2022A&A...666A.142S

Context. In the time of large space surveys that provide tremendous amounts of precise data, it is highly desirable to have a commonly accepted methodology and system for the classification of variable stars. This is especially important for A-F stars, which can show intrinsic brightness variations due to both rotation and pulsations.
Aims. The goal of our study is to provide a reliable classification of the variability of A-F stars brighter than 11 mag located in the northern TESS continuous viewing zone. We also aim to provide a thorough discussion about issues in the classification related to data characteristics and the issues arising from the similar light-curve shape generated by different physical mechanisms.
Methods. We used TESS long- and short-cadence photometric data and corresponding Fourier transform to classify the variability type of the stars. We also used spectroscopic observations to determine the projected rotational velocity of a few stars.
Results. We present a clear and concise classification system that is demonstrated on many examples. We find clear signs of variability in 3025 of 5923 studied stars (51%). For 1813 of these 3025 stars, we provide a classification; the rest cannot be unambiguously classified. Of the classified stars, 64.5% are pulsating stars of g-mode γ Doradus (GDOR) and p-mode δ Scuti types and their hybrids. We realised that the long- and short-cadence pre-search data conditioning simple aperture photometry data can differ significantly not only in amplitude but also in the content of instrumental and data-reduction artefacts, making the long-cadence data less reliable. We identified a new group of stars that show stable light curves and characteristic frequency spectrum patterns (8.5% of the classified stars). According to the position in the Hertzsprung-Russell diagram, these stars are likely GDOR stars but are on average about 200 K cooler than GDORs and have smaller amplitudes and longer periods. With the help of spectroscopic measurements of v sin i, we show that the variability of stars with unresolved groups of peaks located close to the positions of the harmonics in their frequency spectra (16% of the classified stars) can be caused by rotation rather than by pulsations. We show that without spectroscopic observations it can be impossible to unambiguously distinguish between ellipsoidal variability and rotational variability. We also applied our methodology to three previous studies and find significant discrepancies in the classification.
Conclusions. We demonstrate how difficult the classification of variable A-F stars can be when using only photometric data, how the residual artefacts can produce false positives, and that some types cannot actually be distinguished without spectroscopic observations. Our analysis provides collections that can be used as training samples for automatic classification.