No abstract

Prediction of solar flares is an important task in solar physics. The occurrence of solar flares is highly dependent on the structure and the topology of solar magnetic fields. A new method for predicting large (M and X class) flares is presented, which uses machine learning methods applied to the Zernike moments of magnetograms observed by the Helioseismic and Magnetic Imager (HMI) onboard the Solar Dynamics Observatory (SDO) for a period of six years from 2 June 2010 to 1 August 2016. Magnetic field images consisting of the radial component of the magnetic field are converted to finite sets of Zernike moments and fed to the Support Vector Machine (SVM) classifier.Zernike moments have the capability to elicit unique features from any 2-D image, which may allow more accurate classification. The results indicate whether an arbitrary active region has the potential to produce at least one large flare. We show that ALL AUTHORS AND AFFILIATIONS

Prediction of solar flares due to the effects on Earth and satellites is an important topic for scientists. We develop a method and a tool for flare prediction by applying the support vector machine classifier to unique and independent Zernike moments extracted from active region (AR) images. In the analysis, we used the Helioseismic and Magnetic Imager (HMI) line-of-sight magnetograms, the Atmospheric Imaging Assembly (AIA) ultraviolet (UV at 1600 Å) and extreme ultraviolet (EUV at 304, 171, 193, 211, 335, 94, and 131 Å) images for a period of eight years of the solar cycle 24 (2010 June to 2018 September). The power-law behavior for the frequency distribution of the large flaring time window—the time interval between the occurrence of an AR and first large flare (X- and M-class) therein—indicated that most of the large flares appeared within 150 hr. The True Skill Score (TSS) metric for the performance of the win classifier that (uses the outputs of the HMI and AIA at 193, 211, 94, and 131 Å classifiers) was obtained as 0.86 ± 0.04. We also showed that the maximum value of the TSS for prediction of large flares for the win classifiers was about 0.95 ± 0.03 on the flaring day and decreased to 0.76 ± 0.1 within 4 to 10 days before flaring.

Small-scale extreme ultraviolet (EUV) dimming often surrounds sites of energy release in the quiet Sun. This paper describes a method for the automatic detection of these small-scale EUV dimmings using a feature based classifier. The method is demonstrated using sequences of 171Å images taken by STEREO/EUVI on 13 June 2007 and by SDO/AIA on 27 August 2010. The feature identification relies on recognizing structure in sequences of space-time 171Å images using the Zernike moments of the images. The Zernike moments space-time slices with events and non-events are distinctive enough to be separated using a Support Vector Machine (SVM) classifier. The SVM is trained using 150 event and 700 non-event space-time slices. We find a total of 1217 events in the EUVI images and 2064 events in the AIA images on the days studied. Most of the events are found between latitudes -35 • and +35 • . The sizes and expansion speeds of central dimming regions are extracted using a region grow algorithm. The histograms of the sizes in both EUVI and AIA follow a steep power law with slope about -5. The AIA slope extends to smaller sizes before turning over. The mean velocity of 1325 dimming regions seen by AIA is found to be about 14 km s −1 .

The several-million-degree, low-density quiet solar corona requires a total energy-loss flux of about 3 × 105 erg cm−2 s−1. Solar coronal bright points (CBPs) are ubiquitous in the quiet Sun. They may release magnetic energy to heat the solar corona, but their contribution to the energy flux has not been determined yet. We used an automatic identification and tracking method for CBPs, which was developed based on the support vector machine classifier and Zernike moments of extreme ultraviolet (EUV) observations from the Atmospheric Imaging Assembly (AIA) on board the Solar Dynamics Observatory. We applied a spatial synthesis differential emission measure method and a Vertical-Current Approximation Nonlinear Force-Free Field technique to extract the thermal and magnetic energetics of the CBPs, respectively. By analyzing 7.5 yr (within the solar cycle 24) of AIA observations, we show that the average thermal energy and magnetic free energy of 140,000 CBPs are positively correlated with sunspots. However, the number of CBPs and sunspots are highly anti-correlated. We calculate a total energy-loss flux (sum of the radiative and conductive loss flux) of about (4.84 ± 1.60) × 103 erg cm−2 s−1 for the system of CBPs. Therefore, it is about 1.61% ± 0.53% of the total energy-loss flux of quiet corona. By extending the distribution of the magnetic Poynting flux and energy-loss flux for CBPs to nanoflares, the total magnetic Poynting flux and total energy-loss flux are obtained to be in the range of 1.48 × 105 to 1.57 × 106 and 3.86 × 104 to 2.35 × 105 erg cm−2 s−1, respectively.