Multiple Period Searching in Astronomical Time Series with the Lomb-Scargle Periodogram

The Lomb-Scargle periodogram is one of the most used algorithms for detecting periodic components in time series data that are unequally sampled in time. In astronomy, many phenomena, such as variable stars and multiplanetary systems, exhibit behavior with multiple periodic components. This makes it very important to have techniques that allow us to extract estimations of the periods from the observations. While the Lomb-Scargle method is widely used, it is unclear how robust the results are when multiple periodicities are present or expected.

In order to assess the predictive capability, we simulated data sets that mimic the sampling patterns of astronomical observations. The comparison of the computed periodicities with the data set parameters showed that the Lomb-Scargle periodogram can easily produce misleading results. One issue is that the concept of significance is no longer well defined after prewhitening. Aliasing caused by the unequal sampling sometimes causes spurious peaks that are difficult to distinguish from true signals. Additional observational or theoretical evidence should be used to verify the periodogram results.