🔭 astrophysics

The Effect of "Roll-Drift" in ACS/WFC Images

This paper investigates the "roll-drift" effect in Hubble Space Telescope ACS/WFC images caused by 2024 operational changes, utilizing simulations and data reduction techniques to establish a measurable threshold that allows users to quickly identify and flag potentially degraded data.

Original authors: Yotam Cohen

Published 2026-03-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yotam Cohen

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

The Big Picture: A Shaky Hand in Space

Imagine you are trying to take a perfect, crystal-clear photo of a distant star using a camera mounted on a giant, floating telescope (the Hubble Space Telescope). To get a sharp picture, the camera needs to stay perfectly still while the shutter is open.

In 2024, Hubble switched to a new, more efficient way of flying called "Reduced Gyro Mode." Think of this like switching from a high-end, heavy-duty tripod to a lighter, more flexible one. While this new mode saves energy, it has a side effect: occasionally, the telescope gets a little "dizzy."

When this happens, the telescope loses its perfect lock on the target for a split second. It's like trying to take a photo of a friend while your hand is slightly shaking. The result isn't a blurry photo in the traditional sense; it's a "smear." The light from the star gets stretched out, like a drop of ink spreading on wet paper. This is what the scientists call "Roll-Drift."

The Problem: The "Silent" Smear

Usually, the telescope's computer is smart enough to say, "Hey, I just shook! Throw this photo in the trash and try again." But with the new mode, sometimes the computer doesn't realize it shook. These are called "undeclared loss of lock" events.

This is a problem because astronomers might download these photos, think they are perfect, and start doing complex math on them, only to realize later that the data is actually "smeared" and unreliable. It's like baking a cake and not realizing you forgot the flour until you take a bite.

The Investigation: Simulating the Shake

The scientists (led by Yotam Cohen) wanted to figure out how to spot these "silent smears" before they ruin scientific research.

The Experiment: They took a perfect, real photo of stars and used a computer program to artificially "smear" it. They shifted the image by tiny amounts (like 0.1 pixels at a time) and averaged them together, mimicking exactly what happens when the telescope drifts.
The Test: They ran a standard analysis tool (called hst1pass) on these fake smeared images. This tool tries to measure the stars and asks, "How well does this star fit the perfect shape I expect?"

The Discovery: The "Q-Fit" Score

The tool produces a score called "qfit." Think of qfit as a "Perfection Score" for a star's shape.

Low Score (near 0.0): The star looks perfect and round. The photo is good.
High Score (above 0.2): The star looks weird, stretched, or blurry. The photo is bad.

The Findings:

The "Invisible" Danger: For very small amounts of shaking (less than 0.4 pixels), the qfit score stays low. The photo looks fine, and the computer says it's fine. But it's actually slightly degraded. This is dangerous because high-precision science might be ruined without anyone noticing.
The "Red Flag" Threshold: Once the shaking gets bigger (around 1 pixel or more), the qfit score shoots up past 0.2. This is the clear warning sign. If a star has a qfit above 0.2, it's almost certainly been smeared by a roll-drift event.

The Solution: The Automated Watchdog

Now that they know what to look for, the ACS team (the group managing this instrument) has set up an automated security guard.

The Routine: As soon as a new photo comes down from Hubble, the system automatically runs the qfit test on it.
The Alert: If the system sees a photo with a suspiciously high qfit score, it sends an email to the human team.
The Check: A human looks at the photo to confirm, "Yes, this is smeared. Don't use it."

What Should You Do?

If you are a scientist using Hubble data taken after 2024:

Check your scores: Run the hst1pass tool on your own data.
Watch the qfit: If your stars have a qfit score higher than 0.2, your data might be "smeared" and you should be careful using it.
Ask for help: If you aren't sure, contact the Hubble help desk. They are now actively watching for these "shaky" photos to keep everyone's data clean.

In short: The telescope sometimes gets a little dizzy without telling us. The scientists have built a "lie detector" (the qfit score) to catch these dizzy moments so we don't accidentally trust bad photos.

1. Problem Statement

Following the Hubble Space Telescope's (HST) transition to Reduced Gyro Mode (RGM) in June 2024, the observatory has experienced an increased frequency of Loss of Lock (LoL) events. While the observatory systems typically detect and flag these events, a subset occurs as "undeclared loss of lock" events.

Consequence: During an exposure, the telescope pointing drifts slightly, causing the light from the target field to smear across the detector.
Impact: This smearing degrades data quality, potentially rendering images unusable for high-precision science.
Challenge: Because these events are often undeclared, there is no immediate metadata flag to alert users. There is an urgent need for a method to automatically detect and flag these degraded images post-observation to prevent their use in critical scientific analysis.

2. Methodology

The authors employed a simulation-based approach combined with standard data reduction techniques to quantify the impact of roll-drift on measurable parameters.

Simulation of Roll-Drift:
- Source Data: A real, archival ACS/WFC flc (flat-field corrected) image unaffected by roll-drift was selected.
- Process: Using the scipy.ndimage.shift module, the image was shifted incrementally (e.g., 0.1 pixels) in both X and Y directions up to a final displacement (e.g., 5 pixels).
- Smearing Model: To simulate the physical integration of light during a drift, the final "roll-drifted" image for a specific displacement was calculated as the average of all shifted images from zero up to that displacement.
Data Analysis (Photometry):
- The simulated images were processed using hst1pass, a standard PSF-fitting photometry pipeline.
- Configuration: The focus level was set as a free parameter to allow the code to find the "best case" PSF fit for smeared stars, mimicking typical user configurations.
Diagnostic Metric:
- The primary metric analyzed was the qfit parameter. This value measures the goodness-of-fit between the source and the Point Spread Function (PSF) model.
- Baseline: Well-measured stars in undisturbed images typically have qfit $\approx$ 0.02–0.04.
- Degradation: Poorly fit stars (due to smearing) yield higher qfit values (approaching 1.0 for cosmic rays or resolved sources).

3. Key Results

The study established a quantitative relationship between the magnitude of roll-drift and the qfit parameter:

Correlation: There is a clear positive correlation between the amount of simulated roll-drift and the median qfit value. As the image smears, the PSF model fits less accurately, and qfit increases.
Focus Level Behavior:
- Between 0 and 0.6 pixels of drift, the calculated focus level increases steadily as the code attempts to fit elongated PSF models to the smeared stars.
- Between 0.7 and 1.1 pixels, the curve flattens; the elongated PSF models can still fit the stars reasonably well, keeping qfit relatively stable.
- Beyond ~1.1 pixels, the PSF fitting breaks down, and qfit rises sharply.
The Detection Threshold:
- The study identified qfit $\approx$ 0.2 as a critical threshold.
- < 0.2: Generally considered "good" stars.
- > 0.2: Indicates poorly fit stars, likely due to roll-drift.
The "Indistinguishable" Zone:
- Small drifts (0.1 to 0.4 pixels) produce qfit values that fall within the normal distribution of unaffected images (0.0 to 0.075).
- Implication: Images with minor roll-drift may appear visually normal and pass standard quality checks, yet still contain subtle degradation that could affect high-precision measurements.

4. Key Contributions

Quantification of Undeclared LoL: The report provides the first systematic quantification of how undeclared roll-drift affects ACS/WFC data parameters.
Automated Flagging Strategy: The authors identified qfit > 0.2 as a robust, automated trigger for flagging potentially compromised data.
Operational Workflow: The ACS team has implemented an automated routine that runs hst1pass on all new ACS/WFC data immediately upon acquisition.
- The system tabulates qfit values.
- Daily emails are sent to the ACS team to manually inspect exposures with suspect qfit values.

5. Significance and Recommendations

Data Integrity: This work is critical for maintaining the scientific integrity of the HST archive post-RGM transition. It prevents the inadvertent use of smeared data in scientific publications.
User Guidance:
- Users are encouraged to run hst1pass on their own data taken after June 2024.
- They should examine the qfit distribution; if the median qfit exceeds 0.2, the data should be visually inspected for smearing or reported to the STScI Help Desk.
Scientific Caution: The report highlights that even data with "acceptable" qfit values (0.1–0.4 pixels of drift) might suffer from subtle degradation. While usable for some science cases, these images require caution for high-precision photometry or astrometry where the source of discrepancies might otherwise remain unexplained.

In conclusion, this report successfully translates a complex operational anomaly (undeclared roll-drift) into a simple, actionable diagnostic metric (qfit), enabling the HST community to rapidly assess and filter data quality in the new Reduced Gyro Mode era.