[Paper Review] Software tool for automatic detection of solar plages in the Coimbra Observatory spectroheliograms

Teresa Barata, Sara Carvalho+5 more|arXiv (Cornell University)|Nov 20, 2018

Solar and Space Plasma Dynamics55 references23 citations

TL;DR

This paper presents a mathematical morphology-based software tool for automatic detection of solar plages in Ca II K3 spectroheliograms from the Coimbra Observatory, using watershed transform and morphological operators. The method achieves high accuracy in detecting chromospheric plages across solar cycle 24, with strong agreement (r = 0.83) to manual and alternative automated methods, even under non-ideal imaging conditions such as cloud cover.

ABSTRACT

Full-disk spectroheliograms have been taken in Coimbra on a daily basis since 1926 in the Ca II K-line (K1 and K3). Later, in 1989, with the upgrade of the equipment it was possible to start the observations in the H-alpha line. The spectroheliograms of Coimbra constitutes a huge dataset of solar images, which requires an efficient automatic tool to detect and analyse solar activity features. This work presents a mathematical morphology approach applied to the CaII K3 series. The objective is to create a tool based on the segmentation by watershed transform combined with other morphological operators to detect automatically and analyse chromospheric plages during the solar cycle 24. The tool is validated by comparing its results for cycle 23 with those presented by Dorotovic et al. (2007, 2010). The results obtained are in very good agreement with those, including on images obtained in non-ideal meteorological conditions (eg. some clouds in sky). The results were also qualitatively compared with the results obtained through the application of ASSA model to SDO HMI magnetograms.

Motivation & Objective

• To develop an automatic, robust method for detecting chromospheric plages in long-term ground-based spectroheliograms from the Coimbra Observatory.
• To enable quantitative analysis of solar activity, particularly facular area and North-South asymmetry, across solar cycle 24.
• To validate the method against existing manual results from cycle 23 and alternative automated methods.
• To ensure applicability to low-quality images affected by atmospheric noise and cloud artifacts without pre-processing.
• To extend the method's utility to other spectroheliogram datasets, such as those from Kharkiv Observatory.

Proposed method

• The method employs mathematical morphology, specifically the watershed transform, to segment and detect bright plage regions in Ca II K3 spectroheliograms.
• It uses morphological operators including erosion, dilation, opening, and closing to enhance and refine the segmentation of plage structures.
• The algorithm applies a thresholding technique after morphological processing to isolate significant plage regions from the background.
• The method does not require pre-processing for limb darkening correction or atmospheric noise removal, making it robust to imaging artifacts.
• The tool computes quantitative metrics including total facular area, northern and southern hemisphere areas, and their asymmetry.
• The approach is validated using cross-comparison with Dorotovič et al. (2007, 2010) results for cycle 23 and with the ASSA model applied to SDO HMI magnetograms.

Experimental results

Research questions

• RQ1 Can a mathematical morphology-based approach reliably detect solar plages in long-term, ground-based spectroheliograms without pre-processing for atmospheric effects?
• RQ2 How does the performance of the proposed automatic method compare to manual detection results from cycle 23?
• RQ3 To what extent does the method maintain accuracy under non-ideal imaging conditions, such as partial cloud cover?
• RQ4 How well does the automatic detection of facular regions correlate with results from the ASSA model applied to SDO HMI magnetograms?
• RQ5 Can the same morphological parameters be successfully applied to spectroheliograms from different observatories, such as Kharkiv?

Key findings

• The method achieved strong quantitative agreement (r = 0.83) with manual results from Dorotovič et al. (2007, 2010) for solar cycle 23.
• The algorithm successfully detected plages in images with atmospheric artifacts like cloud cover, demonstrating robustness without pre-processing.
• The North-South asymmetry of facular regions computed by the method showed high correlation (r = 0.83) with the AKFEAT algorithm.
• The method was successfully applied to Kharkiv Observatory spectroheliograms using the same parameters, confirming cross-instrument applicability.
• The tool is unaffected by limb darkening and does not require intensity normalization or contrast correction.
• The results from the morphological method qualitatively matched those from the ASSA model applied to SDO HMI data, despite the absence of magnetic polarity data in Coimbra’s images.