Tracking Solar Events for Information Discovery using SDO/AIA Images

Providing situational knowledge of the real-time space weather environment requires the ability to understand, model, and predict space weather events. To accomplish this, we must be able to analyze a large quantity of data from many sources, and one such source is the Solar Dynamics Observatory (SDO). Prior to the launch of the SDO mission, an international consortium of scientists, called the Feature Finding Team (FFT), started to develop software modules to process the SDO's Atmospheric Imaging Assembly (AIA) images. These modules report meta-data about various solar phenomena that are of interest to the solar physics community to a central repository. In order to discover spatiotemporal relationships between various types of solar events, these individual reports need to be linked/tracked to create not just a detection report but a trajectory of an event. These trajectories have lifetimes and motion information that is not captured from an individual detection report and can provide meaningful information for modeling activity and predicting sequences of events. Our work focus on using the meta-data reported from the FFT modules and turning it into spatiotemporal trajectories. We utilize SDO's AIA images to do visual comparisons of multiple solar event reports to determine the most likely trajectory of an event when more than one possible trajectory arises, meaning when multiple objects are present. This process of tracking multiple objects in video data is commonly referred to as multiple hypothesis tracking, and is a difficult and only partially solved problem in computer vision. We utilize data covering the three years January 1, 2012 through December 31, 2014. Our goal is to provide a trajectory dataset for exploration to find meaningful spatiotemporal relationships which can then be used for better understanding and modeling of solar events.