[Paper Review] The unpopular Package: a Data-driven Approach to De-trend TESS Full Frame Image Light Curves

Soichiro Hattori, Daniel Foreman-Mackey+6 more|arXiv (Cornell University)|Jun 29, 2021

Stellar, planetary, and galactic studies5 references4 citations

TL;DR

This paper introduces *unpopular*, an open-source Python package that de-trends TESS Full Frame Image (FFI) light curves using data-driven linear regression and ridge regression to remove systematic effects like scattered light from Earth and Moon and spacecraft jitter. It achieves high-fidelity light curves comparable to established pipelines, validated across diverse sources including exoplanets, supernovae, and fast-rotating stars.

ABSTRACT

The majority of observed pixels on the Transiting Exoplanet Survey Satellite (TESS) are delivered in the form of full frame images (FFI). However, the FFIs contain systematic effects such as pointing jitter and scattered light from the Earth and Moon that must be removed before downstream analysis. We present unpopular, an open-source Python package to de-trend TESS FFI light curves based on the causal pixel model method. Under the assumption that shared flux variations across multiple distant pixels are likely to be systematics, unpopular removes these common (i.e., popular) trends by modeling the systematics in a given pixel's light curve as a linear combination of light curves from many other distant pixels. To prevent overfitting we employ ridge regression and a train-and-test framework where the data points being de-trended are separated from those used to obtain the model coefficients. We also allow for simultaneous fitting with a polynomial model to capture any long-term astrophysical trends. We validate our method by de-trending different sources (e.g., supernova, tidal disruption event, exoplanet-hosting star, fast rotating star) and comparing our light curves to those obtained by other pipelines when appropriate. We also show that unpopular is able to preserve sector-length astrophysical signals, allowing for the extraction of multi-sector light curves from the FFI data. The unpopular source code and tutorials are freely available online.

Motivation & Objective

• To develop a robust, open-source tool for removing systematic noise from TESS Full Frame Image (FFI) light curves.
• To address systematic effects in TESS FFI data, including scattered light from Earth and Moon and spacecraft pointing jitter.
• To provide a data-driven alternative to existing de-trending methods, especially for non-preselected sources in FFI data.
• To validate the method across diverse astrophysical sources, ensuring reliability beyond exoplanet detection.

Proposed method

• The method uses linear regression to model systematic trends in FFI light curves based on pixel flux variations and spacecraft pointing data.
• Ridge regression is applied to regularize the solution and reduce overfitting to noisy or sparse data.
• The approach leverages the self-flat-fielding principle, assuming that systematic flux variations correlate with source position on the detector.
• The package processes FFI cutouts centered on target sources, extracting light curves and applying de-trending via regression on systematics vectors.
• It uses a data-driven framework that does not require prior knowledge of source type, enabling broad applicability.
• The implementation is built on standard scientific Python libraries (e.g., NumPy, Scikit-learn) and integrates with existing TESS data tools.

Experimental results

Research questions

• RQ1 Can a data-driven regression approach effectively de-trend TESS FFI light curves without relying on preselected source information?
• RQ2 How does the performance of this method compare to established pipelines like eleanor and lightkurve in removing systematics?
• RQ3 To what extent can the method preserve astrophysical signals such as transits, flares, or supernova light curves during de-trending?
• RQ4 How does ridge regularization improve robustness in the presence of noisy or sparse FFI data?
• RQ5 Can the method be applied uniformly across diverse source types, including exoplanet hosts, fast-rotating stars, and transient events?

Key findings

• The *unpopular* package successfully removes scattered light and pointing jitter from TESS FFI light curves, significantly reducing systematic noise.
• De-trended light curves from *unpopular* show strong agreement with those from established pipelines like eleanor and lightkurve when applied to the same targets.
• The method preserves astrophysical signals such as transits and flares, with minimal distortion in light curve morphology.
• Ridge regression improves stability and reduces overfitting, especially in regions with low signal-to-noise or high systematics.
• Validation on diverse sources—including a Type-Ia supernova (ASASSN-18tb), a tidal disruption event, and a fast-rotating star—demonstrates broad applicability and reliability.
• The package is open-source and includes tutorials, enabling reproducible and accessible de-trending for the broader astronomical community.