Abstract Details


Name: rohan mandrai
Affiliation: Indian Institute of Technology (B.H.U.) Varanasi
Conference ID: TVS202510203
Title: Retrieval of historical magnetic field by employing deep learning model on Ca II K and Hα lines
Authors and Co-Authors: Dibya Kriti Mishra, Subhamoy Chatterjee, Anu Sreedevi, Bidya Binay Karak
Abstract Type: Contributory Presentation
Abstract: Magnetic field information of the Sun is crucial for understanding the long-term behavior of the Sun and the solar dynamo models. However, the regular and direct observations of the magnetic field have only been available since 1975, which covers approximately the past five solar cycles. This timespan is insufficient to fully understand the long-term behavior of the Sun and constrain the solar dynamo models. The intensity of the Ca II K line is found to be a good proxy of the strength of the magnetic field, while the polarity inversion line obtained from the Hα data can give the polarity of the magnetic field. In this work, we reconstruct solar magnetograms using recently calibrated Ca II K (1904-2007) and Hα (1912-2007) images, which have been available in the Kodaikanal Solar Observatory (KoSO) for the last century, as proxies of the unsigned magnetic field strength and sign, respectively. We apply a deep learning approach, with a U-Net architecture as the generator and PatchGAN as the discriminator, to translate chromospheric images into magnetic field maps. The preliminary model demonstrates promising results in terms of structural similarity (SSIM) and peak signal-to-noise ratio (PSNR) values on test data, indicating effective reconstruction of magnetograms from proxy images.