Wavelength self-calibration and sky subtraction for Fabry–Pérot interferometers: applications to OSIRIS

MNRAS 454, 1387–1392 (2015) doi:10.1093/mnras/stv2230 Wavelength self-calibration and sky subtraction for Fabry–Pérot interferometers: applications to OSIRIS T. Weinzirl,1‹ A. Aragón-Salamanca,1 S. P. Bamford,1 B. Rodrı́guez del Pino,1,2 M. E. Gray1 and A. L. Chies-Santos3,4 1 School of Physics and Astronomy, The University of Nottingham, University Park, Nottingham NG7 2RD, UK de Astrobiologı́a, INTA-CSIC, Villafranca del Castillo, E-28850 Madrid, Spain 3 Departamento de Astronomia, Instituto de Fãsica, Universidade Federal do Rio Grande do Sul, Porto Alegre, R.S. 91501-970, Brazil 4 Departamento de Astronomia, Instituto de Astronomia, Geofı́sica e Ciências Atmosféricas, Universidade de Saõ Paulo, 05508-900 São Paulo, SP, Brazil 2 Centro ABSTRACT We describe techniques concerning wavelength calibration and sky subtraction to maximize the scientific utility of data from tunable filter instruments. While we specifically address data from the Optical System for Imaging and low Resolution Integrated Spectroscopy instrument (OSIRIS) on the 10.4-m Gran Telescopio Canarias telescope, our discussion is generalizable to data from other tunable filter instruments. A key aspect of our methodology is a coordinate transformation to polar coordinates, which simplifies matters when the tunable filter data are circularly symmetric around the optical centre. First, we present a method for rectifying inaccuracies in the wavelength calibration using OH sky emission rings. Using this technique, we improve the absolute wavelength calibration from an accuracy of ∼5 to 1 Å, equivalent to ∼7 per cent of our instrumental resolution, for 95 per cent of our data. Then, we discuss a new way to estimate the background sky emission by median filtering in polar coordinates. This method suppresses contributions to the sky background from the outer envelopes of distant galaxies, maximizing the fluxes of sources measured in the corresponding sky-subtracted images. We demonstrate for data tuned to a central wavelength of 7615 Å that galaxy fluxes in the new sky-subtracted image are ∼37 per cent higher, versus a sky-subtracted image from existing methods for OSIRIS tunable filter data. Key words: instrumentation: interferometers – techniques: imaging spectroscopy – galaxies: clusters: individual: Abell 901/902 – galaxies: distances and redshifts. 1 I N T RO D U C T I O N A Fabry–Pérot interferometer, or etalon, is comprised of two reflecting plates working in a collimated beam. For a specific incidence angle of incoming light, the etalon transmits light of wavelength λ in a circular pattern of radius r around the optical centre. The range of wavelengths transmitted by the filter is adjusted by changing the separation between the reflecting plates. Tunable filter (TF) instruments, often built with Fabry–Pérot interferometers, are proving to be a flexible and cost-effective implementation of spectrophotometry. The ability to precisely tune to an unlimited number of wavelengths in a specified interval circumvents the need to purchase arbitrary narrow-band filters (Bland-Hawthorn & Jones 1998). TF instruments are suitable for studies of emission and absorption lines in any redshift window, and they yield higher resolution (R ∼ 500) than low-resolution grisms (González et al.  E-mail: 2014). However, the varying wavelength across the field of view makes data from TF instruments challenging to deal with. Background sky emission can be highly variable across an image in which bright OH sky emission lines appear as prominent rings (see Section 4 for an example). Full utilization of TF data requires a precise wavelength calibration and robust means of subtracting the complicated sky pattern. In this Letter, we discuss refinements to the wavelength calibration and sky subtraction for TF data from the red mode on the Optical System for Imaging and low Resolution Integrated Spectroscopy instrument (OSIRIS; Cepa 2013; Cepa et al. 2013) on the 10.4-m Gran Telescopio Canarias (GTC) telescope. We specifically consider data of emission-line galaxies from the OSIRIS Mapping of Emission-line Galaxies in A901/2 (OMEGA) survey (Chies-Santos et al. 2015) in the Space Telescope A901/2 Galaxy Evolution Survey (STAGES) field (Gray et al. 2009). Our discussion is generalizable to similar TF instruments. We summarize the most important properties of the data in Section 2. Sections 3 and 4 address the wavelength calibration and sky subtraction, respectively.  C 2015 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society Accepted 2015 August 27. Received 2015 August 26; in original form 2015 July 14 1388 T. Weinzirl et al. 2 T H E O M E G A S U RV E Y 3 WAV E L E N G T H S E L F - C A L I B R AT I O N González et al. (2014) show that the radial dependence of wavelength for the OSIRIS red TF is given by the expression λ = λ0 − 5.04r 2 + a3 (λ)r 3 , (1) where a3 (λ) = 6.0396 − 1.5698 × 10−3 λ + 1.0024 × 10−7 λ2 , (2) λ0 is the effective wavelength at the optical centre (i.e. the wavelength to which the TF instrument is tuned), r is measured in arcmin, and wavelengths are measured in Å. After applying the above calibration to our data, we still found significant wavelength offsets between the spectra of galaxies imaged independently in partially overlapping fields. The magnitude of the offsets varied from field to field, but it was in general enough to affect flux calibration and velocity measurements. Assuming the radial dependence of wavelength in equation (1) is correct (which we will test later in this section), we attempt to update the λ0 term based on the positions of sky rings in the images. Adjusting λ0 in this way essentially corrects for instrument tuning inaccuracies. The high-resolution (R ≈ 35 000 at 7000 Å) spectral atlas of Osterbrock et al. (1996) shows multiple OH emission lines populate the spectral range of our observations. We therefore simulate how the sky spectrum should look given our chosen TF bandwidth (14 Å). We convolve an OSIRIS sky spectrum of resolution higher than our data with a 14 Å FWHM Gaussian kernel. Fig. 1 shows the original sky spectrum and the result of the convolution; central wavelengths of the sky lines in the low-resolution spectrum are measured simply as the local maxima of the peaks. The relative strengths of the night sky emission lines are known to vary with time, and this will affect the adopted convolved wavelength of the blended lines, limiting the accuracy of the wavelength calibrations. To evaluate the variability of this effect, we also convolved an independent sky spectrum taken from Hanuschik (2003) taken MNRAS 454, 1387–1392 (2015) Figure 1. The dashed curve is an intermediate-resolution (R = 2500) sky spectrum from the GTC. The solid curve results after convolving the dashed curve with a Gaussian kernel having an FWHM equal to the resolution of our data (14 Å). The peaks of the sky lines in the convolve (...truncated)