Abstract
It is known that stellar differential rotation can be detected by analyzing the Fourier transform of spectral line profiles, since the ratio of the first and second zero frequencies is a useful indicator. This approach essentially relies on the conventional formulation that the observed flux profile is expressible as a convolution of the rotational broadening function and the intrinsic profile, which implicitly assumes that the local intensity profile does not change over the disk. Although this postulation is unrealistic in the strict sense, how the result is affected by this approximation is still unclear. With the aim of examining this problem, flux profiles of several test lines (showing different center-to-limb variations) were simulated using a model atmosphere corresponding to a mid-F dwarf by integrating the intensity profiles for various combinations of vesin i (projected rotational velocity), α (differential degree), and i (inclination angle), and their Fourier transforms were computed to check whether the zeros are detected at the predicted positions or not. For this comparison a large grid of standard rotational broadening functions and their transforms/zeros were also calculated. It turned out that the situation depends critically on vesin i: In the case of vesin i ≳ 20 km s−1, where rotational broadening is predominant over other line broadening velocities (typically several km s−1), the first/second zeros of the transform are confirmed almost at the expected positions. In contrast, deviations begin to appear as vesin i is lowered, and the zero features of the transform are totally different from those expected at vesin i as low as ∼10 km s−1, which means that the classical formulation is no longer valid. Accordingly, while the zero-frequency approach is safely applicable to studying differential rotation in the former broader-line case, it would be difficult to practice for the latter sharp-line case.