5.2 Detection and Observation of Galaxies

At an advanced stage in the course of the mission feasibility studies, the idea of carrying out galaxy observations with GAIA was first put forward by [Høg, Fabricius, Knude and Makarov 1998a] and [Høg, Fabricius, Knude and Makarov 1998b], who suggested that galaxies could be detected as high-surface-brightness regions in the ASM and observed in the BBP with minimum impact on the expected accuracy for the observation of stars and on the telemetry^5.4. According to the ideas expressed in these papers, when an average surface brightness significantly in excess over the local sky background is measured in a square area of a few arcsecond size in the ASM1, then a galaxy has probably been detected and the samples covering this area and its surroundings in the BBPs should then be transmitted to the ground. Since, in so doing, whenever a galaxy was being observed it would not have been possible to observe stars, it was suggested that detection and observation of galaxies were to be carried out in one of the two Astros only, so as to let the other one to observe stars over the whole mission.

Some practical problems has to be solved in order to follow this general idea. One has to define:

• a local sky background with which the average surface brightness measured in the square areas can be compared. This may basically be determined from ASM1 samples of $2 \times 2$ pixels by trimmed median filtering. For instance, it is presently envisaged to transmit to the ground the values thus obtained over $12 \times 12$ arcsec$^2$. The median of these values over several degrees of scan may be used as reference against which comparison is made. However, the exact method used to determine the local sky background is not of interest for our purposes, since the readnoise is by far the dominant noise source.
• a reasonable strategy for the detection area and level so that useful data are transmitted without being swamped in less interesting data from the Milky Way or the zodiacal light. A trimmed median filtering technique such as that planned for the determination of the local sky background but averaging within a smaller area may be effectively used. The larger the detection area, the fainter the detection limit can be for objects of constant surface brightness, if the error on the sky background is negligible. On the other hand, the detection area should be small enough that a large number of small objects would not be missed.
• a useful sampling scheme for galaxy observations, so as to establish a trade-off between the angular resolution, the readnoise and the telemetry. As discussed in the previous Section in connection with the observations of stars, a larger sample size yields a smaller error on the (average) surface brightness, lower readnoise and telemetry rate but also a lower angular resolution. This aspect is critical for the observations of galaxies, since their potentially very large angular extension, requiring the readout of entire CCDs in some cases.

The most realistic, but time-consuming, approach to the solution of these problems requires devising and implementation of a suitable detection and observation strategy, followed by numerical simulations based on real fields imaged with an higher depth and resolution than it is achievable by GAIA. This approach was thoroughly and successfully followed in the planning of the observation of stars. As for galaxies, in this Chapter their detection and observation will be discussed from a statistical point of view only, and preliminary results from numerical simulations of galaxy detection will be summarized. The observation strategy outlined here, instead, was tested through numerical simulations described in Chapter 6.