Presentation Information
[O12-P74]Atmospheric Turbulence Evaluation Based on Solar Images: Methodological Development and Comprehensive Analysis
*Saki Saki Ochiai1, *Jun Jun Shimano1, Miku Miku Aritake2, Ayano Ayano Sato2, Asumi Asumi Takahashi2, Mio Mio Takahashi2, Misa Misa Tomoyori2 (1. Tokyo Metropolitan Fuji high school Astronomy Club, 2. Tokyo Metropolitan Fuji High School Affiliated Junior High School Astronomy Club)
Keywords:
atmospheric turbulence,seeing,scintillation
1. Purpose
During our study of sunspots, solar images were distorted by atmospheric turbulence. This study aims to identify the ideal observation site for sunspot imaging at our school in Nakano, Tokyo, where atmospheric disturbance is minimized. We focused on two indicators-seeing and scintillation- and quantified turbulence using a newly developed program.
2. Observations
The instruments used for the observations are listed in Table 1, and the observation sites are shown in Table 2. The diffraction limit, L = 1.22lambda/D, for this system is 2.09 arcsec, and the pixel scale is 3.712 arcsec per pixel. According to previous studies (H. Socas-Navarro et al.), seeing conditions tend to improve at higher altitudes or near bodies of water. Therefore, we conducted solar observations on the school rooftop and at the poolside (Table 3).
At each site, we recorded 87.4 fps, 16-bit, 1000-frame sequences, taken seven times per day (7000 frames/day), over four days (56,000 frames total) under clear skies. Before observations, we confirmed that there were no performance differences between the telescopes used (Tokyo Metropolitan Fuji High School, 2025, “Measurement of Seeing Using Solar Images,” Kagawa Soubun). Atmospheric conditions such as pressure, wind direction, and humidity were also recorded (Table 3).
3. Analysis Methods
(1) Circle Detection Using the Least-Squares Method The solar image is assumed to be circular. We divide the image into equal vertical and homrizontal segments and select the 18 lines (9 vertical, 9 horizontal) that intersect the solar disk. For each line, the point of maximum change in the derivative of brightness is taken as a provisional solar limb point. A circle approximating these points is then computed using the least-squares method. (2) Determining Seeing Using the least-squares circle detection, we obtain a circle approximating the solar disk. The second derivative of brightness near the solar limb is calculated, and the maximum value is taken as the actual limb position (dn in Fig. 2). The standard deviation between this limb and the fitted circle (d in Fig. 2) is defined as the seeing value. Larger values indicate stronger turbulence and poorer seeing. (3) Determining Scintillation
For each frame, a circle is fitted using the least-squares method. Within a 37 x 37 pixel region centered on the circle center (1369 pixels), the standard deviation of brightness over the entire video is computed for each pixel. The average of these values is defined as the scintillation index for that video.
Previously, circle detection was performed using the Hough transform. However, parameter sensitivity and processing artifacts produced errors of approximately 3%. Therefore, in this study, we adopted a least-squares method using limb-extracted point sets.
4. Results
Using the above methods, we quantified the magnitude of image distortion in the observed solar images (Table 4). Averaging all seeing measurements, the observed solar radius R relative to the theoretical radius ave(r) yielded: Rooftop: R=ave(r)±14.74″ Poolside: R=ave(r)±18.15″ We also examined the correlation between the standard deviation of seeing over 100 consecutive frames and the scintillation index for each video. No significant correlation was found (Fig. 4).
5. Discussion
The rooftop seeing size was 3.41 arcsec smaller than that of the poolside. However, this difference is smaller than the pixel resolution of 3.712 arcsec, indicating that both locations can produce images of comparable sharpness.
Table 4 shows that seeing conditions varied noticeably between observation days, and similar trends were observed across multiple dates. The 1sigma variation of the seven seeing measurements per day was 0.10-0.80 arcsec, which is extremely small compared with the overall seeing values of 8-20 arcsec. This suggests that seeing remains nearly constant over short timescales within the same day.
Among the four observation days, only July 20 showed better seeing at the poolside.
From Fig. 4, the correlation coefficients between the 100-frame seeing standard deviation and scintillation index were mostly below |0.4-0.6|, and although some values were larger, they were not stable. Thus, no meaningful correlation between seeing and scintillation was identified.
6. References
1.Socas-Navarro, H., et al. (2005). Solar Site Survey for the Advanced Technology Solar Telescope. I. Analysis of the Seeing Data. Publications of the Astronomical Society of the Pacific, 117(837), 1296-1305.
2.Tokyo Metropolitan Fuji High School (2025). Measurement of Seeing Using Solar Images. Kagawa Soubun (National High School Cultural Festival).
During our study of sunspots, solar images were distorted by atmospheric turbulence. This study aims to identify the ideal observation site for sunspot imaging at our school in Nakano, Tokyo, where atmospheric disturbance is minimized. We focused on two indicators-seeing and scintillation- and quantified turbulence using a newly developed program.
2. Observations
The instruments used for the observations are listed in Table 1, and the observation sites are shown in Table 2. The diffraction limit, L = 1.22lambda/D, for this system is 2.09 arcsec, and the pixel scale is 3.712 arcsec per pixel. According to previous studies (H. Socas-Navarro et al.), seeing conditions tend to improve at higher altitudes or near bodies of water. Therefore, we conducted solar observations on the school rooftop and at the poolside (Table 3).
At each site, we recorded 87.4 fps, 16-bit, 1000-frame sequences, taken seven times per day (7000 frames/day), over four days (56,000 frames total) under clear skies. Before observations, we confirmed that there were no performance differences between the telescopes used (Tokyo Metropolitan Fuji High School, 2025, “Measurement of Seeing Using Solar Images,” Kagawa Soubun). Atmospheric conditions such as pressure, wind direction, and humidity were also recorded (Table 3).
3. Analysis Methods
(1) Circle Detection Using the Least-Squares Method The solar image is assumed to be circular. We divide the image into equal vertical and homrizontal segments and select the 18 lines (9 vertical, 9 horizontal) that intersect the solar disk. For each line, the point of maximum change in the derivative of brightness is taken as a provisional solar limb point. A circle approximating these points is then computed using the least-squares method. (2) Determining Seeing Using the least-squares circle detection, we obtain a circle approximating the solar disk. The second derivative of brightness near the solar limb is calculated, and the maximum value is taken as the actual limb position (dn in Fig. 2). The standard deviation between this limb and the fitted circle (d in Fig. 2) is defined as the seeing value. Larger values indicate stronger turbulence and poorer seeing. (3) Determining Scintillation
For each frame, a circle is fitted using the least-squares method. Within a 37 x 37 pixel region centered on the circle center (1369 pixels), the standard deviation of brightness over the entire video is computed for each pixel. The average of these values is defined as the scintillation index for that video.
Previously, circle detection was performed using the Hough transform. However, parameter sensitivity and processing artifacts produced errors of approximately 3%. Therefore, in this study, we adopted a least-squares method using limb-extracted point sets.
4. Results
Using the above methods, we quantified the magnitude of image distortion in the observed solar images (Table 4). Averaging all seeing measurements, the observed solar radius R relative to the theoretical radius ave(r) yielded: Rooftop: R=ave(r)±14.74″ Poolside: R=ave(r)±18.15″ We also examined the correlation between the standard deviation of seeing over 100 consecutive frames and the scintillation index for each video. No significant correlation was found (Fig. 4).
5. Discussion
The rooftop seeing size was 3.41 arcsec smaller than that of the poolside. However, this difference is smaller than the pixel resolution of 3.712 arcsec, indicating that both locations can produce images of comparable sharpness.
Table 4 shows that seeing conditions varied noticeably between observation days, and similar trends were observed across multiple dates. The 1sigma variation of the seven seeing measurements per day was 0.10-0.80 arcsec, which is extremely small compared with the overall seeing values of 8-20 arcsec. This suggests that seeing remains nearly constant over short timescales within the same day.
Among the four observation days, only July 20 showed better seeing at the poolside.
From Fig. 4, the correlation coefficients between the 100-frame seeing standard deviation and scintillation index were mostly below |0.4-0.6|, and although some values were larger, they were not stable. Thus, no meaningful correlation between seeing and scintillation was identified.
6. References
1.Socas-Navarro, H., et al. (2005). Solar Site Survey for the Advanced Technology Solar Telescope. I. Analysis of the Seeing Data. Publications of the Astronomical Society of the Pacific, 117(837), 1296-1305.
2.Tokyo Metropolitan Fuji High School (2025). Measurement of Seeing Using Solar Images. Kagawa Soubun (National High School Cultural Festival).