give significantly different results for two reasons. The first reason is that some short time periods do not display the same distribution of events as the full period. The second reason is that the choice of the correct threshold (when using a Peaks Over Threshold approach) is made difficult with a short data set and leads to inaccurate results. This is a strong result as for accurate estimation of the induced effects of extreme events in radiation belts, we may only rely on short flux data sets from one or another mission (mostly shorter than 20 years).
18 August 2023 in Space Weather
The H2020 SafeSpace project aims at the implementation of a space weather safety prototype, in particular to predict the deep charging hazard. The proposed service is built on a Sun‐to‐Earth chain of physical codes that propagates physical information and uncertainties in order to model the outer radiation belt dynamics. In this paper, we present the inner magnetosphere section of the SafeSpace pipeline that relies on solar wind driven and hourly updated models that describe the trapped electron environment (VLF waves, cold plasma and seed population densities), as well as the physical processes to which the trapped electrons are subjected to, such as radial diffusion and wave particle interactions. Then, this physical configuration is poured into the Salammbô‐EnKF model, a data assimilation radiation belt model which provides a global forecast of the densities across the radiation belts. We have compared the forecasting performance of this new modeling pipeline to a reference model during the St. Patrick's Day storm in 2015. We show that the new SafeSpace implementation shows closer results to the observations in addition to a better forecast within the prediction horizon.
24 December 2021 in Journal of Space Weather and Space Climate
The Ca index is a time-integrated geomagnetic index that correlates well with the dynamics of high-energy electron fluxes in the outer radiation belts. Therefore, Ca can be used as an indicator for the state of filling of the radiation belts for those electrons. Ca also has the advantage of being a ground-based measurement with extensive historical records. In this work, we propose a data-driven model to forecast Ca up to 24 h in advance from near-Earth solar wind parameters. Our model relies mainly on a recurrent neural network architecture called Long Short Term Memory that has shown good performances in forecasting other geomagnetic indices in previous papers. Most implementation choices in this study were arbitrated from the point of view of a space system operator, including the data selection and split, the definition of a binary classification threshold, and the evaluation methodology. We evaluate our model (against a linear baseline) using both classical and novel (in the space weather field) measures. In particular, we use the Temporal Distortion Mix (TDM) to assess the propensity of two time series to exhibit time lags. We also evaluate the ability of our model to detect storm onsets during quiet periods. It is shown that our model has high overall accuracy, with evaluation measures deteriorating in a smooth and slow trend over time. However, using the TDM and binary classification forecast evaluation metrics, we show that the forecasts lose some of their usefulness in an operational context even for time horizons shorter than 6 h. This behaviour was not observable when evaluating the model only with metrics such as the root-mean-square error or the Pearson linear correlation. Considering the physics of the problem, this result is not surprising and suggests that the use of more spatially remote data (such as solar imaging) could improve space weather forecasts.
26 May 2025 in Journal of Space Weather and Space Climate
Cataloging past space weather events, such as Interplanetary Coronal Mass Ejections (ICMEs) and Stream Interaction Regions (SIRs), is essential for both scientific research and operational applications. Firstly, it enables a comprehensive statistical analysis of their intrinsic physical and climatological properties. This, in turn, helps improve our understanding of their impact on the near-Earth space environment, including effects on human activities. It also enables the definition of event-driven space weather scenarii. Such studies benefit directly from the rapid and reproducible expansion of these catalogs, made possible by automatic event detection from in-situ time series data. Previous studies revealed the efficiency of deep-learning based methods for this task over traditional threshold-based techniques. Nevertheless, these methods have never been designed to simultaneously identify multiple event types. In this paper, we present a novel method for the multi-class automatic detection of ICMEs and SIRs. Our approach is inspired by the You Only Look Once (YOLO) family of algorithms, widely used in object detection. It works by directly identifying candidate time intervals, providing quick visual indicators of event occurrences that can be readily used by human observers in an operational space weather context. Thanks to its simple architecture, our method is easily implementable in data visualization tools and could be easily extended to additional event types. Tested on OMNI data between 1995 and 2024, the method detects at its best 644 out of 840 existing ICMEs and 1110 out of 1237 existing SIRs. 174 out of the 876 identified ICMEs and 192 out of the 1358 identified SIRs are actual False Positives for a maximal F1-score of 0.784 for ICMEs and 0.878 for SIRs. Our model performs slightly better at detecting events than other existing deep-learning based methods while achieving comparable results when estimating the events beginning and ending times. A detailed analysis of our method output shows that the great majority of detection errors are short events with a weak in-situ signature, which are expected to be among the least geoeffective event or misclassified events. We also show that the confusion made between ICMEs and SIRs is also shown to be comparable to the one actually made by human observers when manually establishing their catalogs.
4 August 2023 in Annales Geophysicae
In this paper, we present the SafeSpace prototype for a safety warning system, dedicated to the electron radiation-belt-induced internal charging hazard aboard spacecraft. The space weather tool relies on a synergy of physical models associated in a chain that covers the whole Sun–interplanetary-space–Earth's inner magnetosphere medium. With the propagation of uncertainties along the modelling pipeline, the safety prototype provides a global nowcast and forecast (within a 4 d lead time) of the electron radiation belt dynamic as well as tailored indicators for space industry operators. They are meant to inform the users about the severity of the electron space environment via a three-coloured alarm system, which sorts the index intensity according to a representative historical distribution of in situ data. The system was tested during the challenging 2015 St Patrick's Day storm in order to assess its performance. It showed overall good nowcasting and forecasting capabilities due to its broad physics-driven pipeline.