The Solar Dynamics Observatory (SDO) is a space telescope that provides 4k x 4k image series of different layers of the solar atmosphere, sending ~1.5 TB of data per day since 2010. I will present optical flow tracking techniques that was developed in Python to efficiently analyze the circulation of the plasma across the solar surface, and the tools that we use to visualize the plasma flows.
The surface of the sun known as the photosphere is made of buoyant parcels of plasma. Measuring these plasma motions gives us precious information on how the plasma circulates in the solar interior and how it interacts with the sun's magnetic field. This requires tracking optical flows in series of images. By providing images at higher resolution (4k x 4k) and higher cadence (one image every 45s) than even before, the SDO mission presented new image processing challenges. I will present the problems that we faced in tracking optical flows in SDO images and how we implemented efficient solutions in Python to measure the flow fields that drive the solar plasma. First, I will present tracking techniques based on "local correlation" and show its limits in analyzing motions in long time series of high resolution images. Next, I will present more efficient techniques based on the so-called "Balltracking" paradigm that I implemented in Python and Cython alongside validation strategies. Finally, I will show segmentation techniques to better extract flow patterns, combined with existing python tools to better visualize the solar flows.