Computer Vision: Discovery And Opportunity Await

Abstract:

Current solar image archives contain information, lots of information, often so much information that "end-point science" is not immediately clear at the start of a project to survey them. However, in such datasets and their metadata, significant scientific could be hidden just a few queries below the surface. We will discuss quite possibly the largest astronomical database, the "EUV Brightpoint Database", which contains information about over 200 million individual features that are ubiquitously observed in the Sun's corona - EUV Brightpoints (or BPs). While end-point science was not clear in 2002 when the project to catalog the Sun’s BPs in the archive of SOHO's Extreme Ultraviolet Telescope (EIT) images began the impact of those few queries could cause a quite a stir in the field.

Our systematic analysis of BPs, and the magnetic scale on which they appear to form, allowed us to demonstrate that the landmarks of sunspot cycle 23 could be explained in terms of the evolution and interaction of latitudinally and temporally overlapping bands of magnetic activity. Those bands appear to belong to the Sun’s 22-year magnetic activity cycle. The patterns that these bands make closely match helioseismic inference of the Sun’s torsional oscillation - a signature of rotational anomalies taking place the Sun’s interior. The high-latitude origin and start dates preceding sunspot formation by more than a decade - on the same activity band - pose a significant challenge to our understanding of the processes responsible for the production of the Sun’s quasi-decadal variability.

We sincerely doubt that the BP database is alone in containing information of such potential scientific value. Often one just has to get lucky, before being able to formulate the correct queries. We hope that the material presented in this talk can motivate a scientific exploitation of the computer vision databases currently being built from the stunning images of our star in addition to some retrospective investigations.