Development of mathematical morphology and pattern recognition algorithms to characterize solar activity

Solar activity acts as the primary driver of space weather, encompassing phenomena such as flares, coronal mass ejections, and solar energetic particles, which originate from distinct regions like sunspots, faculae, and granules. Predicting space weather is essential due to its significant impact on planetary atmospheres and human activities both in space and on Earth, including communication and navigation systems, power grids, spacecraft operations, astronaut safety, and aviation at high altitudes and latitudes. Sunspots, in particular, harbour intense magnetic fields that can trigger powerful solar eruptions and therefore require continuous monitoring. In recent years, automated identification methods have become increasingly common as the vast amounts of data from both ground- and space-based observatories can no longer be handled manually.

In this thesis, we developed mathematical morphology algorithms for automatic sunspot detection. Our results, compared with those obtained through manual methods, demonstrate comparable accuracy, enabling the extension of this approach to other solar features and image datasets. For instance, we applied a modified version of these algorithms to simulation-generated maps to identify magnetic flux rope structures in two active regions, capturing their distinct dynamics and, in one case, an eruption corroborated by Solar Dynamics Observatory (SDO) observations.

Mathematical morphology was further applied to extensive datasets to identify all coronal off-limb structures visible in SDO/Atmospheric Imaging Assembly (AIA) 304 Å images throughout nearly the entire Solar Cycle 24 (June 2010–December 2021). Statistical analysis of the properties of these features revealed significant trends, including distinct behaviours between high- and low-intensity coronal structures in terms of latitudinal distribution. Additionally, evidence was found for the existence of active longitudes, with coronal off-limb structures exhibiting a preferred longitudinal distribution pattern. These findings deepen our understanding of coronal activity and open new avenues for improving space weather forecasting.