We examine the consequences for the frequency analysis of δ Scuti stars of using pre-whitening if the assumptions of stationarity and sinusoidal light curves were violated. We show through numerical simulation that if the assumption of stationarity is violated then very large numbers of spurious frequencies may be generated which span the entire frequency range of the analysis. We also show that if the light curve is asymmetric, even when regularly spaced, highly significant but entirely spurious frequencies are generated. We apply the results of the numerical simulations to the δ Scuti HD 39641 and show that many statistically significant frequencies are artifacts of the data analysis process. Finally, we propose a method we call restricted range frequency analysis which uses existing tools, but aims to curb the worst features of frequency analysis when assumption violations are present in the light curve.
展开▼
