Reliability of Prewhitening as a Method of Multiple Period Searching in Astronomical Time Series

Many astrophysical phenomena, such as binary star systems, multiplanetary systems, variable stars, and accretion disks, exhibit behavior with multiple periodic components. Methods to extract a single periodic component from noisy data, such as the Lomb-Scargle periodogram, have been well developed. Prewhitening refers to the approach of iteratively searching for a single periodicity at a time and removing it from the time series. Although rigorous multi-frequency periodogram methods now exist, prewhitening is still commonly used and a good understanding of the method is necessary. We generated simulated time series data that mimics the sampling patterns of astronomical observations with varying numbers of periodic components. We attempted to detect these periodicities using prewhitening and the Lomb-Scargle periodogram and compared them with the data set parameters. Results showed the effects that noise and aliasing due to the sampling pattern have on accurately retrieving the period information. Although the period frequently does produce good results, additional validation from theoretical or observational evidence should be used.