Short-Term Solar Flare Prediction Using a Sequential Supervised Learning Method

Solar flares are powered by the energy stored in magnetic fields, so evolutionary information of the magnetic field is important for short-term prediction of solar flares. However, the existing solar flare prediction models only use the current information of the active region. A sequential supervised learning method is introduced to add the evolutionary information of the active region into a prediction model. The maximum horizontal gradient, the length of the neutral line, and the number of singular points extracted from SOHO/MDI longitudinal magnetograms are used in the model to describe the nonpotentiality and complexity of the photospheric magnetic field. The evolutionary characteristics of the predictors are analyzed by using autocorrelation functions and mutual information functions. The analysis results indicate that a flare is influenced by the 3-day photospheric magnetic field information before flare eruption. A sliding-window method is used to add evolutionary information of the predictors into machine learning algorithms, then C4.5 decision tree and learning vector quantization are employed to predict the flare level within 48 hours. Experimental results indicate that the performance of the short-term solar flare prediction model within the sequential supervised learning framework is significantly improved.     

Analysis of EUV limb brightening observations from ATM

Magnetic fields in the low corona are the only plausible source of energy for solar flares. Other energy sources appear inadequate or uncorrelated with flares. Low coronal magnetic fields cannot be measured accurately, so most attention has been directed toward measurements of the photospheric magnetic fields from which coronal developments may be inferred. Observations of these magnetic fields are reviewed. It is concluded that, except possibly for the largest flares, changes in the photospheric magnetic fields in flaring centers are confined to evolutionary changes associated with emergence of new magnetic flux. Flare observations with the 10830 Å line of helium, in particular, are discussed. It is concluded that the brightest flare knots appear near points of emergent magnetic flux. Pre-flare activation and eruptions of H filaments are discussed. It is concluded that the rapid motions in filaments indicate unambiguously that the magnetic fields in the low corona are severely disrupted prior to most flares. The coronal signature of H filament eruptions is illustrated with soft X-ray photographs from the S-054 experiment of the NASA Skylab mission. An attempt is made, by studying X-ray flare morphology, to determine whether flares grow by reconnections between adjacent or intertwined magnetic elements or by triggering, in which each flaring loop drives adjacent loops to unstable states. It is concluded that successive loop brightenings are most easily interpreted as the result of magnetic field reconnections, although better time resolution is required to settle the question. A model of magnetic field reconnections for flares associated with filament activation and emerging magnetic flux is presented.     

Correlation Between Solar Flare Productivity and Photospheric Magnetic Field Properties

From a large number of SOHO/MDI longitudinal magnetograms, three physical measures including the maximum horizontal gradient, the length of the neutral line, and the number of singular points are computed. These measures are used to describe photospheric magnetic field properties including nonpotentiality and complexity, which is believed to be closely related to solar flares. Our statistical results demonstrate that solar flare productivity increases with nonpotentiality and complexity. Furthermore, the relationship between the flare productivity and these measures can be well fitted with a sigmoid function. These results can be beneficial to future operational flare forecast models.     

Pre-flare association of magnetic fields and millimeter-wave radio emission

Observations of radio emission at 3.3 mm wavelength associated with magnetic fields in active regions are reported. Results of more than 200 regions during the years 1967–1968 show a strong correlation between peak enhanced millimeter emission, total flux of the longitudinal component of photospheric magnetic fields and the number of flares produced during transit of active regions. For magnetic flux greater than 10²¹ maxwells flares will occur and for flux of 10²³ maxwells the sum of the H flare importance numbers is about 40. The peak millimeter enhancement increases with magnetic flux for regions which subsequently flared. Estimates of the magnetic energy available and the correlation with flare production indicate that the photospheric fields and probably chromospheric currents are responsible for the observed pre-flare heating and provide the energy of flares.This work was supported in part by NASA Contract No. NAS2-7868 and in part by Company funds of The Aerospace Corporation.     

Magnetic fields in flares and active prominences

The locations of flares and chromospheric absorption features on May 21 and 23, 1967, are compared with a series of H magnetograms. Each of the four major flares included in the study developed as double emission ribbons lying at positions of steep field gradient on opposite sides of the boundary between regions of opposite magnetic polarity. At certain stages, the flare outlines followed closely the isogauss contours of the longitudinal field. A fluctuating field of 75 gauss was measured directly in the importance two flare of May 21. Modifications in the magnetic structure of the active region followed the flares of May 23.     

AR6659光球色球的磁场速度场的演化特征及其与大耀斑的关系

ＡＲ６６５９是２２周以来最重要的一个活动区，它爆发了２２周最强大的高能事件。本文用云南天文台的光球、色球精细结构照片和北京天文台怀柔站的磁场速度场资料，分析了该活动区磁场速度场的二维位形和大耀斑期间的演化特征。本文分析的４个大耀斑均爆发在中性线附近的Ｎ极区磁场梯度大的地方及色球速度场的红移区。偏带观测也显示耀斑物质是向红端移动的。耀斑波沿横场传播在离本黑子群几万至十几万公里的地方激起感生耀斑，在原生耀斑与感生耀斑之间往往有耀斑环相连。此外，本文还从演化特征出发分析了耀斑爆发前活动区等离子体的宏观不稳定性。     

Flares and changing magnetic fields

An observational study of maps of the longitudinal component of the photospheric fields in flaring active regions leads to the following conclusions:

1. The broad-wing Hα kernels characteristic of the impulsive phase of flares occur within 10″ of neutral lines encircling features of isolated magnetic polarity (‘satellite sunspots’).
2. Photospheric field changes intimately associated with several importance 1 flares and one importance 2B flare are confined to satellite sunspots, which are small (10″ diam). They often correspond to spot pores in white-light photographs.
3. The field at these features appears to strengthen in the half hour just before the flares. During the flares the growth is reversed, the field drops and then recovers to its previous level.
4. The magnetic flux through flare-associated features changes by about 4 × 10¹⁹ Mx in a day. The features are the same as the ‘Structures Magnétiques Evolutives’ of Martres et al. (1968a).
5. An upper limit of 10²¹ Mx is set for the total flux change through McMath Regions 10381 and 10385 as the result of the 2B flare of 24 October, 1969.
6. Large spots in the regions investigated did not evince flux changes or large proper motions at flare time.
7. The results are taken to imply that the initial instability of a flare occurs at a neutral point, but the magnetic energy lost cannot yet be related to the total energy of the subsequent flare.
8. No unusual velocities are observed in the photosphere at flare time.

     

Magnetic reconnection,electric field,and particle acceleration in the July 14, 2000 solar flare

A topological model with magnetic reconnection at two separators in the corona is used to account for the recently discovered changes of the photospheric magnetic field in the active region NOAA 9077 during the July 14, 2000 flare. The model self-consistently explains the following observed effects: (1) the magnetic field strength decreases on the periphery of the active region but increases in its inner part near the neutral line of the photospheric magnetic field; (2) the center-of-mass positions of the fields of opposite (northern and southern) polarities converge; and (3) the magnetic flux of the active region decreases after the flare. The topological model gives not only a qualitative interpretation of the flare phenomena (the structure of the interacting magnetic fluxes in the corona, the location of the energy sources, the shape of the flare ribbons and kernels in the chromosphere and photosphere), but also correct quantitative estimates of the large-scale processes that form the basis for solar flares. The electric field emerging in the flare during large-scale reconnection is calculated. The electric field strength correlates with the observed intensity of the hard X-ray bremsstrahlung, suggesting an electron acceleration as a result of reconnection.     

Flare build-up in preflare magnetic loops and nonlinear force-free magnetic fields

In this paper, we extend B. C. Low's study on nonlinear force-free magnetic fields. Based on Low's mathematical method, a revised boundary-value problem of the two-dimensional nonlinear force-free magnetic field is solved analytically. The solution shows that higher magnetic loops evolve towards preflare loops when the gradient of longitudinal magnetic field at the photospheric level and the angle (shear) included between the magnetic field line and magnetic neutral line increase with time. The density, temperature and the current density are higher in the preflare loops than in the high-lying magnetic loops. We believe it is the loops that provide conditions for the eruption of the flare.The original has been published in the Acta Astronomica Sinica 21 (1980), 152, in Chinese. The present paper completes the discussion and revises some of the preliminary results.     

Hβ Doppler-velocity fields within sites of flares in a solar active region

In this paper, we analyze the relationship between photospheric magnetic fields and chromospheric velocity fields in a solar active region, especially evolving features of the chromospheric velocity field at preflare sites. It seems that flares are related to unusually distributed velocity field structures, and initial bright kernels and ribbons of the flares appear in the red-shifted areas (i.e., downward flow areas) close to the inversion line of H Dopplergrams with steep gradients of the velocity fields, no matter whether the areas have simple magnetic structure or a weak magnetic field, or strong magnetic shear and complex structure of the magnetic fields. The data show that during several hours prior to the flares, while the velocity field evolves, the sites of the flare kernels (or ribbons) with red-shifted features come close to the inversion line of the velocity field. This result holds regardless of whether or not the flare sites are wholly located in blue-shifted areas (i.e., upward flow areas), or are far from the inversion line of the Doppler velocity field (V = 0 line), or are partly within red-shifted areas. There are two cases favourable for the occurrence of flares, one is that the gulf-like neutral lines of the magnetic field (B = 0 line) occur in the H red-shifted areas, the other is that the gulf-like inversion lines of the H Doppler velocity field (V = 0 line) occur in the unipolar magnetic areas. These observational facts indicate that the velocity field and magnetic field have the same effect on the process of flare energy accumulation and release.     

Flares,magnetic configurations,and magnetic energy release

Using a newly developed Aerospace digital videomagnetograph, three solar active regions are studied as to their magnetic configurations and their flare productivity. These three regions have very different types of magnetic configurations and different types of flare productivity. We review previous theoretical and experimental research on flares and magnetic energy storage, and discuss various ways to observe magnetic energy release due to flares. Results for six subflares are presented. Five showed no measurable magnetic energy change and one result is questionable.We show three counterexamples to Zirin's (1972) contention that as a rule H plage brightness is proportional to magnetic field strength. Each of these three cases involved two plage regions of the same polarity and equal field strengths with one of the plages adjacent to a neutral line. In all three cases the plage region nearer the neutral line was much brighter.     

A quantitative study relating observed shear in photospheric magnetic fields to repeated flaring

In this paper we present a quantitative evaluation of the shear in the magnetic field along the neutral line in an active region during an epoch of flare activity. We define shear as the angular difference in the photosphere between the potential magnetic field, which fits the boundary conditions imposed by the observed line-of-sight field, and the observed magnetic field. For the active region studied, this angular difference (shear) is non-uniform along the neutral line with maxima occurring at the locations of repeated flare onsets. We suggest that continued magnetic evolution causes the field's maximum shear to exceed a critical value of shear, resulting in a flare around the site of maximum shear. Evidently, the field at the site of the flare must relax to a state of shear somewhat below the critical value (but still far from potential), with subsequent evolution returning the field to the critical threshold. We draw this inference because several flares occurred at sites of maximum photospheric shear which were persistent in location.NOAA, Boulder, Colorado.     

Vector magnetic field evolution,energy storage,and associated photospheric velocity shear within a flare-productive active region

The evolution of vector photospheric magnetic fields has been studied in concert with photospheric spot motions for a flare-productive active region. Over a three-day period (5–7 April, 1980), sheared photospheric velocity fields inferred from spot motions are compared both with changes in the orientation of transverse magnetic fields and with the flare history of the region. Rapid spot motions and high inferred velocity shear coincide with increased field alignment along the B _L= 0 line and with increased flare activity; a later decrease in velocity shear precedes a more relaxed magnetic configuration and decrease in flare activity. Crude energy estimates show that magnetic reconfiguration produced by the relative velocities of the spots could cause storage of 10³² erg day^–1, while the flares occurring during this time expended 10³¹ erg day^–1.Maps of vertical current density suggest that parallel (as contrasted with antiparallel) currents flow along the stressed magnetic loops. For the active region, a constant-, force-free magnetic field (J = B) at the photosphere is ruled out by the observations.Presently located at NASA/MSFC, Huntsville, Ala. 35812, U.S.A.     

How are Emerging Flux,Flares and CMEs Related to Magnetic Polarity Imbalance in Midi Data?

In order to understand whether major flares or coronal mass ejections (CMEs) can be related to changes in the longitudinal photospheric magnetic field, we study 4 young active regions during seven days of their disk passage. This time period precludes any biases which may be introduced in studies that look at the field evolution during the short-term flare or CME period only. Data from the Michelson Doppler Imager (MDI) with a time cadence of 96 min are used. Corrections are made to the data to account for area foreshortening and angle between line of sight and field direction, and also the underestimation of the flux densities. We make a systematic study of the evolution of the longitudinal magnetic field, and analyze flare and CME occurrence in the magnetic evolution. We find that the majority of CMEs and flares occur during or after new flux emergence. The flux in all four active regions is observed to have deviations from polarity balance both on the long term (solar rotation) and on the short term (few hours). The long-term imbalance is not due to linkage outside the active region; it is primarily related to the east–west distance from central meridian, with the sign of polarity closer to the limb dominating. The sequence of short-term imbalances are not closely linked to CMEs and flares and no permanent imbalance remains after them. We propose that both kinds of imbalance are due to the presence of a horizontal field component (parallel to the photospheric surface) in the emerging flux.     

Field-Guided Proton Acceleration at Reconnecting x-Points in Flares

An explicitly energy-conserving full orbit code CUEBIT, developed originally to describe energetic particle effects in laboratory fusion experiments, has been applied to the problem of proton acceleration in solar flares. The model fields are obtained from solutions of the linearised MHD equations for reconnecting modes at an X-type neutral point, with the additional ingredient of a longitudinal magnetic field component. To accelerate protons to the highest observed energies on flare timescales, it is necessary to invoke anomalous resistivity in the MHD solution. It is shown that the addition of a longitudinal field component greatly increases the efficiency of ion acceleration, essentially because it greatly reduces the magnitude of drift motions away from the vicinity of the X-point, where the accelerating component of the electric field is largest. Using plasma parameters consistent with flare observations, we obtain proton distributions extending up to -ray-emitting energies (>1 MeV). In some cases the energy distributions exhibit a bump-on-tail in the MeV range. In general, the shape of the distribution is sensitive to the model parameters.     

A study of the magnetic and velocity fields in an active region

A time sequence of magnetograms and velocity-grams in the H and Fe i 6569 Å lines has been made at a rate of 12 h^–1 of McMath Region 10385 from 26 to 29 October, 1969. The 14 flares observed during this period have been studied in relation to the configuration and changes in the magnetic and velocity fields. There was little correlation between flare position and the evolutionary changes in the photospheric magnetic and velocity field, except at large central meridian distances where the velocity observations suggested shearing taking place at flare locations. At central meridian distances > 30° we found that flares are located in areas of low line-of-sight photospheric velocity surrounded by higher velocity hills. The one exception to this was the only flare which produced a surge. Blue-shifted velocity changes in the photosphere of 0.3 to 1 km s^–1 were observed in localized areas at the times of 8 of 14 flares studied.Visiting Astronomer, Kitt Peak National Observatory.Operated by the Association of Universities for Research in Astronomy, Inc., under contract with the National Science Foundation.     

Photospheric and Chromospheric Activity Associated with 3B Flare of February 27, 1992

We have analyzed the H filtergrams and vector magnetograms of the active region NOAA 7070, in which a 3B/X3.3 flare occurred on February 27, 1992. The average area per sunspot of this active region was in declining phase at the time of the flare. The vector magnetograms indicate that the magnetic field was non-potential at the flaring site. Besides non-potentiality, the longitudinal field gradient was found to be the highest at the region showing initial H brightening. Further, in H filtergrams no appreciable change in the morphology of the filament tracing the magnetic neutral line was noticed in the post-flare stage. Also, the photospheric vector magnetograms show considerable shear in post-flare magnetic field of the active region. In this paper we present the observations and discuss the possible mechanism responsible for the 3B/X3.3 flare.     

Observed nonpotential magnetic fields and the inferred flow of electric currents at a location of repeated flaring

We have analyzed the vector magnetic field of an active region at a location of repeated flaring to determine the nature of the currents flowing in the areas where the flares initiated. The component of electric current density crossing the photosphere along the line-of-sight was derived from the observed transverse component of the magnetic field. The maximum concentrations of these currents occurred exactly at the sites of flare initiation and where the photospheric field was sheared the most. The calculated distribution of current density at the flare sites suggested that currents were flowing out of an area of positive magnetic polarity and across the magnetic inversion line into two areas of negative polarity. This interpretation was reinforced by a calculation of the source field, the magnetic field produced in the photosphere by the electric currents above the photosphere. In the vicinity of the flare sites, the calculated source field exhibited three particular characteristics: (1) maximum magnitudes at the sites of flare initiation, (2) a rotational direction where the vertical current density was concentrated, and (3) a fairly constant angular orientation with the magnetic inversion line. The source field was thus very similar to the field produced by two arcades of currents crossing the inversion line at the locations of greatest magnetic shear with orientations of about 60° to the inversion line. With this orientation, the inferred arcades would be aligned with the observed chromospheric fibrils seen in the H data so that the currents were field-aligned above the photosphere. The field thus exhibited a vertical gradient of magnetic shear with the shear decreasing upward from the photosphere. We estimated the currents in the two arcades by matching the source field derived from observations with that produced by a model of parallel loops of currents. We found that the loops of the model would each have a radius of 4500 km, a separation of 1830 km, and carry a current of 0.15 × 10¹² A. Values of vertical current densities and source fields appearing in the umbrae of the two large sunspots away from the flare sites were shown to lie at or below the level of uncertainty in the data. The main source of this uncertainty lay in the method by which the 180° ambiguity in the azimuth of the transverse field is resolved in umbral areas. We thus concluded that these quantities in large umbrae should be treated with a healthy skepticism. Finally, we found that the source field at the flare sites was produced almost entirely by the angular difference between the observed and potential field and not by the difference in field intensity.     

Coincidence between Hα flare kernels and peaks of observed longitudinal electric current densities

We compare large-scale filtergrams of a hitherto neglected class 1B flare with previously published vector magnetograms and maps of photospheric longitudinal electric current density (Hagyard et al., 1985). The vector magnetic fields were mapped simultaneously with the eruption of this flare. We find a coincidence, to within the ±2″ registration accuracy of the data, between the flare kernels and the locations of maximum shear and of peak values in the longitudinal electric current density. The kernels brighten in a way which implies that the preflare heating and the main release of flare energy are spatially coincident within the limits of resolution (≈2″). A pronounced magnetic shear exists in the vertical direction at the location of the strongest flare kernels. We provide evidence that the electric currents could be maintained by the energy stored in the sheared transverse magnetic field and that the amount of energy released is proportional to the amount stored. These circumstances are consistent with theories in which flares are triggered by plasma instabilities due to surplus electric currents.