First Light on the Sloan Digital Sky Survey
by Gillian Knapp
The Sloan Digital Sky Survey obtained First Light on the night of May 9-10 1998. Conditions were not ideal; the Moon was full and the telescope light baffles were not in place, so that it was impossible to see faint things. Nevertheless, these observations demonstrated that the enormous imaging camera, the very wide field telescope, and the complex computer codes which extract and store the data all worked. Two weeks later, the camera was again placed on the telescope in the dark of the Moon, this time with the light and wind baffles in place. These data approach survey quality: the images are good across the field, the sensitivity is within about a magnitude of design, and the position accuracy likewise. Further, a small amount of the data were successfully run through the automated data reduction software to produce data catalogues, "ready for use".
While First Light was almost literally an overnight success, it was almost ten years of painstaking, careful, brilliant, intense work in the making. All of these components had to be designed and constructed basically from scratch; essentially nothing was off the shelf. Every component was tested exhaustively before being shipped to Apache Point Observatory. Preparations for First Light began in earnest about a year ago, when Chicago optical engineer Shu-I Wang and Fermilab astrophysicist Jim Annis moved to APO with their newborn twin boy and girl to begin final shakedown of the Data Acquisition systems constructed at Fermilab by a team led by Don Petravick and Ron Rechenmacher. A Fermilab team led by Bill Boroski (Angie Prosapio, Steve Bastian, Charlie Briegel, John Anderson, Paul Czarapata and Glenn Federwitz) began the task of installing the telescope drive and control system as the University of Washington team (Pat Waddell, Walter Siegmund, Ed Mannery, French Leger, Larry Carey, Fritz Toevs, Siri Limmongkol and Russ Owen) installed the major telescope optical components, to be followed by the light and wind baffling system. Things really heated up in December 1997 when the huge mosaic camera was shipped from Princeton and met at APO by its construction team, led by Jim Gunn (Princeton). While Gunn and his team (Mike Carr, Princeton, mechanical engineer; Connie Rockosi, electronic engineer and Chicago graduate student; Ernst de Haas, Princeton, engineer; Mamoru Doi, Tokyo; and Maki Sekiguchi, Tokyo) installed and tested the imaging camera, software scientists Robert Lupton (Princeton) and Zeljko Ivezic (Princeton) worked with Annis on writing monitoring software and installing data acquisition and data reduction software for the camera. Over the next few months, as the telescope components were installed one by one, the team tested, measured and shook down the camera. At the beginning of May, the secondary mirror was installed, and the telescope worked essentially at once. First light was achieved after several intense days and nights of very hard work, as problems were discovered, diagnosed and fixed in a fury of effort. During the first dark run, intense effort on the software led by Lupton resulted in running the data essentially all the way through the complex software reduction systems and producing images and catalogues of astronomical objects to a limit 100 times fainter than the previous all-sky survey, the Palomar Observatory Sky Survey. This effort was supported by Data Acquisitions scientists Jon Bakken and Don Holmgren at Fermilab, APO staff Dan Long, Jon Brinkman and Bruce Gillespie, and orchestrated by Project Manager Jim Crocker (Johns Hopkins University).
Although the mosaic camera takes the data, it relies for its calibration on two other instruments. The position calibration is achieved by reference observations of standard stars made with a pair of 11-CCD arrays at the leading and trailing edges of the camera. The position catalogues and the analysis software were produced by scientists at the U.S. Naval Observatory, and Jeff Pier, Jeff Munn and Greg Hennessy of the USNO were at APO to help put the data through the position- measuring (astrometric) software and check the results. These were very encouraging; the astrometric performance of the telescope is already within a factor of two of the best we thought we could do, although it remains to be seen how well astrometry works when the telescope is tracking.
The second necessary calibration is of the magnitude scale, and this is performed by a small second telescope which continually measures stellar magnitudes to faint levels, within reach of the imaging camera, relative to the magnitudes of standard stars. The calibration observations were made by Russ McMillan (APO) and John Briggs (Chicago) and analyzed by Ivezic and Fermilab astronomers Doug Tucker and Heidi Newberg. These data show that the 2.5 m telescope/mosaic camera are within about a magnitude of their desired sensitivity. The construction of the calibration telescope was done by Tim McKay and Allun Smith (University of Michigan).
Activities in the next few months will be to make final adjustments to the telescope and camera optics, finish rewriting the software so that it optimally deals with real data and is more robust, and to investigate in detail the data already taken. While these are not quite of "survey quality" they will support a lot of science analysis as well as analysis to understand and improve the instrumentation and software. Survey observations proper will begin in August-September 1998.