Kepler IRIS Light Curves

• Kepler IRIS Light Curves are high-precision photometric time series derived from Kepler’s superstamp imaging using advanced image subtraction.
• The methodology employs techniques like bilinear interpolation and ensemble/individual centroiding to counteract drift and systematic errors in crowded fields.
• The extensive IRIS catalog enables detailed studies of stellar rotation, variability, and asteroseismology in clusters such as NGC 6791 and NGC 6819.

Kepler IRIS Light Curves are high-precision photometric time series derived from Kepler’s superstamp imaging of crowded stellar fields such as open clusters. Unlike standard Kepler pipeline products, which were limited to individually targeted sources, IRIS light curves are generated for every star falling within the 200×200 pixel superstamps that covered clusters like NGC 6791 and NGC 6819. IRIS leverages image subtraction photometry and advanced algorithms to extract robust light curves even for faint, blended, or previously untargeted objects. The availability of IRIS data dramatically expands the sample size and completeness for variability studies in crowded fields, enabling fundamental advances in stellar rotation, variability, asteroseismology, and time-domain astrophysics.

1. Technical Foundations of Kepler IRIS Photometry

Kepler IRIS light curves are constructed using image subtraction photometry rather than classical aperture techniques (Colman et al., 2021). The data reduction begins with extraction of small (typically 5×5 pixel) postage stamps around each catalog source in the superstamps. For each frame, the stellar centroid is measured using a first-moment centroid calculation:

• $$x_{\mathrm{centroid}} = \frac{\Sigma x_i\, f_i}{\Sigma f_i}$$, $$y_{\mathrm{centroid}} = \frac{\Sigma y_i\, f_i}{\Sigma f_i}$$

Frames are upsampled via bilinear interpolation to better handle sub-pixel drift and pointing variations. The series is then aligned to a common centroid position, and a mean reference image is calculated from the stack of aligned frames. The reference is subtracted from each frame, yielding residual images that isolate variability.

A weighted aperture (typically Gaussian and isolation-parameter scaled), tuned for degree of crowding, is applied to the difference images. This approach reduces the impact of blending, systematic drifts, and background variations more effectively than traditional aperture photometry.

2. Photometric Precision and Systematic Effects

The IRIS methodology yields time series with precision approaching that of the standard Kepler pipeline for isolated stars, and in many instances, it improves detectability for faint or blended sources within the cluster field (Colman et al., 2021). The catalog encompasses light curves for 9,150 targets—over 8,400 of which were previously unobserved photometrically by Kepler.

Temporal systematics, such as pointing drift, thermal-induced focus variation, and inter-quarter detector changes, are mitigated through ensemble centroiding (EC-IRIS) for groups of stars and individual centroiding (TC-IRIS) for brighter, isolated targets. Comparisons of signal-to-noise and noise statistics across the catalog demonstrate that, in cases of severe crowding, the difference imaging and upsampling strategies yield a higher fidelity recovery of genuine variability.

Table: IRIS Photometric Workflow

Step	Method	Purpose
Postage stamps extraction	5×5 pixel stamp centered on catalog source	Source localization
Centroid alignment	First-moment, bilinear interpolation, ensemble/individual centroids	Drift correction
Reference construction	Mean image from aligned frames	Background subtraction
Difference imaging	Subtract reference from each frame	Isolate variability
Weighted aperture	Gaussian mask, scaled by crowding/isolation	Flux extraction

3. Catalogue Scope and Data Products

The IRIS catalog covers 5,342 objects in NGC 6791 and 3,808 in NGC 6819, spanning 17 and 14 quarters of data, respectively. Approximately 91–93% of these stars have continuous coverage across all quarters. Both raw (quarter-by-quarter) and corrected (fully merged and de-trended) light curves are provided as high-level science products via MAST (Colman et al., 2021). Ancillary products include systematics metrics, noise quantifications, and variable star flagging for catalog-level analysis.

For ~382 stars, the IRIS catalog increases the number of available quarters compared to pipeline data, enhancing baseline coverage and enabling studies of long time-scale phenomena such as rotational modulation, spot evolution, and period-drift.

4. Applications in Crowded-Field and Cluster Astrophysics

The IRIS light curves open avenues for cluster population studies that were hitherto inaccessible owing to crowding. Detailed investigations using IRIS data have:

• Measured over 271 robust rotation periods in NGC 6819 (an order of magnitude more than prior studies, which observed only ~30 stars), showing the expected gyrochronological spin-down and revealing a bimodal distribution of fast and slow rotators (Sagynbayeva et al., 2 Oct 2025).
• Identified a “pile-up” of stars at near-constant rotation periods across a range of temperatures, suggesting weakened magnetic braking above a critical Rossby number—directly informing models of angular momentum evolution and gyrochronology at intermediate-to-old cluster ages (Sagynbayeva et al., 2 Oct 2025).

Detection of eclipsing binaries, pulsators, and low-amplitude variability is enhanced, as image subtraction methods are less susceptible to systematics from crowding and background variation.

5. Modeling and Statistical Analysis of IRIS Time Series

Advanced detrending and rotational period determination in IRIS light curves are accomplished by constructing design matrices of common basis vectors via singular value decomposition (SVD) of whitened light curves. Systematic trends are projected out with models such as:

$$\mathbf{y}_\bot = (I - M (M^T \Sigma^{-1} M)^{-1} M^T \Sigma^{-1})\, \mathbf{y}$$

where $M$ is the matrix of basis vectors and $\Sigma^{-1}$ is the inverse noise variance.

Rotation periods are determined using Gaussian Process (GP) modeling. The “rotation kernel” takes the form:

$$k_{\mathrm{rot}}(\tau) = \sigma^2 \left\{ \frac{1}{2}(1+f) e^{-|\tau|/(Q_0P)} \cos\left(\frac{2\pi |\tau|}{P}\right) + \frac{1}{2}(1-f) e^{-|\tau|/(Q_1P)} \cos\left(\frac{4\pi |\tau|}{P}\right) \right\}$$

where $P=$ rotation period, $Q_0$, $Q_1$ are damping factors, and $f$ modulates harmonic amplitudes. This formalism allows for robust isolation of periodic variability even in the presence of red noise and stochastic fluctuations.

Post-processing includes Bayesian mixture modeling to probabilistically distinguish gyrochronological sequence stars from background populations in period–temperature space (using Legendre polynomial basis for the “gyrochrone”).

6. Limitations and Scientific Impact

While IRIS substantially mitigates many instrumental and blending-induced artefacts, residual contamination can persist for extremely crowded stellar fields, especially among faint sources close to bright neighbors. Furthermore, for very low-amplitude variability or periods longer than the mission baseline, detection sensitivity declines.

Despite these limitations, the IRIS catalog elevates NGC 6819 as a benchmark system for spin-down and angular momentum evolution in intermediate-age clusters, revealing scatter and bimodality at fixed temperature and confirming a “stalling” mechanism for magnetic braking near critical Rossby numbers. These insights challenge traditional gyrochronology models, indicating that rotation period becomes a less precise age proxy at older cluster ages, especially above 2–3 Gyr (Sagynbayeva et al., 2 Oct 2025).

The general methodology and catalog design of IRIS have wider implications, providing a paradigm for extracting and analyzing time-domain photometry from crowded fields in space-based (Kepler, TESS, PLATO) and ground-based surveys, informing the next generation of time-domain and stellar population studies.

7. Future Prospects and Extensions

Future directions include adaptations of IRIS techniques for other missions and datasets (crowded TESS sectors, PLATO clusters), extension of GP-based period recovery to fainter and more blended objects, and further integration with multi-wavelength variability catalogs. Ongoing releases and community pipelines (Colman et al., 2021) continue to expand IRIS’s coverage, statistical completeness, and methodological sophistication, promising ongoing advances in rotation-age relations, stellar and planetary variability, and the physical mechanisms governing angular momentum evolution in clusters.