Classification for 969 double-mode RR Lyrae stars from Zwicky Transient Facility

Abstract: RR Lyrae (RRL) variable stars are cornerstone distance indicators. In particular, double-mode RR Lyrae (RRd) stars enable period--luminosity relations (PLRs) that are less sensitive to metallicity, reducing systematic biases in distance measurements. However, their utility has been limited by a global sample of only $\sim$3,000 objects. We develop an automated RRd-screening pipeline and apply it to a cross-matched sample between the Gaia DR3 RRL catalog and ZTF DR22 time-series photometry. The workflow combines Lomb--Scargle period searches, iterative pre-whitening, period-ratio constraints that suppress $\sim$1-day sampling aliases, and amplitude-based quality cuts, enabling large-scale RRd star screening. We produce two ZTF-based catalogs: (i) 39,322 reliable single-mode RRL (40.5\% of the cross-matched set) and (ii) 969 RRd stars. Among the RRd stars, 614 objects are newly identified, substantially enlarging this previously scarce sample; the catalog achieves an estimated completeness of 47.7\%. The PLR derived from the newly discovered RRd stars agrees with the LMC-based relation, though with larger uncertainties. Incorporating these stars will help tighten the RRd PLR and improve distance measurements. Looking ahead, systematic RRd searches with upcoming surveys such as the Legacy Survey of Space and Time (LSST) and the China Space Station Telescope (CSST) should further extend high-accuracy distances across the Local Group and strengthen their cosmological applications.

• The paper presents an automated pipeline that identifies 969 RRd stars, including 614 new discoveries, enhancing the global RRd census.
• It employs Lomb–Scargle period search with Fourier pre-whitening to robustly extract dual pulsation modes from high-quality ZTF photometry.
• The calibrated period–luminosity relation, consistent with LMC standards, reinforces the metallicity-insensitive use of RRd stars in cosmic distance scaling.

Automated Classification of 969 Double-Mode RR Lyrae Stars from ZTF

Scientific Context and Motivation

RR Lyrae variables are critical Population II standard candles for the local cosmic distance scale. While RRab (fundamental mode) and RRc (first-overtone) stars are prevalent, double-mode RR Lyrae (RRd) variables pulsate simultaneously in both modes. The unique property of RRd stars—the metallicity sensitivity of their period ratios—enables construction of period–luminosity relations (PLRs) less affected by [Fe/H] systematics, enhancing the reliability of extragalactic and Galactic distance mapping. Historically, the utility of RRd stars has been hampered by incomplete and inhomogeneous samples (≈2000–3000 global counts), extracted via manual or inefficient semi-automated light curve mining across wide-field time-domain surveys.

This study addresses these limitations by developing and deploying an automated, systematic RRd classification pipeline, utilizing cross-matched Zwicky Transient Facility (ZTF) DR22 time-series photometry and Gaia DR3 RR Lyrae catalogs, and presents an expanded RRd sample of 969 stars—614 of which are newly identified—substantially increasing the global census of confirmed double-mode RR Lyrae stars (2510.07705).

Data Set Construction and Preprocessing

The parent sample is obtained by positional cross-matching of 271,779 Gaia DR3 RR Lyrae candidates with ZTF DR22 photometric records, yielding 97,146 stars with sufficient ZTF $r$-band epochs (declination-limited by ZTF coverage). The workflow includes advanced time-series cleaning: median aggregation of closely clustered exposures, stringent catflag-based quality filters, $3\sigma$ outlier removal from model fits, and a lower threshold of 40 high-fidelity photometric points per source. These steps are crucial for suppressing spurious artifact periods from ground-based cadence and instrument systematics.

Period Search and RRd Identification Pipeline

The core period analysis utilizes the Lomb–Scargle (LS) method, applying a global peak search constrained to physically plausible RRL period domains ($0.2 < P_1 < 1$ d), followed by pre-whitening with a 10th-order Fourier fit to extract residuals. Secondary LS analysis identifies candidate double-mode (RRd) objects with a significant (FAP $< 10^{-5}$) second period, conditioned on period ratios within $0.72 < P_{1\mathrm{O}}/P_F < 0.75$—the canonical RRd Petersen sequence window.

To further suppress ground-based systematics, especially the ZTF $\sim$1-day aliasing, the pipeline excludes candidates with primary/secondary periods near integer frequencies or period sum/difference spacing commensurate with daily sampling harmonics. Secondary period amplitude criteria are dynamically set relative to photometric noise, with the main RRd locus (0.742–0.748) permitting marginal detections to increase completeness without compromising purity. Only sources with fundamental mode periods $0.4 < P_F < 0.6$ d are considered, tightly linking the output catalog to physically meaningful RRd candidates.

The search for triply-periodic variables, conducted via iterative pre-whitening and further LS analysis, yields no robust third-mode detections, in contrast to the more complex pulsational instability seen in $\delta$ Scuti stars [2025ApJ...984...89J]. This aligns with the observed low amplitudes ($<0.1$ mag) of secondary modes in the RRd sample.

Results: Catalogs, Completeness, and Performance

The final ZTF-based variable catalog comprises 39,322 RR Lyrae (≈40% of the cleaned parent set), within which 969 RRd are robustly identified—614 previously uncatalogued (≈63%). The completeness of RRd classification, benchmarked against Gaia DR3 RRd labels (restricted to stars with ZTF photometry), is 47.7%, surpassing the RRL overall completeness (40%). The mismatch is primarily attributed to ZTF’s sparse time sampling and SNR limitations in faint RRLs, with additional Gaia misclassifications flagged where double-mode status cannot be confirmed by ZTF or ASAS-SN.

The period–amplitude and Petersen diagrams reveal that RRd stars with first-overtone dominance (by period and amplitude assignment) trace the canonical RRd sequence with tight clustering in both projections, while fundamental-mode-dominant RRd candidates are both fewer (5%) and less coherent, likely representing a mix of anomalous, blended, or borderline cases. Newly identified RRd stars preferentially populate the low-amplitude, low-metallicity ends of these distributions, reflecting improved sensitivity and search systematics relative to previous studies.

RRd Period–Luminosity Relation Calibration

Applying Gaia parallaxes (with zero-point corrections) and Wesenheit magnitude construction incorporates 90 high-quality RRd stars into a calibration of the PLR: $$M_{W_{G, BP, RP}} = (-2.310\pm0.584)\, (\log P_{\mathrm{F}}+0.35) + (-0.091\pm0.040), \quad \sigma=0.428\,\mathrm{mag}$$ This PLR is consistent in zero point (difference $0.08\pm0.11$ mag) with LMC-based RRd relations, with scatter driven by parallax uncertainties, residual Gaia biases, and the larger distances of the Galactic sample. This result reinforces the metallicity-insensitivity claim for RRd PLRs, promising tighter calibration with future, larger samples and improved parallax systematics [2023NatAs...7.1081C].

Implications, Limitations, and Prospects

This catalog represents the largest, homogeneous, systematically classified RRd sample assembled from ground-based time-domain data, with direct implications for improving Galactic structure mapping and the extragalactic distance ladder. The enhanced census will underpin refined, metallicity-insensitive PLRs, crucial for resolving systematics in the Population II-based extragalactic scale and Hubble constant determinations [2016ApJ...832..210B]. The robust, reproducible pipeline design exposes the dual limitations of time-domain survey cadence and photometric precision, but also demonstrates the efficacy of period ratio and amplitude diagnostics as complementary criteria.

With upcoming surveys such as LSST and CSST poised to provide deeper, higher-cadence, and more uniform variability data, future versions of this pipeline anticipate an order-of-magnitude increase in high-confidence RRd identifications. This extension will allow more statistically secure studies of RRd incidence, period–metallicity relations, and their connections to stellar evolution models, as well as new constraints on the three-dimensional structure of the Milky Way and Local Group galaxies.

Conclusion

This study introduces a rigorous, automated pipeline for RRd classification, applied to the combined Gaia DR3 and ZTF DR22 data, leading to the identification of 969 double-mode RR Lyrae stars, more than half previously unknown (2510.07705). The period–luminosity calibration for the expanded RRd sample affirms consistency with LMC standards and emphasizes the metallicity independence of the RRd PLR. The approach and results have direct implications for future time-domain surveys and cosmological distance scale anchoring, while highlighting the remaining limitations from survey cadence and parallactic calibration.