TESS Time-Series Observations

• TESS time-series observations are high-precision, continuous photometric measurements covering nearly the entire sky, enabling detailed studies of exoplanets and variable stars.
• They employ multiple cadences and sophisticated pipelines—including PCA and machine learning—to extract reliable light curves even in crowded fields.
• The data facilitate ensemble asteroseismology, precise transit modeling, and comprehensive analysis of stellar variability and exoplanet atmospheres.

The Transiting Exoplanet Survey Satellite (TESS) time-series observations constitute a foundational resource for contemporary time-domain astrophysics, providing high-precision, continuous photometric measurements across nearly the entire sky. Designed primarily to enable the detection of exoplanets via the transit method, TESS has also transformed the paper of variable stars, stellar pulsations, binary systems, stellar rotation, and diverse classes of intrinsic and extrinsic variability. Its data, accessible at multiple cadences and complemented by sophisticated analysis pipelines and advanced machine learning techniques, underpin a wide spectrum of research from ensemble asteroseismology to exoplanetary atmosphere characterization.

1. Instrumentation, Observing Strategy, and Data Products

TESS uses four wide-field cameras, each with a 24° × 24° field-of-view, enabling near all-sky coverage in ~27-day sectors. Photometric data are collected at several temporal cadences: standard full-frame images (FFIs) have a 30-minute time resolution in the primary mission, with subsequent mission extensions allowing 10-minute and even 20-second cadences for select targets. Targeted pre-selected stars are monitored at 2-minute or shorter intervals. The combination of long, nearly uninterrupted temporal coverage per sector (particularly near the ecliptic poles, where stars can be observed for up to a year), and the high photometric precision achievable from space, permits the analysis of variability phenomena on time scales from seconds to weeks.

The TESS data pipeline, including the Science Processing Operations Center (SPOC) products, delivers calibrated target pixel files, light curve files from simple aperture photometry (SAP), presearch data conditioning (PDC)-corrected flux series, and cotrending basis vectors for systematic mitigation. These are supplemented by specialized open-source extraction tools (e.g., eleanor (Feinstein et al., 2019)) optimized for the more complex systematics and blending present in TESS FFIs.

2. Methodologies for Light Curve Extraction and Analysis

Data reduction for TESS time-series photometry involves rigorous pre-processing to handle issues unique to its large pixel scale and background structure. Pipelines like eleanor implement astrometric registration, local background subtraction (using both empirical formulas and Principal Component Analysis), and a suite of photometric extraction options (aperture and PSF-based), subsequently decorrelating instrumental systematics using centroid and background regressors and SPOC’s cotrending basis vectors (Feinstein et al., 2019, Caldwell et al., 2020).

For crowded fields and cluster science, difference imaging coupled with forced photometry at Gaia-registered positions enables extraction of robust light curves even under significant blending (e.g., CDIPS pipeline (Bouma et al., 2019)). Systematic trends (“red noise”) are removed with ensemble-based detrending methods such as the Trend Filtering Algorithm (TFA) and PCA. High-cadence data are essential for detecting short-period oscillators (e.g., δ Scuti, roAp stars), while the long, contiguous windows maximize sensitivity to episodic and quasi-periodic variability.

Analytical and machine learning methods are employed for event detection, classification, and parameter estimation:

• Transit events are identified using phase folding, Box Least Squares (BLS), and detailed model fits (e.g., Mandel & Agol formalism), with noise modeling through wavelet denoising and Gaussian Processes for correlated noise (Saha, 2023).
• Machine learning classifiers based on convolutional neural networks (e.g., Astronet-Triage-v2 (Tey et al., 2023)) leverage multi-view representations of phase-folded light curves and ensemble training for efficient transit candidate triage across massive FFI datasets, achieving recall rates up to 99.6% at precision of 75.7%, and robust generalization across mission phases.
• Time series analysis of variable stars utilizes Fourier transforms, periodograms, and autocorrelation functions to recover oscillation frequencies, rotation periods, and complex pulsation/mode structures.

3. Application Domains: Variable Star Research and Exoplanet Science

TESS’s time-series datasets support studies across a broad portfolio of variable phenomena:

Variable Type	Observational Signature	Analysis Approaches
Eclipsing binaries	Periodic primary/secondary eclipses	Phase folding, Fourier, ML triage
Transiting exoplanets	Shallow, periodic flux dips (transits)	Transit modeling, ML triage, MCMC
Pulsating stars (RR Lyrae, δ Scuti, γ Doradus, roAp, β Cephei, SPB, WD/subdwarf pulsators)	Multi-periodic variability (p/g-modes)	Fourier analysis, asteroseismology
Rotational variables	Quasi-periodic spot-induced modulations	Sector-by-sector analysis, periodograms
Eruptive/cataclysmic	Stochastic flares, outbursts	High-cadence event search, statistics

For exoplanet science, TESS time-series observations enable high-SNR transit detection and precise refinement of planetary ephemerides, with follow-up strategies addressing rapid ephemeris deterioration inherent to TESS’s observational cadence (Dragomir et al., 2019, Ivshina et al., 2022). Periodic re-observation and community-organized follow-up campaigns are necessary to maintain scheduling viability for future facilities (JWST, ELTs, LUVOIR).

Machine learning classifiers (Astronet-Triage-v2) have demonstrably improved the recovery of TESS Objects of Interest (TOIs), saving at least 200 additional planet candidates at fixed precision relative to predecessor models, and are integrated into automated workflows for large-scale planet candidate vetting (Tey et al., 2023).

Research in pulsating and variable stars has been significantly advanced due to TESS’s all-sky coverage and uniform cadence. Ensemble asteroseismology—enabled by extensive uniform datasets—has revealed statistical properties of mode selection, instability strips, and the influence of rotation and stellar evolution (see studies of the Pleiades cluster (Bedding et al., 2022)). High-fidelity characterization of RR Lyrae modulation, discovery of pulsating compact objects, and population-scale surveys have all been facilitated.

4. Statistical and Physical Parameter Inference

Bayesian and frequentist approaches underpin virtually all physical inference from TESS time-series data:

• Stellar parameters (e.g., $T_{\rm eff}$, $\log g$, [Fe/H], $v \sin i$, microturbulence) are derived via joint spectral and photometric modeling, incorporating isochrone grids and priors from population synthesis models (Sharma et al., 2017).
• Planetary parameters and transit ephemerides are refined through MCMC sampling of transit models, often augmented by Gaussian Process noise modeling to account for time-correlated systematics (Saha, 2023).
• Timing verification of the TESS photometric series against ground-based standards has established timing accuracy at the $\sim$5 s level, with no significant long-term drift detected (Essen et al., 2020).

The use of diagnostic indicators (e.g., large frequency separation $\Delta\nu$, frequency of maximum oscillation power $\nu_{\rm max}$) enables seismic modeling of stellar interiors. TESS time-series have also been critical in calibrating the relationship between photometric variability and radial velocity jitter, with implications for target selection in RV planet searches (Hojjatpanah et al., 2020).

5. Complementarity, Verification, and Cross-Instrument Synergies

TESS time-series observations are frequently combined with contemporary and archival datasets for improved astrophysical interpretation:

• Simultaneous TESS and Kepler/K2 observations over common sky regions have allowed stringent cross-calibration of photometric precision, systematics, and timing (Barclay et al., 2018).
• Ground-based photometric surveys (e.g., ZTF) offer multi-filter, high-cadence coverage that supplements TESS’s broad bandpass and lower spatial resolution, supporting source deblending and color-based variability analysis (Roestel et al., 2019).
• Spectroscopic follow-up (e.g., LAMOST, HERMES) enables dynamical mass determination in compact binaries, calibration of stellar atmospheres, and physical companion characterization (Zheng et al., 2022, Sharma et al., 2017).

Machine learning methods (CNNs, LSTMs, transformers) are increasingly employed for automated classification, outlier identification, and anomaly detection, which is indispensable in handling the survey’s scale.

6. Impact, Limitations, and Future Prospects

TESS time-series photometry has become indispensable for both population-scale and individual-object time-domain astrophysics (Bognár et al., 8 Sep 2025). The high data quality facilitates ensemble studies of stellar rotation, pulsation, and binarity, contributing to gyrochronology, asteroseismic modeling, and binary evolution theory. The breadth of variable phenomena accessible—from flares on M dwarfs to compact white dwarf and subdwarf oscillators—demonstrates TESS’s versatility.

Limitations include challenges in crowded fields arising from large pixel sizes, complexities in detrending instrumental systematics, and the relatively short baseline for most sky sectors (limiting sensitivity to the longest-period variables and secular phenomena). The need for ephemeris maintenance and targeted follow-up is acute for planning next-generation exoplanet atmospheric studies.

The combination of exhaustive time-series datasets, sophisticated extraction and analysis methodologies, and integration with complementary surveys and machine learning frameworks positions TESS as a cornerstone of time-domain astrophysics. As the mission enters its second extended phase, prospects include deeper ensemble asteroseismology, further expansion of exoplanet inventories, and new discoveries in the variable star and transient domain.