Speaker
Description
The properties and the spatial distribution of the large-scale structures of the solar corona determine the observed solar wind structure at 1~au. Coronal holes are a major source of fast solar wind, an important geo-effective component, and appear as large dark patches in extreme ultraviolet images. The solar observatories provide images of the solar corona at different wavelengths, enabling the identification of coronal hole morphology and other large-scale structures along a given line of sight. It is crucial to understand the properties of coronal holes for effective space weather forecasting. The main goal of this work is to develop a threshold-based coronal hole detection tool across solar cycles 23, 24, and 25, using artificial intelligence. Here, we allow the user to input a specific time within solar cycles 23, 24, and 25, enabling the retrieval of the threshold value used to detect coronal hole contours in line-of-sight extreme ultraviolet images from SDO/AIA and SoHO/EIT. We retrieve data from the heliophysics events knowledge database for the large-scale features such as active regions, solar flares, coronal mass ejections, and filaments, and then we engineer them, which allows us to train the neural network model (Prevision Of Phenomena through Coronal-hole Outline Recognition with Neural-network; POP-CORN). The input of the model is the categorical features of the large-scale structures of the solar corona based on their spatial distribution and additional properties, such as solar flare class based on intensity. A neural network model (POP-CORN) was then trained to achieve higher accuracy. The model determines the threshold needed to detect coronal holes, allowing their boundaries to be identified automatically and consistently in extreme ultraviolet images from solar cycles 23, 24, and 25. To interpret the performance of our neural network model (POP-CORN), we divided the predicted coronal hole results into different phases across the solar cycles 23, 24, and 25. Later, we compare them qualitatively and quantitatively with other coronal hole detection tools. We conclude that the properties of large-scale structures affect the determination in coronal hole regions, and incorporating these properties manually into the training improves coronal hole detection. We find that POP-CORN performs well at detecting coronal hole contours, even when many bright features, such as active regions and solar flares, are present, which makes it hard for threshold-based methods to detect dark regions like coronal holes. In the future, we plan to integrate the coronal hole detection tool into a solar wind model validation pipeline, creating a fully automated validation tool that provides a quantitative score for predictions.
| Numerical model | POP-CORN |
|---|
