REPORT ON THE MEETING OF THE OPTICON 3D SPECTROSCOPY WORKING GROUP

Held: 7-8 December 2000. ESO, Garching

 Jeremy Allington-Smith  j.r.allington-smith AT durham.ac.uk 
 Thomas Becker           tbecker AT aip.de
 Veronique Cayatte       veronique.cayatte AT obspm.fr 
 Yannick Copin           copin AT strw.leidenuniv.nl 
 Nicolas Devillard       ndevilla AT eso.org 
 Pierre Ferruit          ferruit AT obs.univ-lyon1.fr 
 Bianca Garilli          bianca AT ifctr.mi.cnr.it 
 Markus Kissler-Patig    mkissler AT eso.org 
 Evencio Mediavilla      emg AT ll.iac.es 
 Guy Monnet              gmonnet AT eso.org 
 Casiana Munoz Tunon     cmt AT ll.iac.es 
 Ralf Palsa              rpalsa AT eso.org 
 Alvio Renzini           arenzini AT eso.org 
 Martin Roth             mmroth AT aip.de
 Juergen Schmoll         jurgen.schmoll AT durham.ac.uk 
 Juergen Schreiber       schreib AT mpe.mpg.de 
 Lowell Tacconi-Garman   lowell AT mpe-garching.mpg.de 
 Matthias Tecza          tecza AT mpe-garching.mpg.de 
 Niranjan Thatte         thatte AT mpe-garching.mpg.de 
 Paul van der Werf       pvdwerf AT strw.leidenuniv.nl 
 Jeremy Walsh            jwalsh AT eso.org (Chair) 

Alvio Renzini opened the meeting. He highlighted the role of 3D spectroscopy and the possibilities of funding through the EU 5th Framework.

Martin Roth presented the euro3D proposal which had been submitted to the EU 5th framework programme as a Research and Technical Development project (RTD). The proposal was prepared in some haste and submitted in May 1999. The proposal (Martin to make an electronic version available) included software and scientific programmes with IFU's, mainly on galaxies. It failed to gain funding, but the comment was made that it would be better suited to a co-operation network. The suggestion was also made that it should be coordinated with OPTICON.

Alvio presented an outline of OPTICON - OPTical and Infrared CO-ordination Network for astronomy. It consists of 14 partners, both Institutes and national Funding Agencies, with a budget of 1Million Euro's over 4 years. The lead scientist is Gerry Gilmore in Cambridge. The aim of OPTICON is to promote workshops. So far there have been /will be meetings on Science Cases for extremely large telescopes, use of medium sized telescopes, archive infrastructure and others. There is funding for meetings but not mobility of researchers.

The session on 3D spectroscopy INSTRUMENTS proceeded.

Yannick Copin described the SAURON instrument which is a team instrument used on the WHT for the specific purpose of studying the velocity field of early-type galaxies. The current coverage is of the 4760-5400A region for H-beta, [O III] and Mg I lines. There are 1500 lenslets in the main array and 100 additional lenslets for offset sky. The data reduction proceeds smoothly and a pipeline is in preparation.

Evencio Mediavilla described INTEGRAL which a common user instrument on the WHT and connected to the WYFOS spectrometer. It has three spatial scales (0.45, 0.90 and 2.70''). The procedure to correct for differential atmospheric refraction was described. The instrument had been used for a variety of science programmes from Pre-Main Sequence objects to gravitational lenses.

Martin Roth described the Potsdam Multi-Aperture Spectrometer (PMAS) which uses a fibre coupled micro-lens array. Currently it uses a 16x16 array and the science goals are PN, SNR's and supergiants in Local group galaxies. PMAS is aimed at VLT and LBT. The emphasis is on excellent subtraction of the surrounding background to enable studies of line and continuum in point sources and small resolved objects. Martin compared the advance achievable with an IFU compared to a slit with the step from aperture photometry to CCD photometry.

Bianca Garilli described VIMOS and NIRMOS instruments and their IFU's. There are 6400 square microlenses with scale of 54x54''. The microlenses are coupled to four slits feeding each spectrometer. For higher spectral resolution only the central 24x24'' of the IFU and the central fibres in each slit are used. The zeroth order and 2nd order can overlap other spectra on the detector complicating the reduction; the 2nd order has upto 7% of the total flux.

Matthias Tecza presented SINFONI, which comprises the MACAO AO feed and the SPIFFI spectrometer. The SPIFFI design is based on the 3D instrument and slices the images into 32 slitlets, each of 32 spatial pixels and feeds this to a 1024x1024 detector. There are three pixel scales, 0.25, 0.1 and 0.025 with coverage of 8x8, 3.2x3.2 and 1x1''. The spectral resolution of 4500 in J,H,K is achieved by dithering the spectra in wavelength.

Jeremy Walsh introduced the FLAMES facility consisting of the Nasmyth corrector, the OzPoz fibre positioner and the GIRAFFE spectrometer. There is allowance for another spectrometer at a future date. There are 132 single object (MEDUSA) fibres, 15 mini IFU's (20 fibres over 3x2'') and an ARGUS IFU. The ARGUS IFU has 22x14 microlenses coupled to fibres with two scales covering 11.5x7.3 or 6.6x4.2''. There are two spectral resolutions of ~9000 and 28000 available with ARGUS and mini-IFU's.

Markus Kissler-Patig presented an overview of NAOS/Conica. Conica is fed by the Nasmyth AO system NAOS and has a 1x1k InSb array with pixel sizes 0.02 to 0.1''. It is an imaging instrument but also has grisms with resolutions of 50-500 and a Fabry-Perot for the K band. The FP finesse will be ~50 and the resolution 2000. Order separation is by blocking filters. No data reduction tools are yet anticipated for the FP which will be difficult to calibrate well as there will be ghost images, since the F-P plates are not wedged.

Juergen Schmoll presented a number of integral field projects which are being undertaken at Durham.
SMIRFS-IFU - a 72 element uncooled lenslet+fibre feed covering 4.1x5.9'' with fibres in steel tubes.
TEIFU - a 1000 element IFU for 14x15'' (with two x2 magnifications) for the WHT for the Electra Naomi AO system.
GMOS-IFU - two IFU's with 5.0x7.0 and 5.0x3.5 fields with 1000 and 500 elements respectively for object and sky for Gemini North.
IMACS IFU - similar to GMOS IFU for the Magellan telescope.
FMOS-IFU - 15-20 deployable small IFU's of 150-200 lenslets feeding fibres. The instrument will be used in the optical and IR (0.9-1.8micron) feeding floor spectrometers and is in plan for Subaru.
GNIRS-IFU - image slicer for Gemini NIR spectrometer. 21 slices giving 972 or 625 spectra with scales 0.04 and 0.15'' per spatial sample.
GIRMOS - pre-design for 32 3x3'' mini IFU's for 0.9-2.5micron with IFU arms in cryogenic area.
NGST NIRMOS - ESA instrument for NGST involving many groups. IFU for 46x40'' with 0.2'' samples or 3.8x2.6'' at 0.05'' samples using slicers, and feeding 3 spectrometers. The slicers may have aspherical surfaces (diamond turning on metal) or spherical surfaces for polished zerodur.

The session on 3D SOFTWARE followed.

Yannick discussed the XOasis and XSauron packages. Most of the background is in Yannick's thesis. Simple summing of spectra is not sufficient since there is substantial cross-talk between adjacent spectra and an extraction technique taking account of cross-talk by nearest neighbours is used. Physical models of the Oasis and Sauron instruments are used with tunable parameters to match the actual data. Flat fielding is done by lamp flats for the spectral flat fielding and sky flats for the spatial part. Cosmic rays are cleaned out after the extraction processing. The complex procedure of merging of offset fields is also handled. The is a GUI to run the software. The XOasis library is in C with Tcl/Tk and the cfitsio library is used. The input and output data can be in FITS or MIDAS. For display a modified version of the ESO Real Time Display is used. The extracted spectra, variances and associated table information to reconstruct the 3D data is held in a platform independent binary format. The type of processing performed to produce a particular file is recorded in its header. Since Oasis uses hexagonal pixels a FITS 3D format is not well suited so the stacked format is used with tools to create FITS output. A pipeline for SAURON reduction called Palantir is well advanced and uses the STScI OPUS pipeline to control processing.

Thomas Becker described the XPMAS reduction package written in IDL. The package allows full control or fully automatic reduction of PMAS data, but is general enough to be applied to other IFU data. The aim is extraction of high photometric quality. In MPFS there are 256 spectra on a 1x1k detector so the cross-talk is large. In addition the fibre-to-fibre transmission varies by up to 20%. IRAF tasks were found to be too time consuming and often failed so a user procedure to extract all the spectra simultaneously was applied. A demonstration was shown of extraction of a PN from the bulge region of M31. In particular the extraction of emission line spectra of point or small sources from extended emission was emphasized. Experiments had also been tried with two channel (spatial) deconvolution with the point source in one channel and the extended emission in another.

Bianca Garilli outlined the reduction software for the VIMOS IFU which is written as C programs and using some MIDAS tasks. It runs under Python and is written by A. Zanichelli. The package is structured as a pipeline but the individual routines can be put together into recipes. The input is FITS 2D images from each quadrant and the output is FITS 3D frames. The consortium has developed its own GUI which they may distribute. The construction of the images from the spectra uses a table of the IFU links. The relative transmission of the fibres is to be calibrated using sky lines (e.g. 5577A). The fibre transmissions are assumed independent of wavelength. There is no correction for cross-talk. The spectra are extracted by collapsing the cross dispersion profile, with profile weighting and are stored in a stacked format. In order to determine the sky the mode of the wavelength collapsed signals is used. There is also the possibility to use a separate sky spectrum stacked image. Aspects not yet covered are cosmic ray removal (the XOasis method is being considered) and scattered light correction.

Jeremy Allington-Smith reported on extraction software under development at Durham for lenslet arrays. A student, James Turner, is working on a package which is written for IRAF in SPP and C. Depending on the packing of the spectra on the detector there are two paradigms:
MOS - spectra independent so can use long slit techniques for individual spectra
Modified Longslit - spectra not independent.
The GMOS IFU (1500 spatial elements arranged in 2 slits) has 5 spatial pixels per spectrum and uses the MOS paradigm with each spectrum independently traced. Wavelength calibration, mapping of the spectra to the spatial pixels, datacube reconstruction and 3D interpolation are all performed. A fast white light image reconstruction is available to check telescope pointing. Data cube mosaicing (in spectral and spatial dimensions) is robustly performed. Background subtraction using either the spectra at the field edges or through a separate offset exposure are covered and also atmospheric dispersion correction. Durham, UK-ATC (Edinburgh), IoA Cambridge, and Starlink are coordinating a Joint Grant application for 3D spectroscopy support. See this web page for addresses and useful listings. Starlink are trying to bring the diverse 3D tools together in a uniform way.
Another project is CIRPASS with a 499 element IFU for the IR using lenslets and fibres. The software is being written by Rachel Johnson at IoA.

Veronique Cayatte reported on the software being written for GIRAFFE. There is a Baseline DRS written by the Observatoires of Geneve and Lausanne (C code with a Python wrapper) for the basic reduction of the MEDUSA, mini-IFU and ARGUS data. Specific tools for the IFU's and ARGUS will come in the Ancilliary Data Analysis System (ADAS) with tools for cube reconstruction from the stacked 1D format. It is proposed to use some radio astronomy tools such as channel maps for visualization. Basic analysis on the data such as line moments and equivalent widths will be available.

Nicolas Devillard presented the eclipse software package. The is an ANSI C library supported by ESO for basic image processing. It is used in the ESO Data Flow System for some instruments. There are no data analysis tools in eclipse. It features automated processing through dedicated basic tasks and is fast and reliable. The aim is to make it easy to maintain. It can also be used by astronomers at home for tuned reductions. It is written to coding standards with object-oriented design and is open source to enable users to contribute to its development. It is available for all POSIX platforms. It is currently used for ADONIS, Isaac, WFI and SOFI and will be used for NAOS/Conica, VISIR, OmegaCam. It is also in use at other observatories. In the future, it might offer some more extended functionalities, such as automated wavelength calibration, and will offer access to the library via a scripting language. See for details.

Juergen Schreiber described the reduction software planned for SPIFFI which is adapted from the 3D reduction software. The routines are written in ANSI C, are POSIX compatible and run in the HREF="http://www.astro.rug.nl/~gipsy/">GIPSY package. The routines can also be run under Python. The FITS I/O routines are taken from the eclipse package. There is a quicklook reduction to allow adjustment of telescope pointing, an on-line pipeline for best effort reduction and an offline version for full interaction. Bad pixel are handled through a bad pixel mask and tilted spectra are untilted. The spectral dithered data is formed by interleaving. Division by a reference star is performed with adjustment of the reference star spectrum using the ATRAN model atmosphere code. Cube mosaicing is performed. The basic data reduction (removal of instrument and atmospheric signature) ends with a cube. Following analysis is performed in GIPSY.

The discussion session followed.

euro3D

Europe currently has a lead over the US in the 3D Spectroscopy area and a network would exploit this lead. The aim is to provide tools for 3D spectra and pass the knowledge around in Europe. The telescopes and their instrumentation provide the "infrastructure". The guaranteed VLT time available to group members could be mentioned as a strength.

OPTICON had already highlighted the importance of IFU's for extremely large telescopes.

All agreed that 3D spectroscopy is applicable to all branches of astronomy, and that in most cases it is the method of preference for feeding a spectrometer. Almost all new instruments in the near future (with the exception of the mid-IR) will have some form of integral field device. Within 10 years it was suggested that over half of all spectrometry will be taken with IFU's.
It was agreed to include all forms of 3D spectra - from IFU's, FTS's and Fabry-Perots - in the proposal. The FTS instruments are well adapted for study of planetary atmospheres.

For the FLAMES instrument it was agreed to contact Francois Hammer to enquire if he wanted to be involved.

A question was raised about whether the international nature of some of the projects (e.g. Gemini) should be mentioned. Some advice was needed here. Claus Madsen has been appointed by the ESO DG as the ESO contact in Brussels and we will seek his advice.

It was agreed that projects depending upon and exploiting 3D spectroscopy to obtain excellent science should be presented in the Network proposal. However some advice needs to be sought as to whether this should be a major thrust of the proposal. A wide coverage of astronomy should be shown with specific topics in which the groups have expertise should be emphasized. Galaxy dynamics, AGN, distant galaxy formation, abundances of HII regions, PN and individual supergiant stars in the local universe and pre-main sequence evolution are the main themes. Solar System should be mentioned although there is no expertise in this area in the group. It was seen as important not to be too specific but on the other hand not to be shallow in coverage of individual topics. For each science project the specific tools to be developed would be identified.

Astronomical topics to be highlighted, with text required from the named members, for the next meeting:
Stellar populations (Martin)
Nearby supernovae (Emmanuel Pecontal)
High z galaxies (Paul)
Galaxy dynamics (Yannick)
Pre-Main Sequence Objects (Jeremy)
Nearby active galaxies (Pierre)
Lensing (Evencio)
Dwarf galaxies and groups (Veronique)
[Text in ASCII to be sent to Martin by February 9 2001.]

There was discussion of the handling of 3D data in existing packages especially for radio data and a worry that we should not be seen to be re-inventing the wheel. Lowell and Pierre are to come up with a matrix of the availability and extent of 3D spectroscopy software in data reduction packages by next meeting. Please send them suggestions and overviews of the packages you use (and URL's).

The major issues for the network to explore are clearly format and software to handle the data. All should send to Lowell and Pierre the format of the 3D data used or proposed for reduction of the various instrument data and how it is translated to common formats (e.g. FITS). Tools identified for development to be featued in the proposal (with text required from the named members for next meeting) were:
1) Visualization - e.g. 3D viewing (Paul)
2) Line fitting tool (Pierre)
3) Background subtraction (Martin)
4) 3D mosaicing (Lowell)
5) Crowded field spectrophotometry (Martin)
6) Data cube exploration /data mining (Veronique)
7) Guided deconvolution (Pierre)
8) Guided spectral analysis using emission & absorption lines (Yannick)
9) Object extraction (Bianca)

All agreed that the specific tools to remove the instrument signature should not be included in a proposal for funding. These tasks are already well taken care of by existing tools and are not suitable for a network approach.

Existing networks request 2 and 4 yr postdoc positions. The Very Large Telescopes Network has 32 staff years. There was not thought to be a penalty for asking for the maximum allowable. The minimum number of participating institutions is 5. Sub-contracting institutions can be named as such in the proposal, as can names of post-docs. Some advice about the balance between countries is needed. There was a question whether Observatories had to pledge their support as their facilities were involved.

As a network it should include a School at the half-way point at one at the end. "Less favoured areas" such as Canaries or East Germany may be chosen.

It was deemed important to have the support of OPTICON so a draft of the proposal should be ready for the OPTICON Board meeting in Sicily on 6 April 2001.

There should be a Web page for the 3D Spect. WG to share methods and algorithms. Possible also tools for format transformation. Petra Boehm in Potsdam will prepare a Web page (). Please send her your URL's for inclusion.

It was agreed that there should be an article sent to the ESO Messenger about the Working Group and its activities and also a brief presentation at the Scientific Drivers for ESO Future VLT/VLTI Instrumentation conference on 11-15 June. Jeremy to send a note to Guy Monnet.