Automated Sunspot Detection as an Alternative to Visual Observations
Solar Science Observatory, NAOJ
Sunspot numbers present the variation in solar activity for more than 400 years, and to continue their derivation is important. The sunspot number counts are still based on hand-drawn sketches obtained by visual observations. Therefore, automated detection of sunspots on digital white-light images, which enables objective sunspot counting with small manpower, is required.
For this reason, NAOJ changed the sunspot counting method from hand-drawn visual observation to automated sunspot detection based on digital images in 1998. However, the quality of images is not very high, and detections of false spots and missed detections of true spots often occur. Therefore, we have started higher-quality imaging observation with the Solar Flare Telescope at Mitaka and developed another method to detect sunspots of which the performance is comparable to that of visual observations. An example of the results by the new automated detection method is shown in Figure 1. To identify small spots correctly, the quiet-disk component of the Sun, which is used as a reference to identify sunspots using a threshold, is derived accurately. This threshold is determined using an adaptive method to process images obtained under various conditions. To eliminate the seeing effect, our method can process multiple images taken within a short time.
We applied the developed method to digital images captured by three different observers (the Solar Flare Telescope, Kawaguchi Science Museum, and an amateur observer Mr. S. Morita) and compared the detection results with those of visual observations by Specola Solare Ticinese and Kwasan Observatory of Kyoto University. Figure 2(a) presents the number of sunspots detected by these observations during 2021. Because they show significant scatter, the monthly-mean ratios of the number of sunspots with respect to that of the Solar Flare Telescope are shown in Figure 2(b). The numbers of sunspots detected on the digital images are between two visual observations. We concluded that the proposed sunspot detection method has a similar performance to that of visual observation.
This method can process data taken with various instruments, and therefore, it can be widely used by public observatories and amateurs as well as professional observatories as an alternative to hand-drawn visual observation for sunspot counting. The latest data of sunspot detection from continuum images taken with the Solar Flare Telescope using our method are available on a web page.
These results appeared as Hanaoka, Y. “Automated Sunspot Detection as an Alternative to Visual Observations” in Solar Physics (2022, 297, 158; doi:10.1007/s11207-022-02089-z).
 |
Figure 1. Sunspot detection results for some portions in the white-light image on 2014 February 28 taken with the Solar Flare Telescope. The results of sunspot detection for regions shown in panels (a)--(c) are presented in panels (d)--(f). Penumbrae and umbrae are indicated with red and green patches, respectively. |
 |
Figure 2. Comparison of detection results based on digital white-light images and visual observations for 2021. (a) Daily number of sunspots obtained by automated detection for data taken by the Solar Flare Telescoe (SFT), Kawaguchi Science Museum (KSM), and Mr. S. Morita (SM) and those by Specola Solare Ticinese (SST) and Kwasan Observatory (KO) visual observations. (b) Ratio of the number of sunspots detected with KSM, SM, SST, and KO observations with respect to those by the SFT each month. |
|
.