NIRSpec Pipeline Concept: A High Level Description

NIRSpec is a multi-object spectrograph (up to ∼ 100 objects) with a large field of view (∼ 10 square) and a detector response covering nearly a factor of 10 in wavelength (0.6 − 5μm). It will be employed on a segmented off-axis telescope in deep space. The combination of these factors results in a number of unique challenges for the calibration of the NIRSpec instrument. In this paper we outline some of these challenges and present a high-level concept of the NIRSpec pipeline data reduction sequence. 1. Why is NIRSpec so special? In general, the goal of the spectrophotometric calibration of any astronomical instrument is to convert — for every field point, wavelength, and “observing mode” (i.e. combination of filter, grating and band) — the number of electronic counts registered in the detector electronics (measured in digital numbers or DN) to the signal received from the astronomical source (measured in erg/s/cm/Å or similar units). In the case of a multi-object spectrograph (MOS) in general, and NIRSpec on board JWST in particular, there are a number of effects that make this calibration particularly challenging: 1. At variance with a traditional long slit spectrograph, every detector pixel in a MOS must be calibrated for its response over the entire spectral range, because spectra from a large number of celestial objects can illuminate the same detector pixel, albeit at different wavelengths. For a MOS, therefore, the “flat field” response map is a three-dimensional data cube that “stacks” the two-dimensional, (2K × 4K) pixel response maps along the wavelength dimension. In principle, the wavelength axis must be sampled at the spectral resolution of the instrument (up to 2700 in the case of NIRSpec). Possible simplifications of this approach, which would otherwise result in rather large and impractical calibration reference files, must be sought. 2. The two-dimensional NIRSpec field of view (FOV), combined with the use of reflection gratings and the NIRSpec design (see Figure 1), causes light to hit the dispersive elements at rather large (and varying) incidence angles. This leads to a significant curvature of the projected slit apertures on the NIRSpec detectors: while each individual shutter aperture is essentially tilted with respect to the dispersion axis, the tilt angle varies significantly across the FOV. This complicates calibration and data reduction because each shutter spectrum must be rectified differently. 3. Unlike for ground-based MOS, for NIRSpec the telescope PSF is a strong function of wavelength: the FWHM varies nearly by a factor of 10 between the blue and red ends of the NIRSpec spectral range. On the other hand, NIRSpec uses only a single imaging scale for its entire wavelength range, and the Micro-Shutter Array (MSA) offers only a single physical slit width. Although the shutter width (200mas) has been carefully selected to optimize throughput and resolution across the entire