NIRISS Time Series Observation Pipeline Caveats
Unique features of the JWST Science Calibration Pipeline for time-series observations (TSOs) with the NIRISS instrument, and caveats for users, are described in this article. Users should also refer to the TSO pipeline overview for characteristics and caveats that are common to all instruments. This information reflects the status for the JWST pipeline version 1.4.6.
This article describes aspects of the JWST Science Calibration Pipeline that affect processing of time-series observations (TSOs) obtained with the single object slitless spectroscopy (SOSS) mode of NIRISS. Note that early in the mission, our understanding of the observatory performance is evolving quite rapidly, and changes to calibration procedures are expected.
Stage 1 processing for SOSS
Reference pixel correction in SOSS
The reference pixel correction step effectively removes several detector-level artifacts from SOSS data, but does not properly remove 1/f noise in extracted spectra. Analyses performed during commissioning show that there is significant improvement when removing this component using non-illuminated pixels adjacent to the trace (e.g., by taking the median of these background pixels and removing this value from the pixels used to extract the spectrum). For optimal results, we recommend such 1/f treatments be performed on individual groups, prior to ramp-fitting but after the non-linearity step. Although this sequence should produce optimal results for the read-noise-dominated parts of the spectrum, competitive results may be produced by performing the steps in a different order (e.g., after ramp-fitting) for other spectral regions.
Saturation levels in NIRISS detectors as applied to SOSS observations
During commissioning observations based on TSO observations of the exoplanet HAT-P-14b — PID 1541 (for which saturation was reached by design in the up-the-ramp samples, significant deviations from linear ramps were found even after non-linearity corrections at count levels of about 35,000 ADUs (i.e., about 56,000 e– after removal of the superbias). The light curves obtained from pixels exceeding this fluence level exhibited degraded precision. Until the origin of this degradation is better characterized, users are advised to limit the counts in an integration to a maximum of 35,000 ADU / 56,000 e–.
Quantum yield and pipeline-reported errorbars
The quantum yield is an effect over which a given input photon generates more than a single electron (see, e.g, McCullough et al. 2008). The JWST pipeline does not account for these quantum yields in the noise estimates provided in the number of counts and/or count rates for "rateint.fits" products — it simply assumes the quantum yield is 1. Shortward of 2 μm, the HgCdTe detector in NIRISS might produce more than one photo-electron per photon (see, e.g., McCullough et al. 2008, Rauscher et al. 2014), resulting in non-Poissonian statistics. For TSOs, this implies that at about 1 μm, larger-than-expected light curve scatter by a factor of up to 10%–30% could be observed due to this effect. The same enhancement factor at 0.6 μm could be as large as 30%–50%. The magnitude and wavelength dependence of this enhancement are currently under investigation.
Stage 2 processing
SOSS sky background
The sky background associated with SOSS observations has an unusual shape that results from the spectral content of the background (which is typically the zodiacal light), its illumination of the pick-off mirror (POM), and its dispersion by the GR700XD element. The main characteristics include a smooth, raising background level towards longer wavelengths, with a sharp decrease caused by the edges of the POM at about 2.1 μm (which corresponds to column 700 in the trace of order 1). Although the amplitude of the background depends on the strength of the zodiacal signal, its shape as determined during commissioning remains constant to within 2%–3%. The typical peak amplitude of the background flux on the brightest background pixels is approximately 2 ADU/s (i.e., about 3 e–/s).
The JWST calibration pipeline does not currently have a step to remove the background flux, which must be removed manually during a post-processing stage. The NIRISS team has provided a smoothed background measurement obtained during commissioning observations for SUBSTRIP256 and for SUBSTRIP96, obtained by combining and smoothing dithered rates of a field with relatively few stars (Observation 5 of Program ID 1541). During commissioning, it was found that scaling this model background frame using background pixels from a target frame, allows the background component to be removed with an accuracy of up to 2%–3% (e.g., peak rates of 2 ADU/s are reduced to 0.04–0.06 ADU/s); further processing (e.g., removing the residual signal in the background by subtracting the median counts in background pixels at each column) can reduce this number to less than 1%. For applications that require higher precision, dithered exposures to obtain background measurements on a nearby area of the sky are encouraged.
SOSS spectral extraction
NIRISS SOSS spectral extraction needs special care at the red end of the spectrum, where order 1 and order 2 partially overlap—in particular for columns below 400 (i.e., above about 2.4 μm as measured by order 1). While an algorithm has been developed by the NIRISS IDT team and implemented in the JWST pipeline to account for this contamination during spectral extraction (the Algorithm to Treat Order ContaminAtion — ATOCA; Darveau-Bernier et al., in prep; available from the algorithm's Github repository), these routines are still being adapted to the in-flight properties of the instrument. In the meantime, the JWST pipeline is set by default to perform simple box extraction which might include some contamination for both order 1 and order 2.
An estimate of this contamination can be obtained using an exposure with the GR700XD grism in combination with the F277W filter. The F277W filter removes contributions from order 2 in the overlap region with order 1. This exposure can be used to obtain a precise shape for the cross-dispersion profile of order 1 in the contamination region, which can subsequently be scaled and subtracted from the CLEAR+GR700XD exposure to provide an uncontaminated estimate of the spectrum in order 2, as well as an estimate of the uncontaminated flux of order 1 in the overlap region.
SOSS wavelength solution
During commissioning observations, it was observed that the wavelength solution for NIRISS SOSS might shift from visit to visit by about a pixel (i.e., about 0.00045 μm in order 2, 0.001 μm in order 1). The causes of this global shift of the wavelength solution are still being investigated, but users should be aware of this shift if their aim is to perform analyses at the native resolution level of the instrument.
Rauscher, B. J., et al., 2014, PASP, vol. 126, issue 942
New and Better Detectors for the JWST Near-Infrared Spectrograph