Abstract Details


Name: Dibya Kirti Mishra
Affiliation: ARIES
Conference ID: TVS202510152
Title: Automatic Detection of Plages in Historical Kodaikanal Solar Observatory Suncharts Using Machine Learning Technique
Authors and Co-Authors: Bibhuti Kumar Jha, Subhamoy Chatterjee, Dibya Kirti Mishra
Abstract Type: Contributory Presentation
Abstract: The Kodaikanal Solar Observatory (KoSO) hosts over a century of multi-wavelength solar observations, including white-light, Ca II K, and H α images. In addition to these photographic records, KoSO has preserved hand-drawn suncharts (1904–2022) marking solar features such as sunspots, plages, filaments, and prominences on the Stonyhurst grid with colour distinction. We present the first analysis of the newly digitized sunchart dataset (6k × 6k resolution), applying supervised machine learning (ML) techniques for automated feature detection. A Convolutional Neural Network (CNN) was implemented to identify the solar limb and north–south line in suncharts from 1909–2007, enabling the derivation of disk centre coordinates, radius, and P-angle. A similar CNN-based segmentation approach was then applied for plage detection, producing a time series covering nine solar cycles (1916–2007). Plage areas from suncharts were compared with contemporaneous Ca II K full-disk data, demonstrating good consistency and confirming the suitability of this ML approach for historical datasets. The results highlight the potential of KoSO suncharts to bridge temporal gaps in the observatory’s multi-wavelength archive and to contribute to the construction of an extended composite plage area series from 1916–2007.