Technical and Scientific Preparation

For a more complete review of the ROSAT deep surveys see [32]. The detailed preparation for ROSAT Deep Survey observations started in 1984, well in advance of the actual ROSAT mission [87]. At this time the engineering model of the ROSAT Proportional Counter PSPC [66] had already been calibrated in the laboratory and the task was at hand to understand and try to correct for the various electronic and geometric distortions and gain nonlinearities, which in general plague imaging proportional counters. Nonlinearities in the performance of the Einstein IPC have, e.g., set the final sensitivity threshold for Einstein deep surveys [30, 81]. The ROSAT PSPC, even at the raw coordinate level, shows a higher degree of uniformity than the IPC. It was therefore possible to largely correct the significant distortions present in the PSPC data. Figure 25.1 (left) shows the final PSPC ground calibration flat field after correction of the distortion effects. This image immediately indicates another geometrical problem. While it is obviously possible to remove the distortions created by the detector itself, the shadows cast by the complicated wire-mesh in front of the PSPC cannot be corrected for. In the early days we were still hoping that the satellite pointing instability would wash out any residual flat-field inhomogeneities.

Fig. 25.1 The ROSAT PSPC flat-field. Left: in detector coordinates. The web-like shadow is due to the mechanical support structure of the PSPC window and wire mesh. Right: in sky coordinates after the application of the wobble mode

In 1985 we also assembled an international team of astronomers interested in deep X-ray surveys. Riccardo Giacconi, together with Richard Burg, at that time both at STScI, brought the Einstein experience into the team and originally suggested the deep survey in the Lockman Hole (see further) as our prime study area. This hole had just been discovered as the direction with an absolute minimum of interstellar hydrogen column density [46]. Gianni Zamorani brought the experience and the data of the deep optical survey in the Marano field, which we defined as a second study area in the southern hemisphere [97]. Maarten Schmidt, who had previously modeled [72] the contribution of various classes of sources to the X-ray background, was planning the optical identification work in the North, first using the Palomar 200" and later the Keck telescopes. Joachim Triimper, Gisela Hart-ner, and myself were responsible for the ROSAT data, i.e., calibration, simulation, and finally observation and analysis. For more than a decade this group met regularly at 1/2-1 yr intervals. We recall these meetings as extremely productive but also quite exhaustive and often with violent disputes. In particular, we all appreciated Riccardo's rigorous and constructive criticism.

Roughly at the same time we started prototyping the ROSAT scientific analysis software. Early ideas about local and map-detect algorithms, background estimation, etc. were taken over from the Einstein analysis system. However, substantial improvements were incorporated, and the most important of which was probably the rigorous application of Poisson statistics and maximum likelihood estimators in all statistical computations.

To test and understand the analysis algorithms we developed a science simulator system, which turned out to be one of our most powerful tools in the preparation of the science mission. During the course of time the simulation models became more and more realistic, including extended and point sources, time variability, different spectral models, and realistic number counts for the X-ray sources as well as cosmic and solar scattering backgrounds. Orbital variation of the exposure time and background components due to the radiation belts were included as well. Every individual photon was traced through a realistic model of the X-ray mirror system and the detector. Comparing input and output information, the detection algorithms could be tested, calibrated and, if necessary, improved.

In 1985, the first data about the performance of the ROSAT attitude control system became available from dynamical hardware-in-the-loop tests. It turned out that the attitude control was much more stable than specified and anticipated. Simulations including realistic attitude data indicated that sources in the center of the PSPC field could easily get lost behind shadows of the PSPC support grid. Less than a year from the originally planned launch (1987) we were able to convince the funding agency and industry to introduce a "wobble mode" into the attitude control system, which later was to become the standard ROSAT pointing mode. The idea was adapted from the optical drift scan technique previously developed by Maarten Schmidt's team. Instead of wandering around at random, the spacecraft pointing direction was guided smoothly into a linear periodic zig-zag motion with a period of 402 s and an amplitude of ±3 arcmin (± 1 arcmin for the HRI) diagonal with respect to the detector coordinates. The wobble mode, while introducing artificial periodic power into the measured light curves of X-ray sources, turned out to be very valuable for the study of the X-ray background or other diffuse sources. Figure 25.1 shows the PSPC flat-field expected after the wobble motion has been applied to it. The rms variation of the wobbled PSPC flat field is about 3% [31]. The idea of the wobble mode was later picked up by the Chandra Observatory, which uses a Lissajous figure pointing drift as standard observing mode.

0 0

Post a comment