| dc.description.abstract | Coronal Holes (CHs) are expansive, low density, open magnetic field regions which appear dark in extreme-ultraviolet (EUV) images of the solar corona. CHs are associated with the acceleration of the high speed solar wind (HSSW), which in turn impacts the geomagnetic field of Earth, causing geomagnetic storms. To date, suficient automated monitoring of CHs has not been performed, and CH properties have not been fully correlated with the properties of HSSW streams. In order to accurately observe and monitor CHs, a new method of multi-thermal segmentation was created to rapidly, consistently, and effectively identify CHs in the solar atmosphere. This algorithm, named the Coronal Hole Identification via Multi-thermal Emission Recognition Algorithm (CHIMERA), is the first of its kind and has been used to segment CHs from spacecraft data over two solar cycles. CHIMERA continuously runs live at SolarMonitor.org, outputting segmented images and properties every hour. A detailed investigation into the connections between CHs and their corresponding HSSW streams was undertaken using CHIMERA detections and L1 spacecraft measurements from 2016-2017. This comparison focused on the CH width properties of CHs and the duration and velocity properties of the corresponding HSSW streams. A strong correlation was found between CH widths and the aforementioned properties, and from this multiple new equations were derived from empirical measurements of the solar wind. These equations model the longitudinal expansion of both the solar wind and the expansion of the solar wind flux tube, accounting for a differential velocity of projection of plasma between the leading and trailing edge of HSSW streams. Most notably a derivation was obtained for a simple prediction of solar wind duration, of the form ∆tSW ≈ 0.09∆θCH. A long term investigation into the distribution of CH properties was performed for a 22-year time period from 1998 - 2019. This investigation rendered a probabilistic distribution of CH properties, such as area, flux, and magnetic polarity characteristics. These probabilistic distributions can be used to estimate the occurrences of CHs within a solar cycle, and furthermore as a predictor for more extreme CH events. These distributions combined with already established correlations between CHs and the HSSW renders the probabilistic distribution of corresponding HSSW streams, and hence, estimates the average number of geo-effective storms in an 11-year period. Finally, the extensive CHIMERA catalogue of CH segmentations and properties was used to create a more accurate model for solar wind property predictions using machine learning techniques. A collection of machine learning methods applied to CHIMERA detection from 2010 - 2017 found an improvement of solar wind velocity predictions upon the current operational benchmark, a 27-day persistence model. These improvements were quantified through a number of fitting measurements, with persistence modelling having a coherence = 0.52, root mean squared = 93.7 km s−1, and a running artificial neural network model with a coherence = 0.59, root mean squared = 77.4 km s-1. This thesis has culminated in a significant improvement in the identification of CHs and the predictability of their associated HSSW streams. Future work will expand upon these results through the application of more sophisticated machine learning methods on the entire CHIMERA property database, as well as further investigation into to the correlation of CH and HSSW stream properties to better predict the occurrence of geomagnetic activity. Furthermore, investigations into the correlations between CH magnetic polarities and spherical harmonics observed in helioseismology may assist in unveiling the mysteries of the internal structure of the Sun. | en |
