Realistic Detector Geometry Modeling and Its Impact on Event Reconstruction in JUNO

This paper proposes a method to predict the positions of all photomultiplier tubes in the JUNO detector based on limited survey data of its deformed structure, demonstrating that incorporating this realistic geometry into reconstruction models eliminates significant vertex biases while confirming that the physical deformation itself has a negligible impact on energy resolution.

The Gist

Imagine the JUNO detector as a giant, high-tech underwater camera designed to take "photos" of ghostly particles called neutrinos. Its main goal is to solve a cosmic mystery: figuring out the "weight order" of these particles. To do this, the camera needs to be incredibly precise, capable of measuring energy with the accuracy of a master jeweler weighing a diamond.

However, there's a problem. The camera isn't just a perfect sphere sitting in a vacuum; it's a massive, heavy steel structure built deep underground. When you build something that big, gravity and the installation process cause it to sag and twist slightly, just like a heavy mattress sags in the middle when you sit on it.

Here is the story of how the scientists fixed this "sagging camera" problem, explained simply:

1. The Problem: The "Bent" Camera

The JUNO detector is lined with over 17,000 giant light sensors (called Photomultiplier Tubes, or PMTs). Think of these as the pixels on a camera sensor. For the camera to take a perfect picture, every single "pixel" needs to be exactly where the blueprints say it should be.

But in reality, the steel frame holding these sensors warped slightly during construction.

• The Reality: The sensors shifted a few centimeters up, down, or sideways.
• The Risk: If the computer software trying to reconstruct the neutrino "photo" still thinks the sensors are in their perfect, original spots, it will calculate the wrong location for the event. It's like trying to navigate a city using a map that hasn't been updated since the roads were moved; you'll end up in the wrong neighborhood.

2. The Investigation: The "Spot Check"

The scientists couldn't measure every single one of the 17,000 sensors (that would take forever and disturb the delicate installation). Instead, they acted like detectives doing a spot check.

• They measured a small sample of the steel frame and about 800 sensors using super-precise laser trackers (think of them as high-tech tape measures).
• They found that the steel frame had indeed sunk slightly at the top and twisted a bit, much like a heavy bookshelf leaning against a wall.

3. The Solution: The "Smart Prediction" Model

Since they only had data for a few sensors, they needed a way to guess where the other 16,000+ sensors were.

• The Analogy: Imagine you have a trampoline with a few people standing on it. If you measure how much the fabric dips under those few people, you can mathematically predict how much the fabric is dipping under the empty spots nearby.
• The Method: The team created a mathematical model that looked at the measured sensors and the steel frame's shape. They grouped the sensors into "rings" and "layers" (like the rings of an onion) and used the data from the measured ones to predict the exact position of the unmeasured ones.

4. The Test: Does It Matter?

Now, they had to answer the big question: Does this slight bending actually ruin our science?

They ran two types of simulations:

• Scenario A (The "Old Map"): They simulated the real, bent detector but told the computer to use the perfect, original map.
  • Result: The computer got confused! It thought the neutrinos came from the wrong place. The location error was as big as 40 millimeters (about 1.5 inches). That's like missing your target by a whole step.
• Scenario B (The "New Map"): They updated the computer's map to match the real, bent detector using their new prediction model.
  • Result: The computer got it right again! The location error disappeared.

The Surprising Twist: While the location (where the event happened) was very sensitive to the bending, the energy (how much energy the particle had) was surprisingly tough. Even with the wrong map, the energy calculation was only off by a tiny fraction (less than 1%). It's like a slightly blurry photo still telling you exactly how bright the sun was, even if it can't tell you exactly where the sun was in the sky.

5. The Conclusion: Why This Matters

This paper is a victory for precision. It shows that:

1. Big structures bend: Even massive scientific instruments aren't perfectly rigid.
2. You can fix it: By measuring a few points and using smart math, you can reconstruct the "true" shape of the detector.
3. Accuracy is key: If you don't update your map to match reality, you'll get the location of cosmic events wrong. But if you use the "Realistic Geometry" model, JUNO can continue to hunt for neutrinos with the extreme precision needed to solve the universe's biggest mysteries.

In short, the scientists took a "wobbly" camera, figured out exactly how it wobbled, and taught the computer how to see straight again. Now, JUNO is ready to take the sharpest pictures of the neutrino world ever taken.

Technical Summary

1. Problem Statement

The Jiangmen Underground Neutrino Observatory (JUNO) aims to determine the neutrino mass ordering with an unprecedented energy resolution of 3% at 1 MeV. This goal relies heavily on precise event reconstruction (vertex and energy) using data from 17,612 20-inch and 25,600 3-inch photomultiplier tubes (PMTs).

• The Challenge: In the real detector, the central stainless-steel (SS) truss structure undergoes deformations during installation due to gravitational forces and varying loads (e.g., the weight of the acrylic sphere and PMTs).
• The Consequence: These deformations cause PMT positions to deviate from their ideal design coordinates. Since reconstruction algorithms rely on precise PMT coordinates to predict charge and time responses, using the idealized design geometry for real data introduces systematic biases.
• The Data Limitation: While surveys were conducted during installation, they were limited in scope to minimize interference. Only a small fraction of PMTs (808 out of 17,612) and specific points on the SS truss were measured. A method was needed to extrapolate these limited measurements to model the positions of all PMTs realistically.

2. Methodology

The authors developed a comprehensive workflow to model the realistic detector geometry and validate its impact:

A. Survey Data Analysis

• Measurements: Three groups (Dongnan, Dadi, and IHEP) used Total Stations and laser trackers to measure the SS shell and PMT positions.
• Findings:
  • Lower Hemisphere ($z < 0$): Deformations were minimal (<1 cm) due to structural support legs.
  • Upper Hemisphere ($z > 0$): Systematic deformations were observed, including a downward "sinking" (up to ~3 cm) and twisting.
  • Correlation: A strong correlation was found between the deformation of the SS truss and the displacement of the PMTs.

B. Realistic Geometry Modeling

• Extrapolation Strategy: The authors grouped PMTs into 62 concentric rings and 12 layers.
• Displacement Functions:
  • Z-direction: Modeled using a piecewise linear function based on ring number, capturing the downward shift.
  • $\phi$-direction (Azimuth): Calculated average rotation per layer and applied uniform adjustments.
  • $\rho$-direction (Radial): Used linear interpolation of measured displacements as a function of azimuth angle ($\phi$) for specific layer groups.
• Implementation: The model was implemented in the Geant4-based JUNO simulation framework. Special care was taken to adjust the reflective surface surrounding the PMTs to prevent geometric overlaps (which increased from 16 to 93 instances but remained negligible at ~0.5% overlap rate).

C. Validation and Impact Study
Three simulation scenarios were compared to isolate the effects of geometry:

1. Case 1 (Ideal): Designed geometry for both simulation and reconstruction (Reference).
2. Case 2 (Wrong Geometry): Measured (realistic) geometry for simulation, but Designed geometry for reconstruction.
3. Case 3 (Realistic): Measured (realistic) geometry for both simulation and reconstruction.

3. Key Contributions

• Predictive Modeling: Developed a robust method to predict the positions of all 17,612 PMTs using a sparse set of survey data (808 PMTs) by leveraging the correlation with the SS truss deformation.
• Quantification of Bias: Demonstrated that failing to account for detector deformation introduces significant vertex reconstruction biases.
• Algorithm Stability: Proved that incorporating the realistic geometry into the calibration-based PMT response model preserves the stability of reconstruction algorithms without degrading performance.

4. Key Results

• Vertex Reconstruction:
  • Case 2 (Inconsistent Geometry): Using the ideal geometry for real data caused significant biases. The reconstructed vertex was pulled upward by up to 20 mm in the $z$-direction and inward toward the center by up to 40 mm in the $\rho$-direction.
  • Case 3 (Realistic Geometry): When the realistic geometry was used for reconstruction, the bias was eliminated, yielding performance nearly identical to the ideal Case 1.
  • $\phi$-direction: No significant bias was observed, as the azimuthal deformation was minimal.
• Energy Reconstruction:
  • The impact of detector deformation on energy resolution was found to be negligible (<1% variation).
  • Energy non-linearity and non-uniformity were also unaffected (<0.4% variation) by the geometry changes, provided the correct geometry is used in the reconstruction.
• Light Yield: The change in geometry resulted in a marginal increase in light yield (+0.059%), consistent with the slight increase in PMT coverage.

5. Significance

• Critical for JUNO Physics: The study confirms that while detector deformation does not inherently degrade the energy resolution, using the wrong geometry (design values) for real data reconstruction introduces severe vertex biases. These biases could compromise the precision required for neutrino mass ordering determination and other physics analyses.
• Operational Readiness: The proposed modeling method allows the JUNO collaboration to utilize the detector immediately upon data-taking, even before a full survey of every PMT is completed.
• Future Outlook: The paper establishes a framework for updating the geometry model as more survey data becomes available and suggests future work to quantify secondary effects, such as buoyancy-induced deformation of the acrylic sphere, using calibration data.

In conclusion, the paper validates that a realistic geometry model derived from limited survey data is essential to eliminate vertex reconstruction biases in JUNO, ensuring the experiment meets its stringent physics goals.