Following the eruption of a filament from a flaring active region, sunward-flowing voids are often seen above developing post-eruption arcades. First discovered using the soft X-ray telescope aboard Yohkoh, these supra-arcade downflows (SADs) are now
an expected observation of extreme ultra-violet (EUV) and soft X-ray coronal imagers and spectrographs (e.g, TRACE, SOHO/SUMER, Hinode/XRT, SDO/AIA). Observations made prior to the operation of AIA suggested that these plasma voids (which are seen in contrast to bright, high-temperature plasma associated with current sheets) are the cross-sections of evacuated flux tubes retracting from reconnection sites high in the corona. The high temperature imaging afforded by AIAs 131, 94, and 193 Angstrom channels coupled with the fast temporal cadence allows for unprecedented scrutiny of the voids. For a flare occurring on 2011 October 22, we provide evidence suggesting that SADs, instead of being the cross-sections of relatively large, evacuated flux tubes, are actually wakes (i.e., trailing regions of low density) created by the retraction of much thinner tubes. This re-interpretation is a significant shift in the fundamental understanding of SADs, as the features once thought to be identifiable as the shrinking loops themselves now appear to be side effects of the passage of the loops through the supra-arcade plasma. In light of the fact that previous measurements have attributed to the shrinking loops characteristics that may instead belong to their wakes, we discuss the implications of this new interpretation on previous parameter estimations, and on reconnection theory.
The frequency distributions of sizes and fluxes of supra-arcade downflows (SADs) provide information about the process of their creation. For example, a fractal creation process may be expected to yield a power-law distribution of sizes and/or fluxes
. We examine 120 cross-sectional areas and magnetic flux estimates found by Savage & McKenzie for SADs, and find that (1) the areas are consistent with a log-normal distribution and (2) the fluxes are consistent with both a log-normal and an exponential distribution. Neither set of measurements is compatible with a power-law distribution nor a normal distribution. As a demonstration of the applicability of these findings to improved understanding of reconnection, we consider a simple SAD growth scenario with minimal assumptions, capable of producing a log-normal distribution.
