• The Bulletin of The Korean Astronomical Society
• /
• v.42 no.2
• /
• pp.60.1-60.1
• /
• 2017

In this study we apply Convolution Neural Network(CNN) to solar flare occurrence prediction with various parameter options using the 00:00 UT MDI images from 1996 to 2010 (total 4962 images). We assume that only X, M and C class flares correspond to "flare occurrence" and the others to "non-flare". We have attempted to look for the best options for the models with two CNN pre-trained models (AlexNet and GoogLeNet), by modifying training images and changing hyper parameters. Our major results from this study are as follows. First, the flare occurrence predictions are relatively good with about 80 % accuracies. Second, both flare prediction models based on AlexNet and GoogLeNet have similar results but AlexNet is faster than GoogLeNet. Third, modifying the training images to reduce the projection effect is not effective. Fourth, skill scores of our flare occurrence model are mostly better than those of the previous models.

• Bulletin of the Korean Space Science Society
• /
• 2009.10a
• /
• pp.37.2-37.2
• /
• 2009

In this study, we present a new empirical forecasting method of solar proton events based on flare parameters. For this we used NOAA solar energetic particle (SEP) events from 1976 to 2006 and their associated X-ray flare data. As a result, we found that about only 3.5% (1.9% for M-class and 21.3% for X-class) of the flares are associated with the proton events. It is also found that this fraction strongly depends on longitude; for example, the fraction for $30W^{\circ}$ < L < $90W^{\circ}$ is about three times larger than that for $30^{\circ}E$ < L < $90^{\circ}E$. The occurrence probability of solar proton events for flares with long duration (> 0.3 hours) is about 2 (X-class flare) to 7 (M-class flare) times larger than that for flares with short duration (< 0.3 hours). The relationship between X-ray flare peak flux and proton peak flux as well as its correlation coefficient are strongly dependent on longitude. Using these results for prediction of proton flux, we divided the data into 6 subgroups depending on two parameters: (1) 3 longitude ranges (east, center, and west) and (2) flare impulsive times (long and short). For each subgroup, we make a linear regression between the X-ray flare peak flux and the corresponding proton peak flux. The result shows that the proton flux in the eastern region is much better correlated with the X-ray flux than that in the western region.

• Journal of The Korean Astronomical Society
• /
• v.47 no.6
• /
• pp.209-214
• /
• 2014

In this study we develop a set of solar proton event (SPE) forecast models with NOAA scales by Multi Layer Perceptron (MLP), one of neural network methods, using GOES solar X-ray flare data from 1976 to 2011. Our MLP models are the first attempt to forecast the SPE scales by the neural network method. The combinations of X-ray flare class, impulsive time, and location are used for input data. For this study we make a number of trials by changing the number of layers and nodes as well as combinations of the input data. To find the best model, we use the summation of F-scores weighted by SPE scales, where F-score is the harmonic mean of PODy (recall) and precision (positive predictive value), in order to minimize both misses and false alarms. We find that the MLP models are much better than the multiple linear regression model and one layer MLP model gives the best result.

• Journal of Satellite, Information and Communications
• /
• v.6 no.2
• /
• pp.107-111
• /
• 2011

In World War II, the solar eruption (solar flare) was revealed to make a significant effect to radar systems. The radio disturbance in February 28, 1942 was due to increased cosmic ray during solar maximum. Since such phenomena had been disclosed, many studies were accomplished on solar flare and solar particle event. Now various researches about the effects of solar flare on the spacecrafts, the airplanes flying across the pole, the radar systems, and wireless communication systems are studied. In this paper we analyzed the relationship between the harmful effect on the wireless communication by the solar eruption and the period of solar activity from the sunspot number data and the solar radio burst data for last 40 years.

• Journal of The Korean Astronomical Society
• /
• v.29 no.spc1
• /
• pp.317-319
• /
• 1996

Utilizing a Calcium filter, a large two ribbon flare of an importance 2.5Xj31? was recorded at. King Abdul-Aziz University Solar Observatory (KAAUSO) at the 30th of October 1991. This chromosphenc flare observation, which is of special importance since it is rarely reported, was for a flare that occurred near the south west of the equator at the vicinity of a large sunspot group on an active region known as AR 6891. The observed foot points of this flare had a strange behavior in which the separating motion of the ribbons were not typical of most flares, rather were nearly orthogonal. In this article we present the characteristics of the main sunspot group of this active region and try to investigate its evolution and fragmentation with time. Information regarding magnetic fields and velocity fields are necessary to understand the restructuring of the magnetic field pattern and plasma motion, and hence the changes that could lead to the occurrence of such an interesting flare.

• Publications of The Korean Astronomical Society
• /
• v.14 no.2
• /
• pp.103-112
• /
• 1999

We have investigated the solar activity variation with period shorter than 1000 days, through Fourier transformation of solar cycle 21 and 22 data. And real time predictions of the flare maximum intensity have been made by multilinear regression method to allow the use of multivariate vectors of sunspot groups or active region characteristics. In addition, we have examined the evolution of magnetic field and current density in active regions at times before and after flare occurrence, to check short term variability of solar activity. According to our results of calculation, solar activity changes with periods of 27.1, 28.0, 52.1, 156.3, 333.3 days for solar cycle 21 and of 26.5, 27.1, 28.9, 54.1, 154, 176.7, 384.6 days for solar cycle 22. Periodic components of about 27, 28, 53, 155 days are found simultaneously at all of two solar cycles. Finally, from our intensive analysis of solar activity data for three different terms of $1977\~1982,\; 1975\~1998,\;and\;1978\~1982$, we find out that our predictions coincide with observations at hit rate of $76\%,\;63\%$, 59 respectively.

• Journal of The Korean Astronomical Society
• /
• v.36 no.spc1
• /
• pp.45-47
• /
• 2003

A white light flare was observed at the limb on 16 August 1989 in He 10830 ${\AA}$ spectra, H$\alpha$ slit jaw photo-grams, and white light filter-grams of ${\lambda}=5600{\AA}{\pm}800{\AA}$. The kernels of the white light flare are not spatially related with Ha brightenings, suggesting that the flare energy would be released at the photosphere.

• Journal of The Korean Astronomical Society
• /
• v.29 no.2
• /
• pp.217-221
• /
• 1996

We have examined morphological change and movements of individual sunspots within a sunspot group in association with a large solar flare activity (3B/X1.5) appeared on 13 May 1981. For this purpose we measured distance among spots during the period before and after the flare activity and estimated the average velocity of their movement. Our main results are as follows: (1) The longitudinal displacement among sunspots are generally greater than the latitudinal displacement. (2) During the period the spots moved with an average velocity of 1.2 km/s in longitude and 0.86 km/s in latitude. (3) The most notable change took place in the central part placed between the two ribbons of the flare.

• Journal of Astronomy and Space Sciences
• /
• v.31 no.2
• /
• pp.145-148
• /
• 2014

There was a research on the prolongation of solar cycle 23 by the solar cyclic variation of solar, interplanetary geomagnetic parameters by Oh & Kim (2013). They also suggested that the sunspot number cannot typically explain the variation of total solar irradiance any more. Instead of the sunspot number, a new index is introduced to explain the degree of solar activity. We have analyzed the frequency of sunspot appearance, the length of solar cycle, and the rise time to a solar maximum as the characteristics of solar cycle. Then, we have examined the predictability of solar activity by the characteristics of preceding solar cycle. We have also investigated the hemispheric variation of flare index for the periods that the leading sunspot has the same magnetic polarity. As a result, it was found that there was a good correlation between the length of preceding solar cycle and spotless days. When the length of preceding solar cycle gets longer, the spotless days increase. It is also shown that the shorter rise time to a solar maximum is highly correlated with the increase of sunspots at a solar maximum. Therefore, the appearance frequency of spotless days and the length of solar cycle are more significant than the general sunspot number as an index of declining solar activity. Additionally, the activity of flares leads in the northern hemisphere and is stronger in the hemisphere with leading sunspots in positive polarity than in the hemisphere with leading sunspots in negative polarity. This result suggests that it is necessary to analyze the magnetic polarity's effect on the flares and to interpret the period from the solar maximum to solar maximum as the definition of solar cycle.

• The Bulletin of The Korean Astronomical Society
• /
• v.41 no.1
• /
• pp.46.1-46.1
• /
• 2016

The formation process of a current sheet is important for solar flare from a viewpoint of a space weather prediction. We therefore derive the temporal development of the spatial and statistical distribution of electric current density distributed in a flare-producing active region to describe the formation of a current sheet. We derive time sequence distribution of electric current density by applying a nonlinear force-free approximation reconstruction to Active Region 12158 that produces an X1.6-class flare. The time sequence maps of photospheric vector magnetic field used for reconstruction are captured by a Helioseismic and Magnetic Imager (HMI) onboard Solar Dynamic Observatory (SDO) on 10th September, 2014. The spatial distribution of electric current density in NLFFF model well reproduce observed sigmoidal structure at the preflare phase, although a layer of high current density shrinks at the postflare phase. A double power-law profile of electric current density is found in statistical analysis. This may be expected to use an indicator of the occurrence of a solar flare.