Single-star optical turbulence profiling techniques for the SHIMM and other Shack-Hartmann instruments

The Big Picture: Why Do We Care About "Air Soup"?

Imagine you are trying to take a perfect photo of a distant mountain, or trying to send a laser beam from the ground to a satellite. The problem isn't your camera or your laser; it's the air in between.

The atmosphere isn't still. It's full of invisible pockets of hot and cold air swirling around. Scientists call this optical turbulence. Think of it like looking at a campfire through a wavy heat haze. The light gets distorted, making stars twinkle and blurring telescope images.

To fix this, astronomers and engineers need to know exactly where these air pockets are and how strong they are. They need a "weather report" for the air, but instead of rain and wind, they are measuring how much the air is jiggling the light.

The Tool: The SHIMM Camera

The paper focuses on a specific instrument called SHIMM (Shack-Hartmann Image Motion Monitor).

The Analogy: Imagine a camera that doesn't just take a picture of a star; it takes a picture of the star's reflection in a grid of tiny mirrors (like a honeycomb).
How it works: If the air is perfectly still, the reflections stay in a perfect grid. If the air is turbulent, the reflections wiggle and blur. By watching how much they wiggle, the SHIMM can calculate how "rough" the air is at different heights.
The Goal: The paper is about upgrading the software that reads this camera's data to make it much more accurate, especially during the day when the sun makes the air even more chaotic.

The Upgrades: What Did They Fix?

The authors didn't just build the camera; they rewrote the math behind it. Here are the three main upgrades, explained simply:

1. The "Tilt" Correction (Z-Tilt vs. G-Tilt)

The Problem: The old math treated the wiggling of the star like a simple slope (G-Tilt). But the camera actually measures a specific type of "tilt" (Z-Tilt) that is more like a gentle lean than a sharp slope.
The Analogy: Imagine trying to measure how much a boat is rocking.
- Old Method (G-Tilt): You measure the angle of the deck relative to the horizon.
- New Method (Z-Tilt): You measure the actual lean of the boat's hull.
- Why it matters: The old method was like trying to measure a gentle breeze with a ruler meant for a hurricane. It led to errors, especially near the ground. The new math (Z-Tilt) fits the data perfectly, like a custom-made glove.

2. The "Slow Motion" Blur Fix (Exposure Time)

The Problem: Cameras don't take pictures instantly; they take a tiny fraction of a second to capture an image. If the wind is blowing fast, the "wiggle" of the star gets smeared out during that tiny moment, like a photo of a fast-moving car.
The Analogy: Imagine trying to count how many times a hummingbird flaps its wings by taking a photo. If your shutter is too slow, you just see a blur.
The Fix: The authors developed a new math trick. They take two sets of data: one with a very fast "shutter" and one with a slightly slower "shutter" (by combining two fast frames). By comparing the difference, they can mathematically "un-blur" the image and figure out exactly how fast the air was moving, even if the camera wasn't fast enough to catch it perfectly.

3. The "Wind Speed" Detective (Coherence Time)

The Problem: Knowing how rough the air is (turbulence) is only half the battle. To fix the image with a telescope, you also need to know how fast the wind is blowing that rough air.
The Analogy: If you know a storm is coming, you need to know if it's a slow drizzle or a fast hurricane to know how to prepare.
The Fix: They used a clever trick called the FADE method. They looked at a specific part of the star's light pattern (the "defocus" or how blurry the center is). By watching how fast this blurriness changes, they can calculate the wind speed at different altitudes. This allows them to predict how long a telescope can keep a clear image before the wind ruins it again.

The Results: Did It Work?

The team tested all these new math tricks using a super-computer simulation (a "digital twin" of the real world).

The Test: They fed the computer thousands of fake "air profiles" (simulated weather) and asked the SHIMM to measure them.
The Outcome: The SHIMM got it right almost every time.
- Correlation: The results matched the input data with a correlation of nearly 100% (like a perfect fingerprint match).
- Sensitivity: They found the camera can detect very faint turbulence, down to a level of $2 \times 10^{-15}$.
- The Glitch: They noticed that sometimes, if the air near the ground is very rough, the math gets confused and thinks the turbulence is slightly higher up (at 4km) than it really is. It's like a loud shout near a microphone making it hard to hear a whisper right next to it. They identified this "cross-talk" so future users can account for it.

Why Does This Matter?

This isn't just about better stargazing.

Satellite Internet: As we try to beam internet to satellites using lasers, we need to know exactly when the air is too bumpy to send a signal.
Big Telescopes: The next generation of giant telescopes (like the 40-meter class) use "adaptive optics" (mirrors that wiggle to cancel out the air). They need this precise data to work.
24/7 Monitoring: Unlike older tools that only work at night, this new method works day and night, giving us a continuous weather report for the sky.

In short: The authors took a clever camera, gave it a brain upgrade with better math, and proved it can see the invisible "bumps" in the air with incredible precision, helping us see the stars more clearly and talk to satellites more reliably.

1. Problem Statement

Atmospheric optical turbulence (OT) degrades the performance of ground-based astronomy and free-space optical communications. While integrated turbulence strength (seeing) is well-monitored, detailed vertical profiling of the turbulence structure function ( $C_n^2(h)dh$ ) is crucial for optimizing Adaptive Optics (AO) systems, scheduling observations, and modeling laser communication links.

Existing single-star profiling techniques often rely on Differential Image Motion Monitor (DIMM) concepts or specific scintillation sensors. However, there is a need for robust, continuous (24-hour) profiling using Shack-Hartmann Wavefront Sensors (SHWFS), which are already common on telescopes. The Shack-Hartmann Image Motion Monitor (SHIMM) was developed to fill this gap, but its original analysis pipeline had limitations, including:

Systematic overestimation of turbulence strength in the first atmospheric layer.
Use of Gradient (G-) tilt models which did not accurately match thresholded centroid data.
Lack of correction for non-zero exposure times (wind smearing).
Absence of a mature method to estimate coherence time ( $\tau_0$ ) and mean wind speed ( $V_{5/3}$ ) from the profile.

2. Methodology

The paper presents a mature, end-to-end analysis pipeline for the SHIMM instrument, which uses a fast Short-Wave Infrared (SWIR) SHWFS observing a single bright star. The core methodology involves solving an inverse problem to retrieve the $C_n^2(h)dh$ profile from statistical measurements of wavefront slopes and intensity fluctuations.

Key Technical Components:

Inversion Framework: The problem is formulated as $c + e = Wj$ , where $c$ is the vector of measured covariances, $W$ is the weighting function matrix, and $j$ is the unknown turbulence profile vector. The solution is found using Non-Negative Least Squares (BVLS) to ensure physical positivity of turbulence.
Z-Tilt Weighting Functions:
- The authors derived new weighting functions based on Zernike (Z-) tilts rather than Gradient (G-) tilts.
- They demonstrated that thresholded centroids (used in SHWFS) approximate Z-tilts more closely.
- The derivation accounts for square sub-apertures and includes a global tilt subtraction to isolate differential motion from tracking errors.
- Validation showed Z-tilt models fit simulated data significantly better ( $\chi^2_\nu \approx 2.1$ ) than G-tilt or original SLODAR methods ( $\chi^2_\nu \approx 6.7-7.5$ ).
Noise Estimation and Subtraction:
- A method is implemented to estimate and subtract noise bias from the covariance matrix.
- For slope noise, the system uses high-speed (600 Hz) measurements to fit a parabola to the temporal autocorrelation of spot motion, extrapolating to zero lag to remove centroid noise bias.
Finite Exposure Time Correction:
- Real-world sensors have non-zero exposure times, causing "wind smearing" that biases covariance measurements.
- The authors derived a wind-speed filter function based on the "frozen flow" hypothesis.
- They implemented a correction method using measurements at two exposure times ( $\tau_E$ and $2\tau_E$) to mathematically recover the zero-exposure weighting functions, significantly reducing bias for fast-moving layers.
Coherence Time and Wind Estimation (FADE Method):
- The paper integrates the FAst DEfocus (FADE) method.
- By analyzing the structure function of the Zernike defocus coefficient ( $a_4$ ) over time, the system estimates the effective wind velocity ( $V_{5/3}$ ).
- Combined with the retrieved $C_n^2(h)dh$ profile, this allows for the calculation of the coherence time ( $\tau_0$ ).
Layer Selection:
- The authors used the condition number of the weighting function matrix to optimize the vertical spacing of atmospheric layers.
- A "split-linear" spacing (denser layers near the ground, sparser at high altitudes) was found to minimize ill-conditioning compared to linear or logarithmic spacing.

3. Key Contributions

Derivation of Z-Tilt Weighting Functions: A significant theoretical improvement over previous G-tilt models, providing a more accurate physical model for SHWFS centroid data.
Exposure Time Correction: A validated method to correct for wind smearing effects in slope and intensity covariances, essential for accurate profiling in daytime or high-wind conditions.
Integrated Coherence Time Estimation: A novel application of the FADE method within a single-star SHWFS profiler to simultaneously retrieve turbulence profiles and coherence times.
Comprehensive Validation: Extensive end-to-end Monte Carlo simulations using real-world turbulence profiles (from Stereo-SCIDAR at Paranal) to validate the entire pipeline.

4. Results

The techniques were validated using 248 turbulence profiles from the Stereo-SCIDAR database, binned into a 4-layer model (0, 4, 12, 20 km).

Turbulence Parameters ( $r_0, \theta_0$ ): The simulation showed near-perfect agreement with inputs.
- Correlation coefficients ( $r$ ) were $\approx 1.00$ .
- Normalized Root Mean Square Error (NRMSE) was very low (0.02 for $r_0$ and $\theta_0$ ).
- Bias was minimal (e.g., -0.08 cm for $r_0$ ).
Coherence Time ( $\tau_0$ ) and Wind:
- Correlation coefficients were high ( $r = 0.98$ ).
- NRMSE was slightly higher (0.09) due to scatter at large $\tau_0$ values (weak turbulence), but the noise subtraction step was proven critical to removing significant bias.
Profile Sensitivity and Limitations:
- Sensitivity Limit: The system demonstrated a detection limit of approximately $2 \times 10^{-15} , m^{1/3} $for$ C_n^2(h)dh$.
- Cross-Talk: A systematic overestimation of the 4 km layer and underestimation of the ground layer was observed. This indicates "cross-talk" where energy from the dominant ground layer is erroneously assigned to the first free-atmosphere layer.
- High Altitude: The 20 km layer showed underestimation for weak turbulence values, consistent with the sensitivity limit.
- Missing Layers: The non-negative least squares solver occasionally returned zero for weak layers (up to 18.5% of 20 km cases), which correlates with high fractional uncertainty.

5. Significance

This work establishes a mature, robust, and transferable framework for single-star optical turbulence profiling using SHWFS instruments.

Operational Impact: The ability to profile turbulence continuously (24/7) in the SWIR makes SHIMM ideal for site characterization for optical communications ground stations and 24-hour astronomy operations.
Scientific Advancement: The shift from G-tilt to Z-tilt models and the inclusion of exposure time corrections significantly improve the accuracy of turbulence reconstruction, particularly for ground-layer dominated conditions.
Scalability: The techniques are not limited to SHIMM; they are inherently transferable to any high-speed SHWFS instrument on existing telescopes, potentially providing "free" turbulence profiling data for large AO systems without additional hardware.
Forecasting: The demonstrated accuracy supports the use of these measurements for data assimilation into numerical weather prediction models, improving turbulence forecasting for telescope scheduling and network optimization.