Background

DEIMOS has been in regular use at Keck since mid-2002 and is the primary optical spectrograph now employed on the Keck II telescope. This document for observers new to DEIMOS summarizes the key differences between observing with LRIS and observing with DEIMOS.

Preparation

You must be aware of several important considerations prior to your observing run.

Slitmask design software: DEIMOS observers use the IRAF-based DSIMULATOR package provided by Drew Phillips of UCO/Lick Observatory to select targets from an observers's astrometric catalog and generate the files necessary for milling DEIMOS masks. The output files are submitted through a web interface at UCO/Lick.
Slitmask submission deadlines: LRIS and DEIMOS slitmask are milled at the summit, but to keep the operation running smoothly we require that observers submit slitmask design files to Lick a minimum of four weeks in advance of the run unless valid scientific reasons (e.g., the need to observe a target of opportunity such as a supernoa) prevent this. When last-minute mask designs are submitted they must reach UCO/Lick by 3:00AM Pacific Time in order to be milled at Keck that day.

Performance

DEIMOS is performing well. Highlights of the instrument performance include:

Throughput is higher than anticipated due to conservative estimated that were used for the CCD sensitivity. The total system efficiency peaks near 30%, comparable to the LRIS performance. Please see DEIMOS Spectroscopic Throughput Measurements by Ricardo Schiavon for measurements of DEIMOS throughput.
Image quality is excellent, with the observed FWHM being limited by the delivered image from the telescope at essentially all times. Image quality of FWHM=0.5 arcsec is achievable under good seeing conditions.
Grating selection includes 600, 830, 900, and 1200-line gratings, of which any two can be used for observing on any given night. In addition, an imaging mirror is always available. Please note that significant ghosting is seen with the 830 and 900-line gratings.
Flexure is present in both for the TV guider and for the internal optics. An open-loop flexure compensation system is used with the guider to reduce flexure-induced image motion to 0.2 arcsec peak-to-peak (from 1.6 arcsec uncorrected). A closed-loop internal flexure compensation system (FCS) reduces internal flexure to < 0.25 pixels.
Readout of the science mosaic is currently about 40 seconds in imaging mode and 80 seconds in spectral readout mode.
Image size is 8k×8k pixels, or 150MB per raw image in spectroscopic (8-detector) format, and half that for the imaging (4-detector) format.
Motor moves are generally completed quickly and reliably. DEIMOS grating changes requires several minutes, comparable to the time required for an LRIS grating change. For that reason, observers trying to maximize efficiency will want to use zeroth-order imaging for slitmask alignment.

Operation

Slitmask alignment

The mask alignment procedure is very similar to the approach used for LRIS mask alignment, employing several alignment boxes 4 arcsec square. A variant of the xbox slitmask alignment software developed for LRIS is employed with DEIMOS. Rather than spending several minutes to switch between the grating and the mirror to take your alignment images, observers take images in zeroth order with the grating.

Data transport

Numerous options exist for transferring DEIMOS data from Keck to your home institution. In approximate order of popularity, they are:

burning your data onto DVDs (each DVD will hold about 30 DEIMOS images) using the autobackup software at Keck;
storing your data on your laptop's high-capacity hard disk;
transferring your data home via the Internet using ftp, scp, rsync, etc.

Please refer to the instructions on Writing Backups and FTPing Data for further information.

Data Analysis

Software requirements

The UCB DEEP team has developed a nearly complete IDL-based data reduction package for DEIMOS multislit spectroscopy which reduces the data to 2-D and 1-D extracted spectra, cosmic-ray cleaned and sky subtracted, without human intervention (usually). The pipeline works very well on data acquired with the 1200-line grating employed in the DEEP2 survey, reasonable well on the 600-line grating, and somewhat less well on the 830 and 900-line gratings. The pipeline properly subtracts tilted sky lines resulting from tilted slits, and the sky subtraction residuals are usually near the Poisson level.

This code is based on the magnificent SDSS spectral code of D. Schlegel. DEIMOS observers are welcome to use this code, and all DEIMOS observing accounts at Keck are configured to run the pipeline. The code is well documented and the DEEP2 team has generously provided both an online primer and a corresponding cookbook for DEIMOS observers to use.

If you do use this code, beware that it creates standard FITS files for all its output, mostly in compressed gz files (which save a factor of 2.4 in disk space and are in fact faster to read than uncompressed files). The problem is that many existing reduction packages cannot read FITS binary tables, nor can they read the compressed files. Beware, or consider switching reduction platforms. The FITS files are usually multi-HDU, and in IDL the retrieved HDU's become structures. This would be a good opportunity to learn about structures (supported in Fortran-90 and C) if you are not yet familiar with them.

Those intending to use IRAF to reduce their DEIMOS data will need to upgrade to version 2.12 (release in May 2002) in order to work with the DEIMOS data. Familiarize yourself with the tasks in the MSCRED package.

Hardware requirements

DEIMOS generates a healthy data volume (about 1 Gbyte/hr) and and since no computers currently at Keck are fast enough to process DEIMOS images with the pipeline software in "real time", you will most likely end up doing your reductions at home. Suggested requirements for running the pipeline are 1 Gbyte RAM, preferably 2. Expect it to require several hours of processing to fully reduce the data from 3 exposures on a single mask.