Tracking algorithms and machine learning for the characterization of active regions over the solar cycle 24

Since the year 2010, SDO is sending more than a terabyte of solar observations per day.

By offering such an unprecedented large and varied data sets, this mission has propelled the solar physics community into the era of “Big Data” analytics. As an answer to this new technical and scientific challenge, we present here a threefold innovative framework for efficient data mining and analysis of the solar photosphere using SDO/HMI:

(i) A method for tracking the horizontal photospheric flows uses an improved version of “Balltracking”. We will present the most recent version of this feature tracking algorithm, its advantage over other more traditional methods like Local Correlation Tracking (LCT) and how it has been specifically tuned to handle the massive HMI datastream. Coupled with flow segmentation algorithms, it offers an unprecedented view of the evolution of the supergranulation.

(ii) A method for tracking the magnetic flux using HMI data called “Magnetic Balltracking”. We will show how it enables us to accurately track magnetic elements on magnetograms in the Lagrange reference frame, and systematically derive parameters such as the position, velocity, and fragments area and how we use it to automate the detection of flux emergence.

(iii) The above methods define a tracking framework whose output feed databases that become the input of machine learning algorithms for classification purposes. We will show how this expands our knowledge-base e.g. on the properties of large-scale photospheric flows prior to and after the emergence of active regions, and on how the flows interact with the magnetic field over large areas and long time scales.

Through these examples we will demonstrate how this framework contributes to a sensible characterization of the evolution of active regions during the whole solar cycle.