JWST MIRI Medium Resolution Spectrometer Point Fixed Pattern Corrections: Cleaner and Higher Signal-to-Noise Spectra of Point Sources

Technical Summary: JWST MIRI Medium Resolution Spectrometer Point Fixed Pattern Corrections

Problem Statement
The James Webb Space Telescope (JWST) Mid-Infrared Instrument (MIRI) Medium Resolution Spectrometer (MRS) provides spectroscopic coverage from 5–28 µm. While the standard JWST data reduction pipeline corrects for many instrumental signatures, the signal-to-noise ratio (S/N) of the resulting spectra remains limited by fixed pattern noise. This noise arises from several sources, including flat-field uncertainties, wavelength-dependent flux calibration errors, sampling artifacts, and, most significantly, residual fringes.

Residual fringes are particularly problematic for point sources. While the pipeline uses a static fringe flat derived from extended sources to correct the majority of fringe signals, point sources exhibit different fringe patterns that vary with their location on the detector. Consequently, standard pipeline corrections leave residual fringes with amplitudes of 1–2%. Furthermore, the standard pipeline's optional residual fringe correction, which fits Fourier components to the data, risks removing astrophysical signal if the source contains regularly spaced emission or absorption lines (e.g., molecular bands) that mimic the fringe frequency. Additionally, other broad instrumental artifacts, such as a known feature at 5.8 µm, persist in the data, limiting the detectability of weak spectral features and capping the achievable S/N.

Methodology
To address these limitations, the authors constructed Point Fixed Pattern Corrections (PFPCs) using a diverse set of calibration targets observed with the default four-point dither pattern. The dataset includes:

• Flux Calibration Stars: O, A, and G dwarf stars (e.g., 10 Lac, HR 5467, HD 37962) providing high S/N data across the spectrum.
• Asteroids: Two Themis family asteroids (515 Athalia, 526 Jena) which serve as featureless, blackbody-like sources, particularly useful at longer wavelengths where stars are faint.

The data reduction process utilized the jwst pipeline version 2.0.0 with specific configurations to generate 1D extracted spectra for each of the four dither positions and all 12 MRS spectral segments (4 channels × 3 gratings). The PFPCs were derived by dividing the observed spectra by appropriate models:

• Stars: CALSPEC models were used. Regions containing strong absorption or emission lines were masked to prevent model inaccuracies from contaminating the correction.
• Asteroids: A quadratic fit to the spectrum in each segment was used to model the continuum, effectively removing broad features while preserving the instrumental signature.

The individual PFPCs from multiple sources were combined using sigma-clipping to create a robust average correction for each dither position and spectral segment. This approach allowed the authors to mask regions with lines in stellar spectra and fill them with data from asteroids (and vice versa), ensuring high S/N coverage across the entire 5–28 µm range.

Key Contributions

1. Development of PFPCs: The paper introduces a set of 48 wavelength-dependent corrections (4 dithers × 12 segments) specifically designed for point sources observed with the default dither pattern.
2. Hybrid Calibration Strategy: By combining stellar and asteroid observations, the authors created corrections that cover the full MRS wavelength range with high fidelity, overcoming the limitations of using stars alone (faint at long wavelengths) or asteroids alone (lack of absolute flux constraints in some bands).
3. MRS-PFPC Software Package: The authors provide a Python package (MRS-PFPC) that automates the application of these corrections. The workflow involves:
   • Reducing raw data with specific pipeline settings (pfpc proc).
   • Applying the PFPCs to individual dither spectra (pfpc cor).
   • Averaging the corrected dithers and applying a final residual fringe correction if appropriate.
   • Visualizing the results (pfpc plot).

Results

• Artifact Removal: The PFPCs successfully remove both narrow and broad spectral artifacts. A notable example is the removal of a broad, symmetric feature at 5.8 µm, which mimics a carbonyl dust feature but is confirmed to be instrumental in origin. The corrections also improve the continuity between spectral segments, reducing the need for large overlap corrections that distort spectral shapes.
• S/N Improvement:
  • For sources where the standard pipeline residual fringe correction cannot be applied (due to complex line structures), the PFPCs improve the S/N by factors of 2 to 6, with the largest gains in Channel 2.
  • For sources where the pipeline fringe correction can be applied, the PFPCs combined with the pipeline correction yield S/N values exceeding 1000 at shorter wavelengths.
  • The limiting S/N for MRS spectra without PFPCs was quantified at 100–300 (without fringe correction) and 180–430 (with fringe correction). With PFPCs, these limits are significantly raised.
• Stability: The corrections show good agreement across different dither positions and epochs, indicating that a single correction per dither position is sufficient for standard observations, provided Target Acquisition is used.

Significance
The paper demonstrates that the fixed pattern noise in JWST/MIRI MRS point source spectra is a fundamental limit that can be significantly mitigated through targeted post-processing. The PFPCs enable the detection of weaker spectral features and allow for higher S/N observations than previously possible with the standard pipeline alone. This is particularly critical for sources with complex spectra where standard fringe correction algorithms fail. The authors emphasize that the maximum achievable S/N is now often limited by the S/N of the PFPCs themselves rather than the photon noise of the observations, highlighting the importance of continued calibration efforts. The work provides a practical tool for the community to achieve cleaner, higher-fidelity spectra, pushing the calibration of MRS point source data into the sub-1% regime.