A New Method for Aperture Masking Interferometric Imaging: Demonstration with the JWST

TL;DR

The paper adapts radio interferometric self-calibration to JWST’s aperture masking interferometry (AMI) to image near-infrared sources with high dynamic range. By deriving visibilities from interferograms and iteratively solving for both the source structure and element-based gains, the method provides a real-time-like wavefront sensor and robust imaging, validated on WR137 with a calibrator to correct baseline-based distortions. The results show self-calibration yields image dynamic ranges up to approximately $2\times10^2$ and retrieves JWST mirror piston values within roughly $10$–$15$ nm across apertures, while baseline-phase corrections improve the final dynamic range by about 23%. Non-closing errors due to under-sampling, charge migration, and near-field optics limit the ultimate contrast and piston precision, highlighting the need for improved uv-coverage and calibration strategies in JWST-AMI. Overall, the study demonstrates that AMI on JWST can achieve high-fidelity imaging and provides a practical pathway to real-time wavefront sensing in space-based optical interferometry.

Abstract

We present a new method for aperture masking interferometric (AMI) imaging at near-IR wavelengths using radio astronomical techniques. The method starts with derivation of interferometric visibilities from a Fourier transform of the interferograms. An iterative joint optimization process is then employed, using self-calibration of the interferometric element-based complex voltage gains (i.e. electric fields), and CLEAN deconvolution to obtain the source structure. We demonstrate the efficacy of the method using the NIRISS aperture masking interferometer on the James Webb Space Telescope (JWST) at 4.8~$μ$m and 3.8~$μ$m. Due to a number of effects (the large pixel size, charge migration, near-field optics), the method also requires an initial visibility-based amplitude normalization using observations of a well know point-source calibration star. We employ early science observations of the dusty binary Wolf-Rayet star WR137. Images with a dynamic range (peak/rms) of $\sim 240$ on the target, and $\sim 1000$ on the calibrator, are synthesized from a short integration. The self-calibration process determines the photon path-lengths through the optical system to each aperture using data on the target source itself, thereby representing an essentially 'real-time', precise wavefront error sensor. Four independent measures of the JWST mirror segment pistons (two wavelengths for two sources), agree to within 10~nm to 15~nm, comparable to the expected errors based on an analysis of closure phases on the calibrator star. Including a baseline-based phase correction improves the dynamic range of the final images by about 23\%.

Figures (7)

• Figure 1: Left and Center: example JWST AMI interferogram of WR137 at 4.8$\mu$m. The pixel size is 65mas. The grayscale stretch of the center image is saturated to emphasize the first diffraction lobe of the power pattern ('primary beam') set by the size of the hexagonal apertures of the JWST aperture mask. Right: The amplitudes of the u,v samples derived by a Fourier transform of the AMI image. X and Y axes are the interferometer u,v baseline coordinates, in meters.
• Figure 2: Example time series of visibility amplitude and phase for two baselines (1-2 red; 2-4 black) on WR137 at 4.8$\mu$m. The integration time per point is 0.3 s.
• Figure 3: Left: Closure phases for all the independent closure triads in the mask, on the calibration star HD 228337. For a point source, the closure phase should be zero, unless there are baseline-dependent phase errors. The X-axis numbers map to triads as: 1 = (0, 1, 2), 2 = (0, 1, 3), 3 = (0, 1, 4), 4 = (0, 1, 5), 5 = (0, 1, 6), 6 = (0, 2, 3), 7 = (0, 2, 4), 8 = (0, 2, 5), 9 = (0, 2, 6), 10 = (0, 3, 4), 11 = (0, 3, 5), 12 = (0, 3, 6), 13 = (0, 4, 5), 14 = (0, 4, 6), 15 = (0, 5, 6). Center: closure amplitudes for all the independent closure amplitudes quads in the mask on HD 228337. For a point source, the closure amplitudes should be unity. Departure from unity implies non-closing errors. The X-axis numbers map to quads as: 1 = (0, 1, 2, 3), 2 = (0, 1, 3, 4), 3 = (0, 1, 4, 5), 4 = (0, 1, 5, 6), 5 = (1, 2, 3, 4), 6 = (1, 2, 4, 5), 7 = (1, 2, 5, 6), 8 = (1, 2, 6, 0), 9 = (2, 3, 4, 5), 10 = (2, 3, 5, 6), 11 = (2, 3, 6, 0), 12 = (3, 4, 5, 6), 13 = (3, 4, 6, 0), 14 = (4, 5, 6, 0). Right: The normalized measured visibility amplitude on HD 228337 at 4.8$\mu$m (black) and 3.8$\mu$m (red) vs. baseline length. The dotted line shows the model for expected decoherence due to the large pixel size and under-sampling of the spatial fringes at 4.8$\mu$m (Section \ref{['sec:measurements']}).
• Figure 4: Time series of aperture-based amplitude and phase self-calibration gain solutions on WR137 at 4.8 $\mu$m. The color-coding to aperture number is shown on the right plot (see Figure \ref{['fig:piston']}). Note the zero start time is roughly time 100s in Figure \ref{['fig:VisAP']}.
• Figure 5: Top left: Image of WR137 from Lau et al.lau2024 (their figure 5). Contours are at 4.8$\mu$m and color scale is 3.8$\mu$m, with the first solid green contour = 0.33% of the peak intensity. Top right: The self-calibrated and CLEANed image of WR137 at 4.8$\mu$m. Bottom left: The self-calibrated and CLEANed image of WR137 at 3.8$\mu$m. In both cases, the contour levels are a geometric progression in factors of two, with the first level being 0.45% of the peak surface brightness. Dashed contours are negative. In both cases the Gaussian restoring CLEAN beam has FWHM = 70 mas. Bottom right: CLEAN WR137 image at 4.8$\mu$m without self-calibration. In this case, the first contour level is 2.7% of the peak surface brightness.