Development and Demonstration of Computer Vision Software for Solar Transient Events, 15-R8390
Craig E. DeForest
Inclusive Dates: 04/01/13 – Current
Background — Solar corona exists at a million degree Kelvin (K) above the photosphere, the visible surface of the Sun, which is only 6,000 K. This apparently paradoxical phenomenon, known as coronal heating, is one of the most controversial topics in solar physics. Magnetic reconnection, manifesting as small- and large-scale explosive and eruptive phenomena, is the most viable mechanism for coronal heating. The role of small-scale events such as microflares in coronal heating is poorly understood mainly because of a lack of an unbiased tool for characterizing large populations and a lack of observations across the wide temperature regimes of the corona. The Atmospheric Imaging Assembly (AIA) telescope onboard the Solar Dynamics Observatory (SDO) observes the solar atmosphere in visible (4500 Â), ultraviolet (1600 and 1700 Â), and extreme ultraviolet (94, 131, 171, 193, 211, 304, and 335 Â) wavelengths, covering temperature regimes from 5,000 K to 20,000,000 K. Studying microevents in multiple wavelengths will reveal their evolution in these temperature regimes and clarify the physics.
Approach — The project goal is to develop a computer vision code, Automated Microevent-finding Code (AMC), for automatically detecting and characterizing thousands of microevents and transient features in the AIA images without human bias. AMC will be used for a proof-of-concept study addressing a 60-year-old mystery central to solar physics: Can microflares account for the high temperature of the solar corona? AMC will have other applications in solar physics (e. g., implications of small-scale eruptive events such as miniature CMEs, jets, and plumes) and in planetary sciences (for example, the turbulent eddies in the Jovian atmosphere).
To develop AMC, the SWAMIS vision code for magnetic feature tracking is being modified. However, identifying transient brightenings requires a set of selection criteria and threshold values completely different from those for magnetic feature tracking. Also, discriminating the events from other brightenings and dynamics of the corona is a challenge, as the AIA images are high resolution and high cadence. The microevents detected in each wavelength will be associated in multiple wavelengths.
Accomplishments — A test-version of AMC was developed that is capable of identifying the microevents in AIA 1600 Â data and associating them in multiple frames of the time series. A sample of the identified features is shown in the illustration. The image shown is a small slice of the original image of size 4096 x 4096 pixels. Here, the small, irregular bright spots are the microevents detected using the AMC. Thirteen frames were used to define and detect these features.