Temporal filters for isolating steady photospheric flows

Abstract

A variety of temporal filters are tested on artificial data with 60 and 75 s sampling intervals to determine their accuracy in separating the nearly-steady photospheric flows from the p-mode oscillations in Doppler velocity data. Longer temporal averages are better at reducing the residual signal due to p-modes but they introduce additional errors from the rotation of the supergranule pattern across the solar disk. Unweighted filters (boxcar averages) leave residual r.m.s. errors of about 6 m s⁻¹ from the p-modes after 60 min of averaging. Weighted filters, with nearly Gaussian shapes, leave similar residual errors after only 20 min of averaging and introduce smaller errors from the rotation of the supergranule pattern. The best filters found are weighted filters that use data separated by 150 or 120 s so that the p-modes are sampled at opposite phases. These filters achieve an optimum error level after about 20 min, with the r.m.s. errors due to the p-mode oscillations and the rotation of the supergranules both at a level of only 1.5 m s⁻¹.