[Paper Review] Overview of focal plane wavefront sensors to correct for the Low Wind Effect on SUBARU/SCExAO

Sébastien Vievard, Steven P. Bos+16 more|arXiv (Cornell University)|Dec 20, 2019

Adaptive optics and wavefront sensing15 citations

TL;DR

This paper presents four focal plane wavefront sensors—ZAP, Fast and Furious, Neural Network, and LAPD—designed to correct the Low Wind Effect (LWE) on Subaru/SCExAO, which degrades high-contrast imaging by inducing phase discontinuities near telescope spiders. The Fast and Furious algorithm successfully stabilizes the PSF in simulations, achieving a Strehl ratio of 0.9 within 15 iterations, demonstrating real-time, on-sky viable compensation for LWE-induced aberrations.

ABSTRACT

The Low Wind Effect (LWE) refers to a phenomenon that occurs when the wind speed inside a telescope dome drops below $3$m/s creating a temperature gradient near the telescope spider. This produces phase discontinuities in the pupil plane that are not detected by traditional Adaptive Optics (AO) systems such as the pyramid wavefront sensor or the Shack-Hartmann. Considering the pupil as divided in 4 quadrants by regular spiders, the phase discontinuities correspond to piston, tip and tilt aberrations in each quadrant of the pupil. Uncorrected, it strongly decreases the ability of high contrast imaging instruments utilizing coronagraphy to detect exoplanets at small angular separations. Multiple focal plane wavefront sensors are currently being developed and tested on the Subaru Coronagraphic Extreme Adaptive Optics (SCExAO) instrument at Subaru Telescope: Among them, the Zernike Asymmetric Pupil (ZAP) wavefront sensor already showed on-sky that it could measure the LWE induced aberrations in focal plane images. The Fast and Furious algorithm, using previous deformable mirror commands as temporal phase diversity, showed in simulations its efficiency to improve the wavefront quality in the presence of LWE. A Neural Network algorithm trained with SCExAO telemetry showed promising PSF prediction on-sky. The Linearized Analytic Phase Diversity (LAPD) algorithm is a solution for multi-aperture cophasing and is studied to correct for the LWE aberrations by considering the Subaru Telescope as a 4 sub-aperture instrument. We present the different algorithms, show the latest results and compare their implementation on SCExAO/SUBARU as real-time wavefront sensors for the LWE compensation.

Motivation & Objective

• Address the Low Wind Effect (LWE), a thermal-induced wavefront distortion caused by temperature gradients near telescope spiders under low wind conditions (<3 m/s), which degrades high-contrast imaging performance.
• Overcome limitations of traditional wavefront sensors (e.g., pyramid, Shack-Hartmann) that fail to detect LWE-induced phase discontinuities due to pupil plane discontinuities.
• Develop and validate real-time, focal plane wavefront sensing algorithms compatible with ongoing science observations on SCExAO.
• Enable high-contrast exoplanet detection by correcting LWE-induced PSF distortions, particularly at small angular separations.
• Demonstrate the feasibility of integrating multiple wavefront sensing techniques on a single instrument platform for future Extremely Large Telescopes (ELTs).

Proposed method

• Implement the Zernike Asymmetric Pupil (ZAP) wavefront sensor to directly measure LWE-induced phase aberrations in focal plane images using pupil apodization and phase diversity.
• Adapt the Fast and Furious (F&F) algorithm, a temporal phase diversity method that reconstructs wavefront errors by splitting a single PSF into even and odd components, using prior deformable mirror (DM) commands to resolve sign ambiguity.
• Train a neural network on SCExAO telemetry data to predict PSF morphology and wavefront errors in real time, enabling on-sky PSF prediction without requiring additional hardware.
• Apply the Linearized Analytic Phase Diversity (LAPD) algorithm to model the Subaru telescope as a 4-subaperture system, enabling multi-aperture cophasing to correct LWE-induced piston, tip, and tilt errors.
• Integrate all algorithms into SCExAO using existing hardware: the CRED2 near-IR camera for focal plane imaging and real-time access to the deformable mirror (DM) command stream.
• Validate algorithms through monochromatic PSF simulations matching CRED2 camera parameters, with performance measured via Strehl ratio evolution during loop closure.

Experimental results

Research questions

• RQ1 Can focal plane wavefront sensors detect and correct LWE-induced phase discontinuities that are invisible to conventional wavefront sensors?
• RQ2 How effective is the Fast and Furious algorithm in stabilizing the PSF and recovering high Strehl ratios under simulated LWE conditions?
• RQ3 Can a neural network trained on SCExAO telemetry accurately predict PSF morphology in the presence of LWE-induced aberrations?
• RQ4 To what extent can the LAPD algorithm, treating Subaru as a 4-subaperture system, correct for LWE-induced piston, tip, and tilt errors?
• RQ5 What are the hardware and integration requirements for deploying multiple focal plane wavefront sensors simultaneously on SCExAO for comparative on-sky validation?

Key findings

• The Fast and Furious algorithm achieved a Strehl ratio of 0.9 within 15 iterations during closed-loop simulation, indicating effective correction of static LWE modes.
• The initial Strehl ratio in the presence of uncorrected LWE was approximately 0.45, demonstrating significant PSF degradation due to secondary lobes.
• The ZAP wavefront sensor successfully measured LWE-induced aberrations in on-sky tests, confirming its ability to detect pupil-plane phase discontinuities.
• Neural network-based PSF prediction showed promising results in on-sky validation, though quantitative metrics remain to be fully quantified.
• The F&F algorithm enables real-time wavefront correction without interrupting science observations, as it uses only existing camera and DM data streams.
• All four algorithms—ZAP, F&F, Neural Network, and LAPD—demonstrate compatibility with SCExAO’s modular architecture, enabling future comparative on-sky performance evaluation.