2010ApJ...725.1170S

Giant convection cells in the envelopes of massive red supergiants turn over in a time comparable in order of magnitude with the observed long secondary periods in these stars, according to a theory proposed some years ago by Stothers & Leung. This idea is developed further here by using improved theoretical data, especially a more accurate convective mixing length and a simple calculation of the expected radial-velocity variations at the stellar surface. The theory is applied to the two best-observed red supergiants, Betelgeuse and Antares, with more success than in the earlier study. The theory can also explain the long secondary periods seen in the low-mass red giants, thus providing a uniform and coherent picture for all of the semiregular red variables. How the turnover of a giant convection cell might account for the observed slow light and radial-velocity variations, their relative phasing, and the absence of these variations in certain stars is discussed here in a qualitative way, but follows naturally from the theory.