Evaluating local correlation tracking using CO5BOLD simulations of solar granulation

Context. Flows on the solar surface are intimately linked to solar activity, and local correlation tracking (LCT) is one of the standard techniques for capturing the dynamics of these processes by cross-correlating solar images. However, the link between contrast variations in successive images to the underlying plasma motions has to be quantitatively confirmed. Aims. Radiation hydrodynamics simulations of solar granulation (e.g., CO5BOLD) provide access to both the wavelength-integrated, emergent continuum intensity and the three-dimensional velocity field at various heights in the solar atmosphere. Thus, applying LCT to continuum images yields horizontal proper motions, which are then compared to the velocity field of the simulated (non-magnetic) granulation. In this study, we evaluate the performance of an LCT algorithm previously developed for bulk-processing Hinode G-band images, establish it as a quantitative tool for measuring horizontal proper motions, and clearly work out the limitations of LCT or similar techniques designed to track optical flows. Methods. Horizontal flow maps and frequency distributions of the flow speed were computed for a variety of LCT input parameters including the spatial resolution, the width of the sampling window, the time cadence of successive images, and the averaging time used to determine persistent flow properties. Smoothed velocity fields from the hydrodynamics simulation at three atmospheric layers (log τ = −1, 0, and +1) served as a point of reference for the LCT results. Results. LCT recovers many of the granulation properties, e.g., the shape of the flow speed distributions, the relationship between mean flow speed and averaging time, and also – with significant smoothing of the simulated velocity field – morphological features of the flow and divergence maps. However, the horizontal proper motions are grossly underestimated by as much as a factor of three. The LCT flows match best the flows deeper in the atmosphere at log τ = +1. Conclusions. Despite the limitations of optical flow techniques, they are a valuable tool in describing horizontal proper motions on the Sun, as long as the results are not taken at face value but with a proper understanding of the input parameter space and the limitations inherent to the algorithm.