🔭 astrophysics

QLP Data Release Notes 004: TESS-Gaia Light Curve Photometry Implementation

The Quick-Look Pipeline (QLP) has updated its photometry methods starting with Sector 94 by adopting the TESS-Gaia Light Curve package's techniques to improve noise characteristics and detection sensitivity for fainter stars (10.5 < T < 13.5) while maintaining seamless compatibility for existing users.

Original authors: Glen Petitpas, Jack Haviland, Te Han, Willie Fong, Katharine Hesse, Avi Shporer, Jeroen Audenaert, Daniel Muthukrishna, Roland Vanderspek, George R. Ricker

Published 2026-03-24

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Glen Petitpas, Jack Haviland, Te Han, Willie Fong, Katharine Hesse, Avi Shporer, Jeroen Audenaert, Daniel Muthukrishna, Roland Vanderspek, George R. Ricker

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the TESS space telescope as a giant, high-powered camera orbiting Earth, constantly taking panoramic photos of the night sky to find new planets. But these photos are massive, containing millions of stars. To make sense of them, the scientists at MIT use a digital assistant called the Quick-Look Pipeline (QLP). Think of the QLP as a super-fast, automated chef who takes those raw, giant photos and chops them up into individual "light curves"—graphs that show how bright each star is over time. If a star's light dips slightly, it might mean a planet is passing in front of it.

For a long time, this automated chef was very good at cooking for the "VIP guests" (the bright, easy-to-see stars), but it struggled a bit when trying to serve the "faint guests" (dimmer stars). The recipes it used (a method called FIPHOT) were a bit messy, often picking up too much "noise" or static from nearby stars, making it hard to spot the tiny, subtle dips caused by small planets around faint stars.

The Big Upgrade: A New Recipe

In this new update (Data Release Notes 004), the team has swapped out the old recipe for a brand-new, high-tech one called TESS-Gaia Light Curve (TGLC).

Here is how the new method works, using a simple analogy:

The Old Way (FIPHOT): Imagine trying to listen to a whisper in a crowded room. The old method was like putting a large, fuzzy microphone over your ear. It picked up the whisper, but it also picked up a lot of the chatter from the people standing next to you. If you were trying to hear a very quiet whisper (a faint star), the background noise made it impossible.
The New Way (TGLC): The new method is like using a high-tech, directional microphone that knows exactly where the whisperer is standing (using a precise map called Gaia). It focuses only on that specific person and actively cancels out the chatter from everyone else. This allows the chef to hear the faintest whispers clearly.

Why This Matters

Seeing the Invisible: With this new "directional microphone," the QLP can now detect planets around stars that are much dimmer than before. It's like upgrading from a flashlight to a laser pointer; you can now see details in the dark corners of the universe that were previously hidden.
No Disruption for Regular Users: You might worry that changing the recipe would mess up the meal for everyone else. But the scientists were very careful. Even though the main "soup" (the data) is now made with the new, cleaner method, they still serve it in the same three bowls (apertures) as before. If you were used to the old data, the new data will look familiar and work perfectly with your existing tools.
Speed and Efficiency: Surprisingly, this fancy new method is actually faster. By using powerful computer chips (GPUs) to do the heavy lifting, the new system processes the data in half the time it took the old system.

The Result

The team tested this new system by re-running their planet-hunting algorithms. They found that the new method didn't accidentally delete any real planets; it just made the "noise" quieter. This means that signals that were previously too weak to be heard (because they were drowned out by static) can now be detected.

In short: The QLP team has given their planet-hunting software a major upgrade. They've swapped a blurry, noisy lens for a crystal-clear, noise-canceling one. This allows them to find smaller, more distant planets around fainter stars, all while keeping the data format the same so astronomers everywhere can keep using it without a hitch. It's a win-win: clearer data, deeper searches, and faster processing.

1. Problem Statement

The Quick-Look Pipeline (QLP), operated by the MIT TESS Science Office, processes TESS Full-Frame Images (FFIs) to generate light curves for up to 2 million stars per sector to search for transiting exoplanets. Historically, the pipeline's photometric methods (specifically FIPHOT) were optimized for brighter targets ( $T \le 10.5$ ) due to the high volume of human inspection required for fainter signals.

The Limitation: As machine learning capabilities advanced, allowing for deeper searches, the need arose to improve photometric precision for fainter stars (up to the pipeline's generation limit of $T = 13.5$ ).
The Goal: To enhance noise characteristics and detection sensitivity for fainter magnitudes without disrupting existing planet search workflows or breaking backward compatibility for external users.

2. Methodology

The authors implemented a major update to the QLP photometry engine, replacing the legacy FIPHOT method with the TESS-Gaia Light Curve (TGLC) method (Han & Brandt 2023).

Core Algorithm (TGLC):
- Utilizes Gaia DR3 star positions and magnitudes as priors.
- Performs forward modeling of the FFIs using an effective Point Spread Function (ePSF).
- Explicitly removes contamination from nearby stars, a significant source of noise in crowded fields.
Adaptation for QLP Compatibility:
- The native TGLC method uses a single $3\times3$ pixel aperture. However, the QLP planet search relies on depth-aperture tests (checking if transit depth increases with aperture size to identify false positives from blended binaries).
- To maintain backward compatibility, the authors retained the delivery of three apertures for all stars brighter than $T=13.5$ $T = 13.5$ :
  1. Primary: The TGLC $3\times3$ pixel aperture (used for the main light curve).
  2. Secondary: Added $1\times1$ and $5\times5$ pixel apertures to facilitate depth-aperture tests and match previous QLP delivery standards.
Implementation Timeline: This method was implemented starting with Sector 94.
Processing Efficiency: Despite the added complexity of partitioning FFIs and deriving ePSFs for each section, the new Python routines are executed on multiple GPUs, completing the photometry process in approximately half the time of the legacy method.

3. Key Contributions

Photometric Upgrade: Successfully transitioned the QLP from FIPHOT to TGLC, significantly improving noise characteristics for stars in the magnitude range $10.5 < T < 13.5$ .
Backward Compatibility: Engineered a hybrid aperture system (1x1, 3x3, 5x5) that allows existing QLP planet search algorithms (e.g., depth-aperture tests) to function without modification, ensuring a seamless transition for external users.
Pipeline Validation: Conducted rigorous testing to ensure the new data does not negatively impact the existing machine learning classifiers (Astronet-Triage) or signal-to-noise (S/N) thresholds derived from older data.

4. Results

Precision Improvement: Analysis of 950,000 light curves from Orbit-185 shows that the TGLC method (orange in Figure 1) provides superior photometric precision compared to the original FIPHOT method (blue), particularly in the $10.5 < T < 13.5$ range. The discontinuity previously seen at $T=11.5$ (due to aperture switching in the old method) is eliminated.
Search Stability:
- A re-run of the planet search using a mix of old (pre-Sector 89) and new (Sector 89 TGLC) data was performed.
- Out of ~4,890 signals that passed the Astronet-Triage filter, 454 signals were lost when TGLC data was introduced.
- Investigation: Almost all lost signals were false positives that fell just above the noise threshold in the old data. The improved noise characteristics of TGLC lowered the noise floor, pushing these signals slightly below the detection threshold. The remaining lost signals were non-transit-like and would have been rejected during human inspection.
- Conclusion: The transition did not cause a significant loss of valid planet candidates; rather, it improved the signal-to-noise ratio, filtering out marginal false positives more effectively.
Performance: The GPU-accelerated processing reduced the total time required for photometry extraction by 50%.

5. Significance

Deeper Exoplanet Searches: The improved precision allows the QLP to detect smaller planets around fainter stars ( $T$ up to 13.5) that were previously undetectable due to noise limitations.
Future Reprocessing: The authors plan to reprocess the entire TESS archive using TGLC methods to uncover signals currently too weak for detection.
Data Continuity: The resulting High-Level Science Products (HLSP) FITS files remain fully compatible with previous QLP light curves, ensuring that long-term studies and user workflows are not disrupted.
Adaptive Machine Learning: The pipeline will continue to monitor and retrain the Astronet-Triage classifier as the ratio of TGLC data increases, ensuring optimal performance for the evolving dataset.

In summary, this update represents a critical evolution in TESS data processing, leveraging Gaia priors and advanced PSF modeling to push the boundaries of exoplanet detection while maintaining the robustness and compatibility of the existing Quick-Look Pipeline infrastructure.