Solar Flare Prediction Through Anomaly Detection of H[alpha] Wavelength Images, Incorporating Automated Filament Detection and Characterization
Author | : Jezreel Bassett |
Publisher | : |
Total Pages | : 170 |
Release | : 2015 |
ISBN-10 | : OCLC:920576252 |
ISBN-13 | : |
Rating | : 4/5 ( Downloads) |
Download or read book Solar Flare Prediction Through Anomaly Detection of H[alpha] Wavelength Images, Incorporating Automated Filament Detection and Characterization written by Jezreel Bassett and published by . This book was released on 2015 with total page 170 pages. Available in PDF, EPUB and Kindle. Book excerpt: Solar flare prediction is a valuable and sought after commodity for the safety of astronauts and satellites. Our work takes an image processing approach to solving this problem using Improved Solar Observing Optical Network (ISOON) images to produce a robust flare prediction algorithm. Our algorithm operates on images of the chromosphere which is associated with the H[alpha] wavelength. We detect solar structures of importance to flare prediction namely filaments and active regions, then extract their features. Using these features, we use a classification method known as anomaly detection which is not affected by an imbalanced dataset. Our results are highly successful when filament features are paired with features of active regions. We speculate the combination of active regions and filament features produce a catalyst to successfully predict solar flares. We also have data indicating that feature selection within anomaly detection may produce higher accuracies.