🔭 astrophysics

In Situ Measurements of the Reflectances of the LSSTCam Optics and Assessing the Impact of Optical Ghosts

This paper utilizes optical ray tracing simulations tuned to LSST Commissioning data and Collimated Beam Projector measurements to quantify the impact of optical ghosts on LSSTCam, revealing that approximately 0.57% of the focal plane is affected while confirming optical element reflectance estimates of roughly 2%.

Original authors: Aashay Pai, Alex Drlica-Wagner, Lee S. Kelvin, Joshua E. Meyers, Elana K. Urbach, Fritz Mueller, Robert H. Lupton

Published 2026-07-01

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Aashay Pai, Alex Drlica-Wagner, Lee S. Kelvin, Joshua E. Meyers, Elana K. Urbach, Fritz Mueller, Robert H. Lupton

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 you are taking a very high-resolution photograph of the night sky with a giant, super-sensitive camera. You want to see faint, fuzzy clouds of gas and dust (low-surface-brightness science) that are barely visible. However, there's a problem: the camera isn't perfect. Just like a window in your house reflects a tiny bit of light, the glass lenses, filters, and the camera sensor inside this giant telescope reflect a small amount of light back and forth.

When a very bright star shines into the lens, that tiny bit of reflected light bounces around inside the camera like a pinball, creating faint, ghostly copies of the star in the wrong places on the image. These are called Optical Ghosts.

This paper is essentially a report card on how much these "ghosts" mess up the pictures taken by the LSSTCam (the camera for the Vera C. Rubin Observatory) and whether the camera is following the rules set by its builders.

Here is the breakdown of what the authors did, using simple analogies:

1. The Problem: Ghosts in the Machine

Think of the camera as a long hallway lined with mirrors (the lenses and filters). If you shine a flashlight (a bright star) down the hall, you don't just see the light at the end; you see faint reflections bouncing off the mirrors along the way.

The Issue: These reflections look like fake stars. If you are trying to study a faint cloud of gas, a ghost from a bright star could look like part of that cloud, ruining your data.
The Rule: The telescope's designers said, "The area of the photo covered by these ghosts must be less than 1%."

2. The Simulation: A Virtual Pinball Machine

The authors didn't just guess; they built a computer model of the entire camera.

The Setup: They took a list of all the bright stars in the sky (like a "Hall of Fame" of stars) and used a physics simulation (called Batoid) to trace how light would bounce around inside the camera for every single photo the telescope takes.
The Result: They found that, on average, 0.57% of the camera's view is covered by these ghosts.
- Analogy: If the camera's view was a football field, the ghosts would cover less than one small square of grass.
- Pass/Fail: Since 0.57% is less than the 1% limit, the camera passes the requirement.
- The Catch: The ghosts are worse in the "blue" part of the light spectrum (the u band) because the sky is naturally darker there, making the ghosts stand out more. In the "red" and "near-infrared" parts, they are almost invisible.

3. The Real-World Test: The "Flashlight" Experiment

Computer simulations are great, but you need to check if reality matches the math. To do this, the team used a special tool called the Collimated Beam Projector (CBP).

The Analogy: Imagine you are trying to test how much light bounces off a car windshield. Instead of waiting for the sun to hit it, you set up a powerful, focused flashlight right outside the car in a dark garage.
The Experiment: They used a laser flashlight to shoot light into the telescope from a distance. This created artificial "ghosts" on the camera sensor without needing to wait for real stars.
The Measurement: By measuring how bright these artificial ghosts were compared to the original laser beam, they could calculate exactly how reflective the glass and filters are.
The Finding: The real-world measurements showed that the glass reflects about 2% of the light. This matched the engineers' original predictions perfectly. This confirmed that their computer simulation was accurate.

4. The Conclusion

The paper concludes with two main takeaways:

The Camera is Compliant: The optical ghosts affect less than 1% of the sky images, so the telescope meets its safety standards.
The Ghosts are Real: While the camera passes the test, these ghosts are still a real nuisance. If you are looking for extremely faint objects near a very bright star, those ghost reflections will still be there, acting like a "smudge" on your lens that you can't wipe off.

In short: The team built a virtual model to predict where "ghost stars" would appear, checked it with a real laser flashlight in a test lab, and confirmed that the camera is safe to use for science, though astronomers still need to be careful when bright stars are nearby.

Technical Summary: In Situ Measurements of the Reflectances of the LSSTCam Optics and Assessing the Impact of Optical Ghosts

Problem Statement
Optical ghosts are image artifacts resulting from successive reflections of light between optical surfaces (lenses, filters, and detectors). While generally negligible for most astronomical sources, ghosts produced by bright stars can significantly contaminate data, posing challenges for sky background estimation, photometric measurements, and low-surface-brightness science. The LSST System Requirement mandates that the percentage of the image area containing ghosts with a surface brightness gradient amplitude exceeding 1/3 of the sky noise over 1 arcsecond must be less than 1%. However, the wording of this requirement allows for interpretation. The authors adopt a conservative definition: the total area of the focal plane containing a ghost with a flux greater than 1/3 of the sky noise on any detector. The primary challenge addressed is quantifying this impacted area and validating the optical reflectance assumptions used in simulations, as direct on-sky measurements are complicated by uneven flat-fielding, spatial degeneracy of overlapping ghosts, and sky background gradients.

Methodology
The study employs a two-pronged approach combining optical ray-tracing simulations with in situ reflectance measurements.

Simulation Framework (Batoid):
- The authors utilize the Batoid Python package to simulate optical ghosts for specific LSSTCam visits.
- Input Data: Bright stars are identified from the Yale Bright Star Catalog within 1.9 degrees of the boresight. Magnitudes are transformed from the V-band to LSST bands (u, g, r, i, z, y) using established photometric transformation equations (converting via the DES system where possible; otherwise, V-band magnitudes are used as proxies).
- Ghost Modeling: Ghosts are modeled based on ray paths between specific optical elements (e.g., Lens 3 to Filter, Detector to Filter). The simulation calculates the morphology, flux, and surface brightness of ghosts for each visit.
- Impact Calculation: For each ghost, the surface brightness ( $S_n$ ) is calculated by dividing the total ghost flux by the ghost area on a specific detector. The ratio of this surface brightness to the median sky noise is computed. If this ratio exceeds 1/3, the area is marked as "impacted." The total impacted area is the union of all such areas across all ghosts and detectors, accounting for overlaps.
In Situ Reflectance Measurements (Collimated Beam Projector):
- To validate the reflectance values used in simulations, the authors used the Collimated Beam Projector (CBP), a 30 cm Schmidt telescope fitted with a laser, mounted on the Rubin Observatory dome.
- Procedure: The CBP projects a monochromatic, point-like source onto the primary mirror, creating artificial ghosts on the focal plane. Data was collected during commissioning in "no filter," g, r, and i bands.
- Flux Extraction: The authors binned the data and used Batoid to generate morphological templates for 16 prominent ghosts. They employed non-negative least squares fitting to extract the flux of each ghost relative to the source. To avoid bias from the saturated source beam, a second fit was performed with the source region masked.
- Reflectance Derivation: The ghost fluxes ( $f_{A,B}$ ) and source flux ( $f_{src}$ ) are related by the product of the reflectances of the two optical elements involved ( $R_A R_B$ ). By linearizing the system of equations in log-space ( $\log(f_{A,B}/f_{src}) = \log R_A + \log R_B$ ), the authors solved for the individual reflectances of the optical surfaces (Filter, Detector, Lenses) using a least-squares matrix inversion.

Key Results

Ghost-Impacted Area:
- Applying the simulation framework to ~3,100 visits from the weekly 37 intermittent Data Release Pipeline (DRP), the average fraction of the focal plane significantly impacted by ghosts is 0.57% when averaged across all bands.
- This value is well below the 1% system requirement threshold.
- Band Dependence: The impact varies significantly by band. The u-band shows the highest impact (~~8%), driven by lower sky background and higher filter reflectance. The r-band (~~0.2%) and i-band (<0.1%) show minimal impact.
- Star Brightness and Position: The impacted area is strongly dependent on stellar magnitude but only weakly dependent on the star's off-axis angle. Brighter stars create larger impacted areas.
- Worst-Case Scenarios: Visits selected visually for severe ghosting showed significantly higher impacted areas, confirming that individual visits with very bright stars can exceed the average significantly.
Reflectance Measurements:
- The measured reflectances for the Filter (F1, F2) and Detector (D) generally agree with the ~2% estimates used in systems engineering throughput predictions.
- The authors note that saturation of the source beam in the CBP data could theoretically lead to an overestimation of reflectances (since the measured source flux would be underestimated). However, the agreement between the measured values and engineering simulations, along with consistency between masked and unmasked fits, suggests the bias is small. Thus, the measured values are treated as conservative upper bounds.

Significance and Claims
The paper claims to provide a validated framework for assessing the impact of optical ghosts on LSSTCam data, demonstrating compliance with the LSST system requirement that less than 1% of the image area be affected by ghosts exceeding the 1/3 sky noise threshold.

The study establishes that while optical ghosts are a non-negligible source of contamination for low-surface-brightness science—particularly in the u-band and in the presence of very bright stars—the overall impact on the survey is within acceptable limits. Furthermore, the in situ measurements using the CBP successfully validate the optical reflectance assumptions used in the ray-tracing simulations, confirming that the systems engineering predictions of ~2% reflectance are accurate. This validation is crucial for the reliability of future ghost impact assessments and science case evaluations.

1. The Problem: Ghosts in the Machine

2. The Simulation: A Virtual Pinball Machine

3. The Real-World Test: The "Flashlight" Experiment

4. The Conclusion

More like this