202. Optimization of Solar White-Light Flare Identification Methods and Its Application

Contributed by Yijun Hou. Posted on September 24, 2024

Yingjie Cai1,2, Yijun Hou1,2,3,4, Ting Li1,2,4, Jifeng Liu1,2,5
1. National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, China
2. School of Astronomy and Space Science, University of Chinese Academy of Sciences, Beijing 100049, China
3. Yunnan Key Laboratory of the Solar physics and Space Science, Kunming 650216, China
4. State Key Laboratory of Solar Activity and Space Weather, Beijing 100190, China
5. Institute for Frontiers in Astronomy and Astrophysics, Beijing Normal University, Beijing 102206, China

White-light flares (WLFs) are energetic activity in stellar atmosphere. However, the observed solar WLF is relatively rare compared to stellar WLFs or solar flares observed at other wavelengths, which has posed a significant obstacle to in-depth investigations into the mechanisms behind their formation[1,2], as well as to conducting further statistical comparisons between solar and stellar WLFs[3]. According to the previous finding of weak C-class WLFs[4], we tend to believe that WLFs are not a rare phenomenon in the Sun, and the previous scarcity in the number of solar WLFs could be attributed to limitations in observations or flaws in identification methods. Hence, we aim to improve the solar WLF identification method through conducting an in-depth analysis of optical continuum observations during flares from the SDO/HMI, by far the best telescope for solar flare research. It’s worth noting that although the HMI continuum intensity observation is strictly not equivalent to white-light observation, it can be used as a proxy of white-light and has been widely employed in the solar community.

Figure 1| Background WL emission with inherent fluctuation caused by constant convective motion in different regions. (a): Spatial distribution of the pixels with WL emission enhancement larger than 5%. (b1): Temporal variation of the maximum fluctuation value among all the pixels in quiet Sun region and sunspot region at each moment (δmax1). (b2): Similar to (b1), but for the mean fluctuation value (δmean1). (c1): The proportion histogram of the maximum fluctuation value of every pixel during one hour (δmax2). (c2): Similar to (c1), but for the mean fluctuation value (δmean2).

Traditional WLF identification methods are essentially based on the difference imaging technique, which can highlight WL emission enhancement signals produced by WLFs. Then a fixed threshold of the difference needs to be set for screening enhancement signals. However, as shown in Figure 1, the background fluctuation in the sunspot region is much weaker than that in the quiet Sun region. Therefore, a fixed threshold higher than the background fluctuation in the quiet Sun region usually employed in the traditional method will inevitably result in the following drawbacks: 1) WLF-related signals weaker than the background fluctuation of quiet Sun region will be impossible to be identified; 2) occasional strong WL emission enhancements produced by the constant convection motion will be incorrectly introduced. As a result, in the optimized method, we propose a concept of intrinsic threshold for each pixel defined according to background WL emission with inherent fluctuation caused by constant convective motion. Furthermore, we also impose constraints defined by the typical temporal and spatial distribution characteristics of WLF-related signals.

Figure 2| Performance comparison between the traditional method and the optimized method. (a1), (b1): Spatial and temporal distributions of the WL emission enhancement signals identified by the traditional method and optimized method, respectively. The signal’s appearance time and occurrence number are marked by different colors. The red contours encompass the flare ribbon region observed on the AIA 1600 Å images during the X1.6 flare. (a2), (b2): Corresponding WL light curve profiles during the flare. The red vertical lines denote flare’s start, peak, and end times of GOES SXR 1-8 Å.

To further intuitively compare the effectiveness of the traditional method and the optimized method, we apply them to an X1.6-class WLF. As shown in Figures 2a1 and 2b1, the optimized method identifies a large amount of WL emission enhancement signals within the X1.6 flare ribbon region covering the main sunspot while the traditional method identifies much fewer signals. Moreover, we calculate WL emission light curve of the flare as follows: after determining the qualified pixels having WL emission enhancement signals identified by each method, we sum their intensity together for each moment during the flare and then obtained the WL light curve. It is obvious that the WL light curve obtained by the optimized method shows a significant rapid rise phase and a gradual decay phase while that of the traditional method shows abnormal fluctuations (Figure 2a2 and 2b2).

Figure 3| Percentage of WLFs among different GOES energy levels and two flare types (eruptive and confined). (a): Percentage of WLFs among C-class, M-class, and X-class flares, respectively. (b): Percentage of WLFs among confined and eruptive C-class, M-class, and X-class flares, respectively.

By applying the optimized WLF identification method to 90 flares (30 C-, 30 M-, and 30 X-class flares), we construct a medium solar WLF sample. As show in Figure 3a, the percentages of WLF among the C-class, M-class, and X-class flares are 30%, 60%, and 93.3%, respectively. It is obvious that WLFs are more frequently related to energetic solar eruptions like X-class flares. However, we also identify 9 WLFs among 30 C-class flares, which is the highest identification rate of C-class WLFs (30%) to date to our best knowledge. Additionally, it is found that the proportion of WLFs in confined flares is obviously higher than that in eruptive flares across various GOES energy levels (Figure 3b).

It has been demonstrated that the improved WLF identification method performs well when applied to flares with different energy levels. Based on white-light observations with higher resolution and new channels from the new-generation solar telescopes, we could have a chance to analyze the response of solar WLFs at different wavelengths and establish a sufficiently large database of solar WLFs spanning across the X-, M-, and C-classes, which will lay a base for future statistical studies on solar and stellar WLFs.

For more details of this work, please refer to our publication Ref [5].

References:

[1] Song, Y., & Tian, H. 2018, ApJ, 867, 159
[2] Castellanos Dur´an, J. S., & Kleint, L. 2020, ApJ, 904, 96
[3] Namekata, K., Sakaue, T., Watanabe, K., et al. 2017, ApJ, 851, 91
[4] Jess, D. B., Mathioudakis, M., Crockett, P. J., & Keenan, F. P. 2008, ApJL, 688, L119
[5] Cai, Y., Hou, Y., Li, T., & Liu, J. 2024, ApJ, in press, https://arxiv.org/abs/2408.05381

Leave a comment

Your email address will not be published. Required fields are marked *