Sunspot extraction and hemispheric statistics of YNAO sunspot drawings using deep learning

Abstract

The sunspot drawings around the globe provide long historical records for understanding the long-term trends in solar activity cycle. Yunnan Astronomical Observatory (YNAO) in China contributes the relatively continuous sunspot drawings from 1957 to 2015. This paper proposes a new deep learning method named as SPR-Mask to extract pores, spots, umbrae and penumbrae in the YNAO sunspot drawings. SPR-Mask consists of three parts: backbone, shared head and mask branch. Especially, it adopts a scale-aware attention network (SAAN) and a PointRend module in the mask branch to improve the accuracy of target edge segmentation. Besides that, each sunspot belonging to northern or southern (N-S) hemisphere is determined by transforming its cartesian coordinates to spherical coordinates after extracting P , B0 and L0 handwritten in sunspot drawings using a revised Lenet-5 deep learning method. The precision, recall and AP of SPR-Mask are 0.92, 0.93, and 0.92, respectively. The test results show the SPR-Mask method has a good performance. The numbers and areas of pores, spots, umbrae and penumbrae for N-S hemisphere are presented and analyzed separately. The YNAO data are also compared with Royal Greenwich Observatory (RGO), Kanzelhöhe Observatory (KSO) and Purple Mountain Astronomical Observatory (PMO) data. The results show they have similar trends, high correlations and similar N-S asymmetries. All data of YNAO are public shared at https://github.com/yzs64/YNAO sd/, which are abundant complementary to the other sunspot catalogues in the world.

Extraction of Sunspots from Chinese Sunspot Drawings Based on Semisupervised Learning

“…We also compared the fully supervised and semisupervised methods for detecting and segmenting the drawings of the six Chinese stations. The fully supervised methods include the HTC method (Xu et al 2021) and the SPR-Mask method (Yang et al 2023), which both use the labeled data set of PMO for training. The HTC method is based on the hybrid cascade model for the drawings of PMO.…”

“…A hole represents a hollow region within the spot, but it does not belong to the spot. Holes are relatively rare in the spot (Yang et al 2023). In this work, we first revised the data set by Xu et al (2021).…”

“…Historical sunspot observations provide the most direct data resource for investigating the dynamic changes in solar activity (Lin et al 2019). Conventional sunspot observations not only illustrate the fluctuation of sunspots but also provide valuable insights into comprehending other solar activity phenomena, including solar magnetic field dynamics, solar rotation patterns, and occurrences of flares (Lin et al 2019;Xu et al 2021;Yang et al 2023).…”

“…They built a data set consisting of 2460 images, of which a total of about 39,000 samples were labeled as the training set, divided into four classes: pore, spot, umbra, and hole. Yang et al (2023) proposed a learning method called SPR-Mask and trained it using the data set by Xu et al (2021) to extract the YNAO sunspot drawings. They provided the number and area of each sunspot component in the northern and southern hemispheres, respectively.…”

SIPNet & SAHI: Multiscale Sunspot Extraction for High-Resolution Full Solar Images

“…In recent years, some deep learning methods have been used in the field of sunspots, for example, sunspot extraction from Chinese sunspot drawings [21,22]. Chola [23] adopted AlexNet for classifying sun images into an active sun or quiet sun.…”