Prediction Of Coronal Mass Ejections Research Articles

An Algorithm for the Determination of Coronal Mass Ejection Kinematic Parameters Based on Machine Learning

Coronal mass ejections (CMEs) constitute the major source of severe space weather events, with the potential to cause enormous damage to humans and spacecraft in space. It is becoming increasingly important to detect and track CMEs, since there are more and more space activities and facilities. We have developed a new algorithm to automatically derive a CME’s kinematic parameters based on machine learning. Our method consists of three steps: recognition, tracking, and the determination of parameters. First, we train a convolutional neural network to classify images from Solar and Heliospheric Observatory Large Angle Spectrometric Coronagraph observations into two categories, containing CME(s) or not. Next, we apply the principal component analysis algorithm and Otsu’s method to acquire binary-labeled CME regions. Then, we employ the track-match algorithm to track a CME’s motion in time-series images and finally determine the CME’s kinematic parameters, e.g., velocity, angular width, and central position angle. The results of four typical CME events with different morphological characteristics are presented and compared with a manual CME catalog and several automatic CME catalogs. Our algorithm shows some advantages in the recognition of CME structure and the accuracy of the kinematic parameters. This algorithm can be helpful for real-time CME warnings and predictions. In the future, this algorithm is capable of being applied to CME initialization in magnetohydrodynamic simulations to study the propagation characteristics of real CME events and to provide more efficient predictions of CMEs’ geoeffectiveness.

Physics-driven Machine Learning for the Prediction of Coronal Mass Ejections’ Travel Times

Coronal Mass Ejections (CMEs) correspond to dramatic expulsions of plasma and magnetic field from the solar corona into the heliosphere. CMEs are scientifically relevant because they are involved in the physical mechanisms characterizing the active Sun. However, more recently, CMEs have attracted attention for their impact on space weather, as they are correlated to geomagnetic storms and may induce the generation of solar energetic particle streams. In this space weather framework, the present paper introduces a physics-driven artificial intelligence (AI) approach to the prediction of CMEs’ travel time, in which the deterministic drag-based model is exploited to improve the training phase of a cascade of two neural networks fed with both remote sensing and in situ data. This study shows that the use of physical information in the AI architecture significantly improves both the accuracy and the robustness of the travel time prediction.

Improving the Arrival Time Estimates of Coronal Mass Ejections by Using Magnetohydrodynamic Ensemble Modeling, Heliospheric Imager Data, and Machine Learning

The arrival time prediction of coronal mass ejections (CMEs) is an area of active research. Many methods with varying levels of complexity have been developed to predict CME arrival. However, the mean absolute error (MAE) of predictions remains above 12 hr, even with the increasing complexity of methods. In this work we develop a new method for CME arrival time prediction that uses magnetohydrodynamic simulations involving data-constrained flux-rope-based CMEs, which are introduced in a data-driven solar wind background. We found that for six CMEs studied in this work the MAE in arrival time was ∼8 hr. We further improved our arrival time predictions by using ensemble modeling and comparing the ensemble solutions with STEREO-A and STEREO-B heliospheric imager data. This was done by using our simulations to create synthetic J-maps. A machine-learning (ML) method called the lasso regression was used for this comparison. Using this approach, we could reduce the MAE to ∼4 hr. Another ML method based on the neural networks (NNs) made it possible to reduce the MAE to ∼5 hr for the cases when HI data from both STEREO-A and STEREO-B were available. NNs are capable of providing similar MAE when only the STEREO-A data are used. Our methods also resulted in very encouraging values of standard deviation (precision) of arrival time. The methods discussed in this paper demonstrate significant improvements in the CME arrival time predictions. Our work highlights the importance of using ML techniques in combination with data-constrained magnetohydrodynamic modeling to improve space weather predictions.

The characteristics of flare- and CME-productive solar active regions

Solar flares and coronal mass ejections (CMEs) cause immediate and adverse effects on the interplanetary space and geospace. In an era of space-based technical civilization, the deeper understanding of the mechanisms that produce them and the construction of efficient prediction schemes are of paramount importance. The source regions of flares and CMEs exhibit some common morphological characteristics, such as δ-spots, filaments and sigmoids, which are associated with strongly sheared magnetic polarity inversion lines, indicative of the complex magnetic configurations that store huge amounts of free magnetic energy and helicity. The challenge is to transform this empirical knowledge into parameters/predictors that can help us distinguish efficiently between quiet, flare-, and CME-productive (eruptive) active regions. This paper reviews these efforts to parameterize the characteristics of eruptive active regions as well as the importance of transforming new knowledge into more efficient predictors and including new types of data. Magnetic properties of active regions were first introduced when systematic ground-based observations of the photospheric magnetic field became possible and the relevant research was boosted by the provision of near real time, uninterrupted, high-quality observations from space, which allowed the study of large, statistically significant samples. Nonetheless, flare and CME prediction still faces a number of challenges. The magnetic field information is still constrained at the photospheric level and accessed only from one vantage point of observation, thus there is always need for better predictors; the dynamic behavior of active regions is still not fully incorporated into predictions; the inherent stochasticity of flares and CMEs renders their prediction probabilistic, thus benchmark sets are necessary to optimize and validate predictions. To meet these challenges, researchers have put forward new magnetic properties, which describe different aspects of magnetic energy storage mechanisms in active regions and offer the opportunity of parametric studies for over an entire solar cycle. This inventory of features/predictors is now expanded to include information from flow fields, transition region and coronal spectroscopy, data-driven modeling of the coronal magnetic field, as well as parameterizations of dynamic effects from time series. Further work towards these directions may help alleviate the current limitations in observing the magnetic field of higher atmospheric layers. In this task, fundamental and operational research converge, with promising results which could stimulate the development of new missions and lay the ground for future exploratory studies, also profiting from and utilizing the long anticipated observations of the new generation of instruments.

Joint Geoeffectiveness and Arrival Time Prediction of CMEs by a Unified Deep Learning Framework

Fast and accurate prediction of the geoeffectiveness of coronal mass ejections (CMEs) and the arrival time of the geoeffective CMEs is urgent, to reduce the harm caused by CMEs. In this paper, we present a new deep learning framework based on time series of satellites’ optical observations that can give both the geoeffectiveness and the arrival time prediction of the CME events. It is the first time combining these two demands in a unified deep learning framework with no requirement of manually feature selection and get results immediately. The only input of the deep learning framework is the time series images from synchronized solar white-light and EUV observations. Our framework first uses the deep residual network embedded with the attention mechanism to extract feature maps for each observation image, then fuses the feature map of each image by the feature map fusion module and determines the geoeffectiveness of CME events. For the geoeffective CME events, we further predict its arrival time by the deep residual regression network based on group convolution. In order to train and evaluate our proposed framework, we collect 2400 partial-/full-halo CME events and its corresponding images from 1996 to 2018. The F1 score and Accuracy of the geoeffectiveness prediction can reach 0.270% and 75.1%, respectively, and the mean absolute error of the arrival time prediction is only 5.8 h, which are both significantly better than well-known deep learning methods and can be comparable to, or even better than, the best performance of traditional methods.

Large non-radial propagation of a coronal mass ejection on 2011 January 24

Understanding the deflection of coronal mass ejections (CMEs) is of great interest to the space weather community because of their implications for improving the prediction of CME. This paper aims to shed light into the effects of the coronal magnetic field environment on CME trajectories. We analyze the influence of the magnetic environment on the early development of a particular CME event. On 2011 January 24 an eruptive filament was ejected in association with a CME that suffered a large deflection from its source region and expected trajectory. We characterize the 3D evolution of the prominence material using the tie-pointing/triangulation reconstruction technique on EUV and white-light images. To estimate the coordinates in 3D space of the apex of the CME we use a forward-modeling technique that reproduces the large-scale structure of the flux rope-like CME, the Graduated Cylindrical Shell model. We found that the deflection amounts to 42° in latitude and 20° in longitude and that most of it occurs at altitudes below 4R⊙. Moreover, we found a non-negligible deflection at higher altitudes. Combining images of different wavelengths with the extrapolated magnetic field obtained from a potential field source surface model we confirm the presence of two magnetic structures near the erupting event. The magnetic field environment suggests that field lines from the southern coronal hole act as a magnetic wall that produces the large latitudinal deflection; while a nearby pseudostreamer and a northward extension of the southern coronal hole may be responsible for the eastward deflection of the CME.

Statistical Relation between Solar Flares and Coronal Mass Ejections with Respect to Sigmoidal Structures in Active Regions

Abstract Statistical dependences among features of coronal mass ejections (CMEs), solar flares, and sigmoidal structures in soft-X-ray images were investigated. We applied analysis methods to all the features in the same way in order to investigate the reproducibility of the correlations among them, which may be found from previous statistical studies. Samples of 211 M-class and X-class flares, observed between 2006 and 2015 by the Hinode/X-ray telescope, Solar and Heliospheric Observatory/Large Angle and Spectrometric Coronagraph, and Geostationary Operational Environmental Satellite, were examined statistically. Five kinds of analysis were performed: occurrence rate analysis, linear-correlation analysis, association analysis, the Kolmogorov–Smirnov test, and the Anderson–Darling test. These give three main results. First, the sigmoidal structure and long-duration events (LDEs) have a stronger dependence on CME occurrence than large X-ray-class events in on-disk events. Second, for the limb events, a significant dependence exists between LDEs and CME occurrence, and between X-ray-class events and CME occurrence. Third, 32.4% of on-disk flare events have sigmoidal structure and are not accompanied by CMEs. However, the occurrence probability of CMEs without sigmoidal structure is very small, 8.8%, in on-disk events. While the first and second results are consistent with previous studies, we provide for the first time a difference between the on-disk and limb events. The third result, that non-sigmoidal regions produce fewer eruptive events, is also different from previous results. We suggest that sigmoidal structures in soft X-ray images will be a helpful feature for CME prediction in on-disk flare events.

Modeling observations of solar coronal mass ejections with heliospheric imagers verified with the Heliophysics System Observatory.

We present an advance toward accurately predicting the arrivals of coronal mass ejections (CMEs) at the terrestrial planets, including Earth. For the first time, we are able to assess a CME prediction model using data over two thirds of a solar cycle of observations with the Heliophysics System Observatory. We validate modeling results of 1337 CMEs observed with the Solar Terrestrial Relations Observatory (STEREO) heliospheric imagers (HI) (science data) from 8 years of observations by five in situ observing spacecraft. We use the self‐similar expansion model for CME fronts assuming 60° longitudinal width, constant speed, and constant propagation direction. With these assumptions we find that 23%–35% of all CMEs that were predicted to hit a certain spacecraft lead to clear in situ signatures, so that for one correct prediction, two to three false alarms would have been issued. In addition, we find that the prediction accuracy does not degrade with the HI longitudinal separation from Earth. Predicted arrival times are on average within 2.6 ± 16.6 h difference of the in situ arrival time, similar to analytical and numerical modeling, and a true skill statistic of 0.21. We also discuss various factors that may improve the accuracy of space weather forecasting using wide‐angle heliospheric imager observations. These results form a first‐order approximated baseline of the prediction accuracy that is possible with HI and other methods used for data by an operational space weather mission at the Sun‐Earth L5 point.

The UCSD kinematic IPS solar wind boundary and its use in the ENLIL 3‐D MHD prediction model

AbstractThe University of California, San Diego interplanetary scintillation (IPS) time‐dependent kinematic 3‐D reconstruction technique has been used and expanded upon for over a decade to provide predictions of heliospheric solar wind parameters. These parameters include global reconstructions of velocity, density, and (through potential field modeling and extrapolation upward from the solar surface) radial and tangential interplanetary magnetic fields. Time‐dependent results can be extracted at any solar distance within the reconstructed volume and are now being exploited as inner boundary values to drive the ENLIL 3‐D MHD model in near real time. The advantage of this coupled system is that it uses the more complete physics of 3‐D MHD modeling to provide an automatic prediction of coronal mass ejections and solar wind stream structures several days prior to their arrival at Earth without employing coronagraph observations. Here we explore, with several examples, the current differences between the IPS real‐time kinematic analyses and those from the ENLIL 3‐D MHD modeling using IPS‐derived real‐time boundaries. Future possibilities for this system include incorporating many different worldwide IPS stations as input to the remote sensing analysis using ENLIL as a kernel in the iterative 3‐D reconstructions.

A Space Weather mission concept: Observatories of the Solar Corona and Active Regions (OSCAR)

Coronal Mass Ejections (CMEs) and Corotating Interaction Regions (CIRs) are major sources of magnetic storms on Earth and are therefore considered to be the most dangerous space weather events. The Observatories of Solar Corona and Active Regions (OSCAR) mission is designed to identify the 3D structure of coronal loops and to study the trigger mechanisms of CMEs in solar Active Regions (ARs) as well as their evolution and propagation processes in the inner heliosphere. It also aims to provide monitoring and forecasting of geo-effective CMEs and CIRs. OSCAR would contribute to significant advancements in the field of solar physics, improvements of the current CME prediction models, and provide data for reliable space weather forecasting. These objectives are achieved by utilising two spacecraft with identical instrumentation, located at a heliocentric orbital distance of 1 AU from the Sun. The spacecraft will be separated by an angle of 68� to provide optimum stereoscopic view of the solar corona. We study the feasibility of such a mission and propose a preliminary design for OSCAR.

Evolution and Consequences of Interacting CMEs of 9 – 10 November 2012 Using STEREO/SECCHI and In Situ Observations

Understanding of the kinematic evolution of Coronal Mass Ejections (CMEs) in the heliosphere is important to estimate their arrival time at the Earth. It is found that kinematics of CMEs can change when they interact or collide with each other as they propagate in the heliosphere. In this paper, we analyze the collision and post-interaction characteristics of two Earth-directed CMEs, launched successively on 2012 November 9 and 10, using white light imaging observations from STEREO/SECCHI and in situ observations taken from WIND spacecraft. We tracked two density enhanced features associated with leading and trailing edge of November 9 CME and one density enhanced feature associated with leading edge of November 10 CME by constructing J-maps. We found that the leading edge of November 10 CME interacted with the trailing edge of November 9 CME. We also estimated the kinematics of these features of the CMEs and found a significant change in their dynamics after interaction. In in situ observations, we identified distinct structures associated with interacted CMEs and also noticed their heating and compression as signatures of CME-CME interaction. Our analysis shows an improvement in arrival time prediction of CMEs using their post-collision dynamics than using pre-collision dynamics. Estimating the true masses and speeds of these colliding CMEs, we investigated the nature of observed collision which is found to be close to perfectly inelastic. The investigation also places in perspective the geomagnetic consequences of the two CMEs and their interaction in terms of occurrence of geomagnetic storm and triggering of magnetospheric substorms.

Acceleration and deceleration of coronal mass ejections during propagation and interaction

A major challenge to the space weather forecasting community is accurate prediction of Coronal Mass Ejections (CMEs) induced Shock Arrival Time (SAT) at Earth's environment. In order to improve the current accuracy, one of the steps is to understand the physical processes of the acceleration and deceleration of a CME's propagation in the heliosphere. We employ our previous study of a three‐dimensional (3D) magnetohydrodynamic (MHD) simulation for the evolution of two interacting CMEs in a realistic ambient solar wind during the period 28–31 March 2001 event to illustrate these acceleration and deceleration processes. The forces which caused the acceleration and deceleration are analyzed in detail. The forces which caused the acceleration are the magnetic pressure term of Lorentz force and pressure gradient. On the other hand, the forces which caused the deceleration are aerodynamic drag, the Sun's gravity and the tension of magnetic field. In addition the momentum exchange between the solar wind and the moving CMEs can cause acceleration and deceleration of the CME which are now analyzed. In this specific CME event 28–31 March 2001 we have analyzed those forces which cause acceleration and deceleration of CME with and without interaction with another CME. It shows that there are significant momentum changes between these two interacting CMEs to cause the acceleration and deceleration.

Machine Leaning-Based Investigation of the Associations between CMEs and Filaments

In this work we study the association between eruptive filaments/prominences and coronal mass ejections (CMEs) using machine learning-based algorithms that analyse the solar data available between January 1996 and December 2001. The support vector machine (SVM) learning algorithm is used for the purpose of knowledge extraction from the association results. The aim is to identify patterns of associations that can be represented using SVM learning rules for the subsequent use in near real-time and reliable CME prediction systems. Timing and location data in the US National Geophysical Data Center (NGDC) filament catalogue and the Solar and Heliospheric Observatory/Large Angle and Spectrometric Coronagraph (SOHO/LASCO) CME catalogue are processed to associate filaments with CMEs. In the previous studies, which classified CMEs into gradual and impulsive CMEs, the associations were refined based on the CME speed and acceleration. Then the associated pairs were refined manually to increase the accuracy of the training dataset. In the current study, a data-mining system is created to process and associate filament and CME data, which are arranged in numerical training vectors. Then the data are fed to SVMs to extract the embedded knowledge and provide the learning rules that can have the potential, in the future, to provide automated predictions of CMEs. The features representing the event time (average of the start and end times), duration, type, and extent of the filaments are extracted from all the associated and not-associated filaments and converted to a numerical format that is suitable for SVM use. Several validation and verification methods are used on the extracted dataset to determine if CMEs can be predicted solely and efficiently based on the associated filaments. More than 14 000 experiments are carried out to optimise the SVM and determine the input features that provide the best performance.

A Numerical Study of the Breakout Model for Coronal Mass Ejection Initiation

A leading theory for the initiation of coronal mass ejections (CMEs) is the breakout model, in which magnetic reconnection above a filament channel is responsible for disrupting the coronal magnetic field. We present the first simulations of the complete breakout process including the initiation, the plasmoid formation and ejection, and the eventual relaxation of the coronal field to a more potential state. These simulations were performed using a new numerical code that solves the numerical cavitation problems that prevented previous simulations from calculating a complete ejection. Furthermore, the position of the outer boundary in the new simulations is increased out to 30 R☉, which enables determination of the full structure and dynamics of the ejected plasmoid. Our results show that the ejection occurs at a speed on the order of the coronal Alfven speed and hence that the breakout model can produce fast CMEs. Another key result is that the ejection speed is not sensitive to the refinement level of the grid used in the calculations, which implies that, at least for the numerical resistivity of these simulations, the speed is not sensitive to the Lundquist number. We also calculate, in detail, the helicity of the system and show that the helicity is well conserved during the breakout process. Most of the helicity is ejected from the Sun with the escaping plasmoid, but a significant fraction (of order 10%) remains in the corona. The implications of these results for observation and prediction of CMEs and eruptive flares is discussed.

CMEs Associated with Eruptive Prominences: How to Predict?

Solar prominences can be viewed as pre-eruptive states of coronal mass ejections (CMEs). Eruptive prominences are the phenomena most related to CMEs observed in the lower layers of the solar atmosphere. The most probable initial magnetic configuration of a CME is a flux rope consisting of twisted field lines which fills the whole volume of the dark cavity stretched in the corona along the photospheric polarity inversion line. Cold dense prominence matter accumulates in the lower parts of helical flux tubes, which serve as magnetic traps in the gravitation field. Coronal cavity is rather inconvenient feature for observation owing to reduced emission, so prominences and filaments are the best tracers of the flux ropes in the corona long before the beginning of eruption. Thus, the problem of the CME prediction can be reduced to the analysis of the filament equilibrium and estimation of the stability store. The height of a prominence (or a filament when observed against the disk) increases with its age and the death of a filament is usually an eruption which is followed by a CME. The filament height, then, can be a measure of its age and its readiness for eruption. In inverse-polarity models the equilibrium height of a filament is related to the value of the filament electric current. The stronger the electric current, the greater the height of the filament. However, the equilibrium and stability of a filament depend not only on its current but also on the characteristics of the external magnetic field. In order to estimate the probability of eruption, we should therefore compare the observed prominence height with a value characterizing the photospheric magnetic field. This value is the critical height, which can be found in the distribution of the magnetic field vertical gradient above the polarity inversion line. We had analyzed three dozens of filaments and found that eruptive prominences were near the limit of stability a few days before eruptions. We believe that the comparison of the real heights of prominences with the calculated critical heights could be a basis for predicting filament eruptions and following CMEs.To search for other articles by the author(s) go to: http://adsabs.harvard.edu/abstract_service.html

A measure from line‐of‐sight magnetograms for prediction of coronal mass ejections

From a sample of 17 vector magnetograms of 12 bipolar active regions we have recently found (1) that a measure of the overall nonpotentiality (the overall twist and shear in the magnetic field) of an active region is given by the strong shear length LSS, the length of the portion of the main neutral line on which the observed transverse fields is strong (>150 Guass (G)) and strongly sheared (shear angle >45°), and (2) that LSS is well correlated with the coronal mass ejection (CME) productivity of the active regions during the ±2‐day time window centered on the day of the magnetogram. In the present paper, from the same sample of 17 vector magnetograms, we show that there is a viable proxy for LSS that can be measured from a line‐of‐sight magnetogram. This proxy is the strong gradient length LSG, the length of the portion of the main neutral line on which the potential transverse field is strong (>150 G), and the gradient of the line‐of‐sight field is sufficiently steep (greater than ∼50 G/Mm). In our sample of active regions, LSG is statistically significantly correlated with LSS (correlation confidence level >95%), and LSG is as strongly correlated with active region CME productivity as is LSS (correlation confidence level ∼99.7%). Because LSG can be measured from line‐of‐sight magnetograms obtained from conventional magnetographs, such as the magnetograph mode of the Michelson Doppler Imager (MDI) on board the Solar and Heliospheric Observatory, it is a dependable substitute for LSS for use in operational CME forecasting. In addition, via measurement of LSG, the years‐long, nearly continuous sequence of 1.5‐hour cadence full disk line‐of‐sight magnetograms from MDI can be used to track the growth and decay of the large‐scale nonpotentiality in active regions and to examine the role of this evolution in active region CME productivity.

A prospective method for predicting coronal mass ejections from vector magnetograms

Two quantitative measures of the global nonpotentiality of predominantly bipolar active regions are explored for their usefulness as indicators of whether an active region is likely to produce coronal mass ejections (CMEs). The two measures, length of strong‐field strong‐shear main neutral line (Lss) and global net current (IN), are obtained directly from vector magnetograms of the regions. From measurements of Lss and IN in four active regions for which the global nonpotentiality and CME productivity are assessed independently of Lss and IN, it appears that Lss and IN are each a useful indicator of an active region's likely CME productivity. The results also suggest that Lss and IN may be more reliable predictors of the CME productivity of active regions than is the presence or absence of sigmoidal structure in coronal X‐ray images of the regions and that there might be threshold values of Lss and IN above which CME production is likely and below which CME production is unlikely. While these results need to be certified by an expansion of this pilot study to more active regions, they show the promise of Lss and IN as quantitative measures of global nonpotentiality for CME prediction.

Coronal and interplanetary magnetic field geometry: Streak lines

This paper discusses coronal and interplanetary magnetic field models, including static and dynamical aspects. The seminal work of Parker [1958] led to an Archimedian spiral field geometry in the interplanetary medium. The early coronal and interplanetary magnetic field models were static models developed to deal with the lack of magnetic field observations throughout most of the heliosphere. Since we are still in this deplorable state, with a particular lack of observations in the inner corona, these static models are still being run since they enable the field throughout the heliosphere to be estimated with only photospheric measurements. By now extending the static models into the temporal domain, a better understanding of the field line dynamics is just beginning to emerge. This understanding is needed to calculate cosmic ray properties, as well as other dynamic events, such as coronal mass ejections. Presently, a number of poorly understood aspects of the coronal field exist: the overall large‐scale field structure, the magnetic field winding angle, and rotational properties of the corona and photosphere. This last topic relates to changing magnetic field patterns. We provide a new paradigm involving “streak lines” for a better understanding of interplanetary field line behavior. The concept of streak lines was developed for understanding fluid dynamics. Use of this concept may lead to future developments and possible prediction of coronal mass ejections, helping us understand some of the causes of “space weather.”