arXiv subjects

Manolis K. Georgoulis

Publications and source records attributed to Manolis K. Georgoulis.

At least 37 records · Page 2Linked to original sources

arXiv2021

All-Clear Flare Prediction Using Interval-based Time Series Classifiers

An all-clear flare prediction is a type of solar flare forecasting that puts more emphasis on predicting non-flaring instances (often relatively small flares and flare quiet regions) with high precision while still maintaining valuable predictive results. While many flare prediction studies do not address this problem directly, all-clear predictions can be useful in operational context. However, in all-clear predictions, finding the right balance between avoiding false negatives (misses) and reducing the false positives (false alarms) is often challenging. Our study focuses on training and testing a set of interval-based time series classifiers named Time Series Forest (TSF). These classifiers will be used towards building an all-clear flare prediction system by utilizing multivariate time series data. Throughout this paper, we demonstrate our data collection, predictive model building and evaluation processes, and compare our time series classification models with baselines using our benchmark datasets. Our results show that time series classifiers provide better forecasting results in terms of skill scores, precision and recall metrics, and they can be further improved for more precise all-clear forecasts by tuning model hyperparameters.

cs.LG↗

arXiv2021

How to Train Your Flare Prediction Model: Revisiting Robust Sampling of Rare Events

We present a case study of solar flare forecasting by means of metadata feature time series, by treating it as a prominent class-imbalance and temporally coherent problem. Taking full advantage of pre-flare time series in solar active regions is made possible via the Space Weather Analytics for Solar Flares (SWAN-SF) benchmark dataset; a partitioned collection of multivariate time series of active region properties comprising 4075 regions and spanning over 9 years of the Solar Dynamics Observatory (SDO) period of operations. We showcase the general concept of temporal coherence triggered by the demand of continuity in time series forecasting and show that lack of proper understanding of this effect may spuriously enhance models' performance. We further address another well-known challenge in rare event prediction, namely, the class-imbalance issue. The SWAN-SF is an appropriate dataset for this, with a 60:1 imbalance ratio for GOES M- and X-class flares and a 800:1 for X-class flares against flare-quiet instances. We revisit the main remedies for these challenges and present several experiments to illustrate the exact impact that each of these remedies may have on performance. Moreover, we acknowledge that some basic data manipulation tasks such as data normalization and cross validation may also impact the performance -- we discuss these problems as well. In this framework we also review the primary advantages and disadvantages of using true skill statistic and Heidke skill score, as two widely used performance verification metrics for the flare forecasting task. In conclusion, we show and advocate for the benefits of time series vs. point-in-time forecasting, provided that the above challenges are measurably and quantitatively addressed.

astro-ph.SR↗

arXiv2021

A Readily Implemented Atmosphere Sustainability Constraint for Terrestrial Exoplanets Orbiting Magnetically Active Stars

With more than 4,300 confirmed exoplanets and counting, the next milestone in exoplanet research is to determine which of these newly found worlds could harbor life. Coronal Mass Ejections (CMEs), spawn by magnetically active, superflare-triggering dwarf stars, pose a direct threat to the habitability of terrestrial exoplanets as they can deprive them from their atmospheres. Here we develop a readily implementable atmosphere sustainability constraint for terrestrial exoplanets orbiting active dwarfs, relying on the magnetospheric compression caused by CME impacts. Our constraint focuses on a systems understanding of CMEs in our own heliosphere that, applying to a given exoplanet, requires as key input the observed bolometric energy of flares emitted by its host star. Application of our constraint to six famous exoplanets, (Kepler-438b, Proxima-Centauri b, and Trappist-1d, -1e, -1f and -1g), within or in the immediate proximity of their stellar host's habitable zones, showed that only for Kepler-438b might atmospheric sustainability against stellar CMEs be likely. This seems to align with some recent studies that, however, may require far more demanding computational resources and observational inputs. Our physically intuitive constraint can be readily and en masse applied, as is or generalized, to large-scale exoplanet surveys to detect planets that could be sieved for atmospheres and, perhaps, possible biosignatures at higher priority by current and future instrumentation.

astro-ph.EP↗

arXiv2020

A Comparison of Flare Forecasting Methods. IV. Evaluating Consecutive-Day Forecasting Patterns

A crucial challenge to successful flare prediction is forecasting periods that transition between "flare-quiet" and "flare-active". Building on earlier studies in this series (Barnes et al. 2016; Leka et al. 2019a,b) in which we describe methodology, details, and results of flare forecasting comparison efforts, we focus here on patterns of forecast outcomes (success and failure) over multi-day periods. A novel analysis is developed to evaluate forecasting success in the context of catching the first event of flare-active periods, and conversely, of correctly predicting declining flare activity. We demonstrate these evaluation methods graphically and quantitatively as they provide both quick comparative evaluations and options for detailed analysis. For the testing interval 2016-2017, we determine the relative frequency distribution of two-day dichotomous forecast outcomes for three different event histories (i.e., event/event, no-event/event and event/no-event), and use it to highlight performance differences between forecasting methods. A trend is identified across all forecasting methods that a high/low forecast probability on day-1 remains high/low on day-2 even though flaring activity is transitioning. For M-class and larger flares, we find that explicitly including persistence or prior flare history in computing forecasts helps to improve overall forecast performance. It is also found that using magnetic/modern data leads to improvement in catching the first-event/first-no-event transitions. Finally, 15% of major (i.e., M-class or above) flare days over the testing interval were effectively missed due to a lack of observations from instruments away from the Earth-Sun line.

astro-ph.SR↗

arXiv2019

Challenges with Extreme Class-Imbalance and Temporal Coherence: A Study on Solar Flare Data

In analyses of rare-events, regardless of the domain of application, class-imbalance issue is intrinsic. Although the challenges are known to data experts, their explicit impact on the analytic and the decisions made based on the findings are often overlooked. This is in particular prevalent in interdisciplinary research where the theoretical aspects are sometimes overshadowed by the challenges of the application. To show-case these undesirable impacts, we conduct a series of experiments on a recently created benchmark data, named Space Weather ANalytics for Solar Flares (SWAN-SF). This is a multivariate time series dataset of magnetic parameters of active regions. As a remedy for the imbalance issue, we study the impact of data manipulation (undersampling and oversampling) and model manipulation (using class weights). Furthermore, we bring to focus the auto-correlation of time series that is inherited from the use of sliding window for monitoring flares' history. Temporal coherence, as we call this phenomenon, invalidates the randomness assumption, thus impacting all sampling practices including different cross-validation techniques. We illustrate how failing to notice this concept could give an artificial boost in the forecast performance and result in misleading findings. Throughout this study we utilized Support Vector Machine as a classifier, and True Skill Statistics as a verification metric for comparison of experiments. We conclude our work by specifying the correct practice in each case, and we hope that this study could benefit researchers in other domains where time series of rare events are of interest.

cs.LG↗

arXiv2019

Which Photospheric Characteristics are Most Relevant to Active-Region Coronal Mass Ejections?

We investigate the relation between characteristics of coronal mass ejections and parameterizations of the eruptive capability of solar active regions widely used in solar flare prediction schemes. These parameters, some of which are explored for the first time, are properties related to topological features, namely, magnetic polarity inversion lines (MPILs) that indicate large amounts of stored non-potential (i.e. free) magnetic energy. We utilize the Space Weather Database of Notifications, Knowledge, Information (DONKI) and the Large Angle and Spectrometric Coronograph (LASCO) databases to find flare-associated coronal mass ejections and their kinematic characteristics while properties of MPILs are extracted from Helioseismic and Magnetic Imager (HMI) vector magnetic-field observations of active regions to extract the properties of source-region MPILs. The correlation between all properties and the characteristics of CMEs ranges from moderate to very strong. More significant correlations hold particularly for fast CMEs, which are most important in terms of adverse space-weather manifestations. Non-neutralized currents and the length of the main MPIL exhibit significantly stronger correlations than the rest of the properties. This finding supports a causal relationship between coronal mass ejections and non-neutralized electric currents in highly sheared, conspicuous MPILs. In addition, non-neutralized currents and MPIL length carry distinct, independent information as to the eruptive potential of active regions. The combined total amount of non-neutralized electric currents and the length of the main polarity inversion line, therefore, reflect more efficiently than other parameters the eruptive capacity of solar active regions and the CME kinematic characteristics stemming from these regions.

astro-ph.SR↗

arXiv2019

A Comparison of Flare Forecasting Methods. III. Systematic Behaviors of Operational Solar Flare Forecasting Systems

A workshop was recently held at Nagoya University (31 October - 02 November 2017), sponsored by the Center for International Collaborative Research, at the Institute for Space-Earth Environmental Research, Nagoya University, Japan, to quantitatively compare the performance of today's operational solar flare forecasting facilities. Building upon Paper I of this series (Barnes et al. 2016), in Paper II (Leka et al. 2019) we described the participating methods for this latest comparison effort, the evaluation methodology, and presented quantitative comparisons. In this paper we focus on the behavior and performance of the methods when evaluated in the context of broad implementation differences. Acknowledging the short testing interval available and the small number of methods available, we do find that forecast performance: 1) appears to improve by including persistence or prior flare activity, region evolution, and a human "forecaster in the loop"; 2) is hurt by restricting data to disk-center observations; 3) may benefit from long-term statistics, but mostly when then combined with modern data sources and statistical approaches. These trends are arguably weak and must be viewed with numerous caveats, as discussed both here and in Paper II. Following this present work, we present in Paper IV a novel analysis method to evaluate temporal patterns of forecasting errors of both types (i.e., misses and false alarms; Park et al. 2019). Hence, most importantly, with this series of papers we demonstrate the techniques for facilitating comparisons in the interest of establishing performance-positive methodologies.

astro-ph.SR↗

arXiv2019

A Comparison of Flare Forecasting Methods. II. Benchmarks, Metrics and Performance Results for Operational Solar Flare Forecasting Systems

Solar flares are extremely energetic phenomena in our Solar System. Their impulsive, often drastic radiative increases, in particular at short wavelengths, bring immediate impacts that motivate solar physics and space weather research to understand solar flares to the point of being able to forecast them. As data and algorithms improve dramatically, questions must be asked concerning how well the forecasting performs; crucially, we must ask how to rigorously measure performance in order to critically gauge any improvements. Building upon earlier-developed methodology (Barnes et al, 2016, Paper I), international representatives of regional warning centers and research facilities assembled in 2017 at the Institute for Space-Earth Environmental Research, Nagoya University, Japan to - for the first time - directly compare the performance of operational solar flare forecasting methods. Multiple quantitative evaluation metrics are employed, with focus and discussion on evaluation methodologies given the restrictions of operational forecasting. Numerous methods performed consistently above the "no skill" level, although which method scored top marks is decisively a function of flare event definition and the metric used; there was no single winner. Following in this paper series we ask why the performances differ by examining implementation details (Leka et al. 2019, Paper III), and then we present a novel analysis method to evaluate temporal patterns of forecasting errors in (Park et al. 2019, Paper IV). With these works, this team presents a well-defined and robust methodology for evaluating solar flare forecasting methods in both research and operational frameworks, and today's performance benchmarks against which improvements and new methods may be compared.

astro-ph.SR↗

arXiv2018

Photospheric Shear Flows in Solar Active Regions and Their Relation to Flare Occurrence

Solar active regions (ARs) that produce major flares typically exhibit strong plasma shear flows around photospheric magnetic polarity inversion lines (MPILs). It is therefore important to quantitatively measure such photospheric shear flows in ARs for a better understanding of their relation to flare occurrence. Photospheric flow fields were determined by applying the Differential Affine Velocity Estimator for Vector Magnetograms (DAVE4VM) method to a large data set of 2,548 co-aligned pairs of AR vector magnetograms with 12-min separation over the period 2012-2016. From each AR flow-field map, three shear-flow parameters were derived corresponding to the mean ( ), maximum (S_max) and integral (S_sum) shear-flow speeds along strong-gradient, strong-field MPIL segments. We calculated flaring rates within 24 hr as a function of each shear-flow parameter, and also investigated the relation between the parameters and the waiting time (τ) until the next major flare (class M1.0 or above) after the parameter observation. In general, it is found that the larger S_sum an AR has, the more likely it is for the AR to produce flares within 24 hr. It is also found that among ARs which produce major flares, if one has a larger value of S_sum then τ generally gets shorter. These results suggest that large ARs with widespread and/or strong shear flows along MPILs tend to not only be more flare productive, but also produce major flares within 24 hr or less.

astro-ph.SR↗

arXiv2018

Testing and Improving a Set of Morphological Predictors of Flaring Activity

Efficient prediction of solar flares relies on parameters that quantify the eruptive capability of solar active regions. Several such quantitative predictors have been proposed in the literature, inferred mostly from photospheric magnetograms and/or white-light observations. Two of them are the Ising energy and the sum of the total horizontal magnetic field gradient. The former has been developed from line-of-sight magnetograms, while the latter uses sunspot detections and characteristics, based on continuum images. Aiming to include these parameters in an automated prediction scheme, we test their applicability on regular photospheric magnetic field observations provided by the Helioseismic and Magnetic Imager (HMI) instrument onboard the Solar Dynamics Observatory (SDO). We test their efficiency as predictors of flaring activity on a representative sample of active regions and investigate possible modifications of these quantities. The Ising energy appears to be an efficient predictor, and the efficiency is even improved if it is modified to describe interacting magnetic partitions or sunspot umbrae. The sum of the horizontal magnetic field gradient appears to be slightly more promising than the three variations of the Ising energy we implement in this article. The new predictors are also compared with two very promising predictors: the effective connected magnetic field strength and the total unsigned non-neutralized current. Our analysis shows that the efficiency of morphological predictors depends on projection effects in a nontrivial way. All four new predictors are found useful for inclusion in an automated flare forecasting facility, such as the Flare Likelihood and Region Eruption Forecasting (FLARECAST), but their utility, among others, will ultimately be determined by the validation effort underway in the framework of the FLARECAST project.

astro-ph.SR↗

arXiv2018

Forecasting Solar Flares Using Magnetogram-based Predictors and Machine Learning

We propose a forecasting approach for solar flares based on data from Solar Cycle 24, taken by the Helioseismic and Magnetic Imager (HMI) on board the Solar Dynamics Observatory (SDO) mission. In particular, we use the Space-weather HMI Active Region Patches (SHARP) product that facilitates cut-out magnetograms of solar active regions (AR) in the Sun in near-real-time (NRT), taken over a five-year interval (2012 - 2016). Our approach utilizes a set of thirteen predictors, which are not included in the SHARP metadata, extracted from line-of-sight and vector photospheric magnetograms. We exploit several Machine Learning (ML) and Conventional Statistics techniques to predict flares of peak magnitude >M1 and >C1, within a 24 h forecast window. The ML methods used are multi-layer perceptrons (MLP), support vector machines (SVM) and random forests (RF). We conclude that random forests could be the prediction technique of choice for our sample, with the second best method being multi-layer perceptrons, subject to an entropy objective function. A Monte Carlo simulation showed that the best performing method gives accuracy ACC=0.93(0.00), true skill statistic TSS=0.74(0.02) and Heidke skill score HSS=0.49(0.01) for >M1 flare prediction with probability threshold 15% and ACC=0.84(0.00), TSS=0.60(0.01) and HSS=0.59(0.01) for >C1 flare prediction with probability threshold 35%.

astro-ph.SR↗

arXiv2017

Non-neutralized electric currents in solar active regions and flare productivity

We explore the association of non-neutralized currents with solar flare occurrence in a sizable sample of observations, aiming to show the potential of such currents in solar flare prediction. We use the regularly produced high quality vector magnetograms by the Helioseismic Magnetic Imager and more specifically the Space weather HMI Active Region Patches (SHARP). Through a newly established method, that incorporates detailed error analysis, we calculate the non-neutralized currents contained in active regions (AR). Two predictors are produced, namely the total and the maximum unsigned non-neutralized current. Both are tested in AR time-series and a representative sample of point-in-time observations during the interval 2012-2016. The average values of non-neutralized currents in flaring active regions are by more than an order of magnitude higher than in non-flaring regions and correlate very well with the corresponding flare index. The temporal evolution of these parameters appears to be connected to physical processes, such as flux emergence and/or magnetic polarity inversion line formation that are associated with increased solar flare activity. Using Bayesian inference of flaring probabilities, it is shown that the total unsigned non-neutralized current outperforms significantly the total unsigned magnetic flux and other well established current-related predictors. Thus, it shows good prospects for inclusion in an operational flare forecasting service. We plan to use the new predictor in the framework of the FLARECAST project along with other highly performing predictors.

astro-ph.SR↗

arXiv2016

Magnetic helicity estimations in models and observations of the solar magnetic field. Part I: Finite volume methods

Magnetic helicity is a conserved quantity of ideal magneto-hydrodynamics characterized by an inverse turbulent cascade. Accordingly, it is often invoked as one of the basic physical quantities driving the generation and structuring of magnetic fields in a variety of astrophysical and laboratory plasmas. We provide here the first systematic comparison of six existing methods for the estimation of the helicity of magnetic fields known in a finite volume. All such methods are reviewed, benchmarked, and compared with each other, and specifically tested for accuracy and sensitivity to errors. To that purpose, we consider four groups of numerical tests, ranging from solutions of the three-dimensional, force-free equilibrium, to magneto-hydrodynamical numerical simulations. Almost all methods are found to produce the same value of magnetic helicity within few percent in all tests. In the more solar-relevant and realistic of the tests employed here, the simulation of an eruptive flux rope, the spread in the computed values obtained by all but one method is only 3%, indicating the reliability and mutual consistency of such methods in appropriate parameter ranges. However, methods show differences in the sensitivity to numerical resolution and to errors in the solenoidal property of the input fields. In addition to finite volume methods, we also briefly discuss a method that estimates helicity from the field lines' twist, and one that exploits the field's value at one boundary and a coronal minimal connectivity instead of a pre-defined three-dimensional magnetic-field solution.

astro-ph.SR↗

arXiv2015

25 Years of Self-Organized Criticality: Numerical Detection Methods

The detection and characterization of self-organized criticality (SOC), in both real and simulated data, has undergone many significant revisions over the past 25 years. The explosive advances in the many numerical methods available for detecting, discriminating, and ultimately testing, SOC have played a critical role in developing our understanding of how systems experience and exhibit SOC. In this article, methods of detecting SOC are reviewed; from correlations to complexity to critical quantities. A description of the basic autocorrelation method leads into a detailed analysis of application-oriented methods developed in the last 25 years. In the second half of this manuscript space-based, time-based and spatial-temporal methods are reviewed and the prevalence of power laws in nature is described, with an emphasis on event detection and characterization. The search for numerical methods to clearly and unambiguously detect SOC in data often leads us outside the comfort zone of our own disciplines - the answers to these questions are often obtained by studying the advances made in other fields of study. In addition, numerical detection methods often provide the optimum link between simulations and experiments in scientific research. We seek to explore this boundary where the rubber meets the road, to review this expanding field of research of numerical detection of SOC systems over the past 25 years, and to iterate forwards so as to provide some foresight and guidance into developing breakthroughs in this subject over the next quarter of a century.

nlin.AO↗

arXiv2015

Analysing the effects of apodizing windows on local correlation tracking using Nirvana simulations of convection

We employ different shapes of apodizing windows in the local correlation tracking (LCT) routine to retrieve horizontal velocities using numerical simulations of convection. LCT was applied on a time sequence of temperature maps generated by the Nirvana code with four different apodizing windows, namely--Gaussian, Lorentzian, trapezoidal and triangular, with varying widths. In terms of correlations (between the LCT-retrieved and simulated flow field), the triangular and the trapezoidal perform the best and worst, respectively. On segregating the intrinsic velocities in the simulations on the basis of their magnitudes, we find that for all windows, a significantly higher correlation is obtained for the intermediate and high-velocity bins and only modest or weak values in the low-velocity bins. The differences between the LCT-retrieved and simulated flow fields were determined spatially which show large residuals at or close to the boundary of granules. The extent to which the horizontal flow vectors retrieved by LCT compare with the simulated values, depends entirely on the width of the central peak of the apodizing window for a given $σ$. Even though LCT suffers from a lack of spatial content as seen in simulations, its simplicity and speed could serve as a viable first-order tool to probe horizontal flows--one that is ideal for large data sets.

astro-ph.SR↗

arXiv2013

Interpreting Eruptive Behavior in NOAA AR 11158 via the Region's Magnetic Energy and Relative Helicity Budgets

In previous works we introduced a nonlinear force-free method that self-consistently calculates the instantaneous budgets of free magnetic energy and relative magnetic helicity in solar active regions (ARs). Calculation is expedient and practical, using only a single vector magnetogram per computation. We apply this method to a timeseries of 600 high-cadence vector magnetograms of the eruptive NOAA AR 11158 acquired by the Helioseismic and Magnetic Imager onboard the Solar Dynamics Observatory over a five-day observing interval. Besides testing our method extensively, we use it to interpret the dynamical evolution in the AR, including eruptions. We find that the AR builds large budgets of both free magnetic energy and relative magnetic helicity, sufficient to power many more eruptions than the ones it gave within the interval of interest. For each of these major eruptions, we find eruption-related decreases and subsequent free-energy and helicity budgets that are consistent with the observed eruption (flare and coronal-mass-ejection [CME]) sizes. In addition, we find that (1) evolution in the AR is consistent with the recently proposed (free) energy - (relative) helicity diagram of solar ARs, (2) eruption-related decreases occur {\it before} the flare and the projected CME-launch times, suggesting that CME progenitors precede flares, and (3) self-terms of free energy and relative helicity most likely originate from respective mutual-terms, following a progressive mutual-to-self conversion pattern that most likely stems from magnetic reconnection. This results in the non-ideal formation of increasingly helical pre-eruption structures and instigates further research on the triggering of solar eruptions with magnetic helicity firmly placed in the eruption cadre.

astro-ph.SR↗

arXiv2012

Non-neutralized Electric Current Patterns in Solar Active Regions: Origin of the Shear-Generating Lorentz Force

Using solar vector magnetograms of the highest available spatial resolution and signal-to-noise ratio we perform a detailed study of electric current patterns in two solar active regions: a flaring/eruptive, and a flare-quiet one. We aim to determine whether active regions inject non-neutralized (net) electric currents in the solar atmosphere, responding to a debate initiated nearly two decades ago that remains inconclusive. We find that well-formed, intense magnetic polarity inversion lines (PILs) within active regions are the only photospheric magnetic structures that support significant net current. More intense PILs seem to imply stronger non-neutralized current patterns per polarity. This finding revises previous works that claim frequent injections of intense non-neutralized currents by most active regions appearing in the solar disk but also works that altogether rule out injection of non-neutralized currents. In agreement with previous studies, we also find that magnetically isolated active regions remain globally current-balanced. In addition, we confirm and quantify the preference of a given magnetic polarity to follow a given sense of electric currents, indicating a dominant sense of twist in active regions. This coherence effect is more pronounced in more compact active regions with stronger PILs and must be of sub-photospheric origin. Our results yield a natural explanation of the Lorentz force, invariably generating velocity and magnetic shear along strong PILs, thus setting a physical context for the observed pre-eruption evolution in solar active regions.

astro-ph.SR↗

arXiv2012

The Magnetic Energy - Helicity Diagram of Solar Active Regions

Using a recently proposed nonlinear force-free method designed for single vector magnetograms of solar active regions we calculate the instantaneous free magnetic energy and relative magnetic helicity budgets in 162 vector magnetograms corresponding to 42 different active regions. We find a statistically robust, monotonic correlation between the free magnetic energy and the relative magnetic helicity in the studied regions. This correlation implies that magnetic helicity, besides free magnetic energy, may be an essential ingredient for major solar eruptions. Eruptive active regions appear well segregated from non-eruptive ones in both free energy and relative helicity with major (at least M-class) flares occurring in active regions with free energy and relative helicity exceeding 4x10^{31} erg and 2x10^{42} Mx^2, respectively. The helicity threshold agrees well with estimates of helicity contents of typical coronal mass ejections.

astro-ph.SR↗