The TESS All-Sky Rotation Survey: Periods for 944,056 Stars Within 500 pc

Abstract

Stellar rotation is a fundamental tracer of stellar magnetic evolution, age, and activity, with broad implications for Galactic archaeology and exoplanet characterization. The Transiting Exoplanet Survey Satellite (TESS) provides high-precision time-series photometry across the sky, enabling rotation measurements for an unprecedented number of stars. We present the TESS All-Sky Rotation Survey (TARS), an all-sky catalog of stellar variability periods for 944,056 stars with T < 16 and distances within 500 pc. We estimate that 94% of these periods are rotation periods. This catalog increases the number of rotation period measurements for stars with T < 16 within 100 pc by a factor of 2.1 and within 500 pc by 3.7. We also present a method to correct half-period aliases in TESS data and show that it reliably recovers periods as long as 25 days from a single TESS sector. TARS represents the largest homogeneous catalog of stellar rotation periods to date, providing a foundation for studies of stellar ages, young associations, and Galactic structure. We make the light curves used in our analysis available as a HLSP through MAST. Beyond the default TARS catalog, we provide code that allows users to generate rotation period catalogs with adjustable completeness and reliability thresholds. This code and all rotation period measurements are available through Zenodo.

Figures (16)

• Figure 1: Selection function.Top left: Sky coverage. We used TESS images from 2018 July--2025 September. Top right: Histogram of number of TESS observations per star; numbers above each bin give the star count. The median star has four sectors of TESS data. Lower left: The TESS magnitude $T$ as a function of the distance to each star in our survey, colored by the median de-reddened Gaia DR3 $G_{\rm BP} - G_{\rm RP}$ color in each bin. Spectral types are approximated from 2013ApJS..208....9P. Lower right: Histograms across our sample of magnitude, distance, temperature, and color.
• Figure 2: Example vetting plot. Probabilities from our systematics and alias classifiers are listed. One such vetting plot is available for each of the 39,061,674 light curves through the TARS High-Level Science Product at MAST. An interactive vetting plot explorer is also https://lgbouma.com/tars_viz/.
• Figure 3: Systematics masquerade as period detections.Top: The period of the highest peak in the Lomb-Scargle periodogram across all TESS sectors. The sinusoidal structures across sectors are aliases of the TESS data gap. Bottom left: Our systematics classifier identifies and separates systematics from other rotation signals. The systematics classifier is trained so rotation measurements with a high probability are more likely to be real signals and less likely to be a systematic. Sector 6 experienced momentum dumps every 3.125 days. The blue and gray stars show the Lomb-Scargle period and power of the two light curves in the bottom right panel. Bottom right: TIC 231925886 and TIC 302869120 both have Lomb-Scargle peak periods of $\sim2.9$ days in Sector 6, but TIC 231925886 has a systematics classifier score of 1.0 (high probability of being a rotation signal) while TIC 302869120 has a systematics classifier score of 0.0 (high probability of being a systematic).
• Figure 4: Systematics classifier performance vs. rotation period.Top left: Reliability, defined as the fraction of random forest-selected signals that yield rotation periods consistent across $>$$5$ sectors, shown for several random forest probability thresholds. The classifier returns reliable detections across the full period range. Top right: Completeness, defined as the fraction of all signals in the class that exceed the random forest probability threshold cuts. Completeness remains high for $P_{\rm rot}\lesssim 10$ days, but decreases at longer periods. These panels suggest that the systematics classifier is reliable at longer periods, however the yield of such measurements declines beyond $\sim$10--13 days. Bottom left: The reliability of the negative class as a function of $P(\rm systematic)$. Bottom right: The completeness of the negative class as a function of $P(\rm systematic)$.
• Figure 5: Aliases can be identified at the cost of completeness.Top left: Reliability, defined here as the fraction of retained measurements corresponding to the true rotation period rather than an alias, remains uniformly high across the period range probed by TESS. Our ground truth periods are taken from mcquillanROTATIONPERIODS342014. Top right: This gain in reliability comes at the expense of completeness, which decreases steadily for more stringent cuts. Bottom left: The random forest classifier likewise identifies aliases robustly across period space, with high reliability indicating that most measurements flagged as aliases are indeed aliases. Bottom right: As for the true-period selection, imposing a high probability threshold for alias identification reduces completeness, removing a large fraction of potential alias detections.