Wavefront Error Recovery and Companion Identification with the James Webb Space Telescope

TL;DR

The paper presents a Bayesian WFE-recovery framework that uses WFSC-derived OPD priors to estimate the wavefront error during JWST/NIRISS observations while simultaneously searching for faint companions, eliminating the need for a reference star. By modeling the WFE with a Hexike-based, segment-wise parameterization and performing nested sampling, the method yields posterior distributions for both WFE and companion parameters and demonstrates near-photon-noise-limited contrast in AMI simulations. Comparisons with traditional calibrator-based interferometric analysis show competitive sensitivity, and a simulated HD 206893 B-like companion is reliably recovered, illustrating significant observing-time savings and improved close-in companion detection. The approach is broadly applicable to JWST imaging modes and could influence future missions by reducing reliance on reference stars and enhancing PSF calibration and exoplanet detection capabilities.

Abstract

The James Webb Space Telescope is orders of magnitude more sensitive than any other facility across the near to mid-infrared wavelengths. Many approved programs take advantage of its highly stable point spread function (PSF) to directly detect faint companions using diverse high-contrast imaging (HCI) techniques. However, periodic re-phasing of the Optical Telescope Element (OTE) is required due to slow thermal drifts distorting to the primary mirror backplane along with stochastic tilt events on individual mirror segments. Many programs utilize observations of a reference star to remove the stellar contribution within an image which can typically take half of the total allocated time. We present a high-contrast imaging technique for the NIRISS instrument that uses the measured wavefront error (WFE) from a phase calibration observation (performed roughly every 48 hours) as prior information in a Bayesian analysis with nested sampling. This technique estimates the WFE of a given observation and simultaneously searches for faint companions, without using a reference star. We estimate the wavefront error for both full aperture and aperture masking interferometry (AMI) imaging modes using three low order Zernike coefficients per mirror segment, using the Hexike basis, to generate synthetic PSFs and compare to simulations. We compare our technique to traditional interferometric analysis in realistic NIRISS F430M simulations both relative to the photon noise limit, and through recovering an injected companion with $Δ$F430M= 8 mag at 0.2''. With future testing, this technique may save significant amounts of observing time given the results of our current implementation on NIRISS simulations.

Figures (15)

• Figure 1: Input transmission to POPPY for the full pupil (left) and AMI (right) observing modes. X and y units are pixels.
• Figure 2: An example realization of high frequency wavefront errors of the JWST OTE. Each mirror segment was measured in lab individually and then combined with a zero gravity model to predict performance. Units are meters. X and y units are pixels. High frequency errors on the OTE are static (unless impacted by events like micro-meteorite collisions) while low frequency errors (e.g. tip/tilt, defocus) can change for individual segments.
• Figure 3: The difference in the OTE OPD after simulating some input amount of drift equivalent to rms residual WFE $\sim$ 5 nm. Differences in the WFE are both specific to individual segments and affect the segments connected to each wing (leftmost and rightmost three segments).
• Figure 4: Left: First co-added frame from simulated data cube produced by ami_sim in the AMI observing mode for NIRISS. This frame is representative of 155 co-added images each with eight groups. Center: Image produced with the maximum likelihood estimate of the WFE for the simulated image. Right: Residuals of first simulated frame minus our maximum likelihood estimated image. Note: Colorbar scales are the same for the left and center image, but different for the right image.
• Figure 5: Left: Simulated WFE of target for AMI observation. Center: Maximum likelihood estimate of the WFE. Right: Residuals of simulation minus estimate. Note: Colorbar scales are the same for the left and center image, but different for the right image.