• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.1
• /
• pp.42.1-42.1
• /
• 2021

In this study, we present a visual explanation of a deep learning solar flare forecast model and its relationship to physical parameters of solar active regions (ARs). For this, we use full-disk magnetograms at 00:00 UT from the Solar and Heliospheric Observatory/Michelson Doppler Imager and the Solar Dynamics Observatory/Helioseismic and Magnetic Imager, physical parameters from the Space-weather HMI Active Region Patch (SHARP), and Geostationary Operational Environmental Satellite X-ray flare data. Our deep learning flare forecast model based on the Convolutional Neural Network (CNN) predicts "Yes" or "No" for the daily occurrence of C-, M-, and X-class flares. We interpret the model using two CNN attribution methods (guided backpropagation and Gradient-weighted Class Activation Mapping [Grad-CAM]) that provide quantitative information on explaining the model. We find that our deep learning flare forecasting model is intimately related to AR physical properties that have also been distinguished in previous studies as holding significant predictive ability. Major results of this study are as follows. First, we successfully apply our deep learning models to the forecast of daily solar flare occurrence with TSS = 0.65, without any preprocessing to extract features from data. Second, using the attribution methods, we find that the polarity inversion line is an important feature for the deep learning flare forecasting model. Third, the ARs with high Grad-CAM values produce more flares than those with low Grad-CAM values. Fourth, nine SHARP parameters such as total unsigned vertical current, total unsigned current helicity, total unsigned flux, and total photospheric magnetic free energy density are well correlated with Grad-CAM values.

• Journal of Sensor Science and Technology
• /
• v.9 no.3
• /
• pp.163-170
• /
• 2000

Optical frequency modulated fiber optic interferometric sensor was developed to sense the mechanical quantities, such as displacement, strain, force etc. It has been difficult to distinguish whether the increase of the mechanical quantities or the decrease of the quantities measured by the conventional fiber optic interferometric sensors because their signals only have a sinusoidal wave pattern related to the change of mechanical quantities. In this study, in order to measure the mechanical quantifies with the distinction of the changing direction of the quantities, the fiber of optic Michelson interferometric sensor was simply constructed by the laser light modulated with saw tooth wave pattern. The output signal of the sensor was controlled as the sinusoidal wave. The signal processing was based on the counting of the wave number of the output signal during constant time duration. The strain was determined by the cumulative value of the wave number producted by the gage factor. In order to verify the strain measurement capability of this sensor, the strain increase-decrease test was performed by universal testing machine installed with the aluminum specimen bonded with the fiber optic sensor and electrical strain gage. In the result of the test, the strain from the fiber optic sensor had a good agreement with the values from the electrical strain gage.

• Journal of The Korean Astronomical Society
• /
• v.37 no.1
• /
• pp.41-53
• /
• 2004

It has been a big mystery what drives filament eruptions and flares. We have studied in detail an X1.8 flare and its associated filament eruption that occurred in NOAA Active Region 9236 on November 24,2000. For this work we have analyzed high temporal (about 1 minute) and spatial (about 1 arcsec) resolution images taken by Michelson Doppler Imager (MDI) onboard the Solar and Heliospheric Observatory, Hoc centerline and blue wing ($-0.6{\AA}$) images from Big Bear Solar Observatory, and 1600 ${\AA}$ UV images by the Transition Region and Corona Explorer (TRACE). We have found that there were several transient brightenings seen in H$\alpha$ and, more noticeably in TRACE 1600 ${\AA}$ images around the preflare phase. A closer look at the UV brightenings in 1600 ${\AA}$ images reveals that they took place near one end of the erupting filament, and are a kind of jets supplying mass into the transient loops seen in 1600 ${\AA}$. These brightenings were also associated with canceling magnetic features (CMFs) as seen in the MDI magnetograms. The flux variations of these CMFs suggest that the flux cancellation may have been driven by the emergence of the new flux. For this event, we have estimated the ejection speeds of the filament ranging from 10 to 160 km $s^{-1}$ for the first twenty minutes. It is noted that the initiation of the filament eruption (as defined by the rise speed less than 20 km $s^{-1}$) coincided with the preflare activity characterized by UV brightenings and CMFs. The speed of the associated LASCO CME can be well extrapolated from the observed filament speed and its direction is consistent with those of the disturbed UV loops associated with the preflare activity. Supposing the H$\alpha$/UV transient brightenings and the canceling magnetic features are due to magnetic reconnect ion in the low atmosphere, our results may be strong observational evidence supporting that the initiation of the filament eruption and the preflare phase of the associated flare may be physically related to low-atmosphere magnetic reconnection.