The Integral Field Spectrograph OASIS

Roland Bacon, Gilles Adam, Yannick Copin, Eric Ensellem,
Pierre Ferruit, Arlette Pécontal, Emmanuel Pécontal

Centre de Recherche Astronomique de Lyon
Electronic-mail : bacon@obs.univ-lyon1.fr

Abstract:

This paper presents OASIS, the new integral field spectrograph built by the Observatoire de Lyon for CFHT. OASIS is designed to be coupled with the AOB system and offers high-spatial imaging spectroscopy capabilities using an array of about 1200 microlenses. A special data reduction software has been extensively developed by Lyon and is available to the general community. Preliminary results from two observing runs in early 1998 are presented.

Introduction

For most of the astronomical targets which present small scale structures embedded in high diffuse background or crowded environment, the spatial resolution is essential. Understanding the physical processes related to theses structures relies generally on getting spectrographic information at high spatial resolution. This goal is now reachable thanks to the success of the adaptive optics. Today, the main problem is not to get the maximum of light into a long slit, but to get all the valuable information into a 2D area. Given the success of the pioneer instrument TIGER, we have designed and realized the integral field spectrograph OASIS dedicated to the PUEO adaptive optics bonnette.

The challenge was to provide a facility coupling the AOB bonnette (PUEO) with a multi-mode integral field spectrograph, while having an easy to use user interface and a powerful data reduction package. We present briefly in this paper the main characteristics of the instrument as well as its first preliminary scientific results.

Instrument

The instrument design follows the TIGER concept described in Bacon et al (1994): a spectrographic mode using a microlens array as the spatial sampler followed by a classical refractor spectrograph, and a direct imaging mode. Various sampling sizes and spectrographic resolutions are available (see table 1). Note that OASIS can also be used at F/8; change from AOB to F/8 is simply done by remotely moving a mirror within the AO bonnette.

OASIS has been funded by the CNRS, the MENRT and the Région Rhône-Alpes and is operated at CFHT as a guest instrument.

 

Operation

Operating an integral field spectrograph coupled with an adaptative optics bonnette is not a trivial task. Therefore, a significant part of the effort has been focused on the realisation of a user friendly interface. The common user will generally interact with the instrument by setting scenarios, using observational parameters such as sampling size, field of view, spectral resolution and wavelength range. The control software convert the designed scenario into the corresponding instrumental configuration (including focus offsets) which is then sent to the instrument. Keywords are written in the FITS header of the output images and will be used later in the data reduction process. The first tests conducted by external observers in January and March 98 have shown that interactions with the instrument are fairly easy.

First Light

OASIS has been mounted on the OHP 1.93m telescope in January 97. We essentially conducted extensive flexure tests as well as first operational tests. Quantitative measurements demonstrate that flexures were well within the specifications, i.e., less than 1 pixel in one hour integration time with a zenital angle between 0 to 60^o. More extensive tests and debugs then occurred in Lyon, and the instrument was finally shipped to Hawaii in May and setup on the CFH telescope in July 97.

Unfortunately, the dome has never been opened during the first 4 engineering nights because of bad weather. Finally OASIS got its first light at CFHT during a second engineering run a month later. We present in figure 1 results obtained from a 30mn exposure of the central part of the globular cluster M13. PUEO loop was closed on a V 15.2 star offcentered 17 arcsec from center of the field. OASIS configuration was 0.11 arcsec sampling giving a 4$\times$3 arcsec² field of view and a spectral range of 0.8-1 $\mu$m with a resolution of 1000. The reconstructed image obtained by integrating over the full wavelength range each of the 1200 spectra has a resolution of 0.25 arcsec FWHM. Such a resolution is a factor of 2 better than what is possible to achieve with SIS and 4 to 6 times better than in usual spectrographic exposure. Furthermore the 3D capabilities allow us to observe the 4$\times$3 arcsec² field of view in a single 30mn shot; for comparison, a classical long slit spectrograph would require 35 hours of exposure to record the same information.

 

Figure 1:   Left: OASIS mounted at CFHT. Lower right panel: Reconstructed image of the central part of M13. Upper right panel: Same image at 1 arcsec resolution.

Data Reduction

As shown in figure 2, the OASIS output image looks quite different from the now classical spectrograms given by long-slit spectrographs. One might thus conclude that the OASIS data reduction is far more complex than the long-slit one. However, one should notice that long-slit data reduction, although being widely used by the community, is not a complete trivial task and complex procedures have been developed to take into account the effect of optical distortions and aberrations, to reject cosmic events and bad pixels, or to optimize spectra extraction. In fact OASIS's data differ from spectrograms only by the discrete and hexagonal arrangement of spectra. It is thus conceivable to write a simple task to reorder the spectra along one of the image axis, giving it a quasi spectrogram appearance. If this task is able to save the initial location of the spectra, the remaining data processing can follow the classical long slit data reduction processes.

 

Figure 2:   Typical OASIS exposures

However, as emphasised in Bacon et al. (1994), integral field spectrographs could theoretically achieve high spatial resolution and spectrophotometric accuracy. This ambitious goal cannot be achieved with a simple adaptation of the long slit data reduction scheme as it was never intended to really deal with spatial resolution nor spectrophotometry. The OASIS data reduction package has been developed to fully exploit the characteristics of a true spectro-imaging instrument; that is the imaging and photometric quality of an imaging device and the spectral quality of a spectrograph. It includes all the necessary tools to fully reduce OASIS data from raw CCD exposures to wavelength, flux calibrated and cosmic removed datacubes. Tools to export and import data from datacube to common formats are provided, as well as merging functions which allow the user to combine individual exposures in a higher S/N and/or larger field of view datacube. At present time, the supported image and table formats are FITS and MIDAS (95NOV). A graphical user interface has been built using TCL/TK. Although additional simple analysis tools are available, our objective is not to provide a complete data analysis package such as IRAF or MIDAS. Simple mouse interactions allow the user to display a spectrum at a given location in the field using the reconstructed image.

The complete package consist of 285000 lines of C and F77 code, plus 34000 lines of TCL/TK code and 9000 lines of HTML documentation. Binary distributions are available for various platforms (currently Sun/OS, AIX, HP-UX and Linux) and can be downloaded from the WEB server in Lyon. An email hot-line service is maintained, as well as necessary updates.

OASIS Performances

The purpose of this section is to give information about OASIS performances using the first scientific exposures obtained during the commissionning runs in April 98. Note that, because of the exceptionally bad seeing conditions (1 - 2.5 arcsec) which prevailed during most of the run, these results are not representative of what can be expected from OASIS in normal Mauna Kea seeing.

Spatial performance

The ability of an IFS to retrieve the morphological structures of an object is a good estimate of its overall spatial performances. We display in figure 3 the reconstructed image of six 30mn exposures of NGC 3377, obtained by summation of all spectra from the datacube over the whole wavelength range, as well as a 10mn direct image exposure through the same filter. The agreement between both images is excellent, showing that the image quality of the spatial stage (enlarger, field lens and lens array) is quite good and that data processing does not degrade the spatial resolution.

 

Figure 3:   Upper left panel: Reconstructed image of NGC 3377. Lower left panel: Direct image.Upper left panel: Stellar velocity dispersion field. Lower right panel: Stellar velocity field

 

Figure 4:   A comparison of major axis kinematical data of NGC 3377. Solid line: OASIS data. Symbol: SIS data

Spectrographic performance

Spectrographic quality is the ability of the instrument to recover precisely the spectral shape and wavelength information. A quantitative comparison can be done using absorption line kinematics. Measuring line of sight stellar velocities is a demanding task for wavelength calibration and spectral PSF, and thus a good measure of the spectrographic quality of the instrument. 2D stellar kinematics derived from the NGC 3377 exposures (spatial resolution 0.65 arcsec FWHM) is displayed in figure 3. The cut along the major axis agrees well with the kinematics obtained by Kormendy et al. (1998) with SIS ($\sim$0.5'' FWHM).

Spatial resolution

Finally the spatial resolution is a key factor. A good comparison can be done using higher spatial resolution image, such as the one given by HST. A comparison between HST narrow band image of the [OIII] emission in NGC 4151 nucleus and the OASIS [SIII] reconstructed image is given in figure 5 (panel A). Although the spatial resolution (0.35 arcsec FWHM) is clearly lower, most of the structures observed by HST are recovered by OASIS. However, the real advantage is seen in panels B and C of this figure, where the kinematics of the NLR is evident. Such results are complementary to HST narrow band WFPC2 images and STIS spectrograms.

 

Figure 5:   Panel A: OASIS reconstructed image of the SIII emission line superimposed on the HST WPC2 OIII image of NGC 4151. Panel B: Velocity slice of [SIII] emission at $V = -435 \pm 66 km.s^{-1}$ Panel C: Velocity slice of [SIII] emission at $V = +493 \pm 66 km.s^{-1}$

The ability of the instrument to detect faint diffuse structures nearby a bright ponctual source is also an important feature. It is directly related to the spatial resolution and the amount of diffuse light. A concrete example is given in figure 6. It shows the continuum and [SII] emission around DG Tau taken from an OASIS 30 mn exposure (0.9 arcsec FWHM) obtained by Cabrit and Lavalley. Despite the very bright continuum peak, the [SII] morphology and kinematics is measured down to a small distance of the star.

 

Figure 6:   Left: Continuum reconstructed image. Right: [SII] reconstructed image contours superimposed on the H$\alpha$ reconstructed image. (Image courtesy of Cabrit and Lavalley)

Throughput

Total throughput of the instrument, CCD, PUEO, telescope and atmosphere is 15% at F/20 and 20% at F/8 (average values). Two examples of transmission values are given in table 2. Note that the first tests of OASIS were done with the LORAL 3 CCD (and the 50-50 AOB beamsplitter) which has a lower efficiency than the EEV.

 

Config Name MR1 MR2 MR3

Central Wavelength (Å) 5200 6600 8750

Sky + Telescope 0.74 0.74 0.73

PUEO + Beamsplitter 0.73 0.76 0.82

OASIS 0.37 0.41 0.30

EEV CCD 0.78 0.74 0.34

Total F/20 0.16 0.17 0.06

Total F/8 0.21 0.22 0.07

Prospects

The instrument OASIS is the achievement of ten years of pionering work with TIGER which has largely contributed to the success of the integral field spectroscopy concept in the astronomical community. For some targets OASIS is able to provide spatial resolution competitive with HST but with the decisive advantage of 2D spatial coverage. On other targets where AOB will only be able to provide a modest gain in spatial resolution (or directly at F/8 focus), OASIS will still be very competitive, measuring simultaneously 1200 spectra over a homogeneously sampled sky area.

Acknowledgements.$\;$Many CFHT people were deeply involved in the successfull final integration of the instrument; William Rambold (now with Gemini), Gregory Barrick, Barney Magrath, Pierre Martin, Jim Thomas, Bernt Grundseth, Rohendra Atapattu are some of them. Pierre Martin and Gregory Barrick are respectively the OASIS instrument scientist and project engineer. Thanks to Sylvie Cabrit and Claudia Lavalley for the permission to use their DG Tau reconstructed images.