We have made remarkable progress in recent years in our understanding of galaxy formation and evolution. Deep, high redshift observations have constrained the star formation history of the Universe and have unveiled the evolution of galaxy morphologies. These observations have helped to build and solidify our current model of hierarchical structure and galaxy formation - the Lambda CDM paradigm. Yet, the nature of look-back observations does not permit study of the evolution of individual galaxies: our understanding of the galaxy formation process is incomplete. Which high redshift galaxy building blocks end up in what kind of local galaxies? How much of the stellar content of the different galaxy components (bulge, thin and thick disk, halo) is created in situ and how much is accreted? How does the current accretion rate compare to Lambda CDM predictions? One of the best ways to address these questions is to look at the fossil record preserved in the resolved stellar populations of local galaxies.

The discovery of the Sagittarius Dwarf galaxy (Ibata et al. 1994) decisively demonstrated that satellite accretion is ongoing in the Milky Way. Observations of the stellar halo population of the Andromeda galaxy (M31) have revealed that it too contains a highly structured halo with low surface brightness loops and spurs (Ibata et al. 2001; Ferguson et al. 2002; Zucker et al. 2004a,b; Fig. 1). In contrast, M33 shows virtually no halo nor signs of accretion: it seems to consist of a pure quiescent disk formed in situ (McConnachie et al. 2004). Yet, despite the similarly violent histories of the Milky Way and M31 halos, they are remarkably different. The Milky Way halo seems to consist mainly of old, metal poor stars; while the halo of M31 contains a significant population of intermediate age stars (~50%) and overall has a higher metallicity (Brown et al. 2003). Clearly the formation histories of the halos of the Milky Way, M31, and M33 have been very different. A larger sample is therefore needed before we can draw general conclusions about disk galaxy assembly.

The GHOSTS (Galaxy Halos, Outer disks, Substructure, Thick disks and Star clusters) survey provides the definitive HST analysis of extra-planar stellar populations of nearby disk galaxies. Using deep ACS, WFPC2 and WFC3 images we target several edge-on galaxies covering a range of masses (Vrot=80-260 km/s). When put on the same physical scale, our observations reach the same depth and spatial resolution as the Sloan observations of M31, which revealed several tidal streams and the MV = -8 dwarf galaxy Andromeda IX (Zucker et al. 2004a,b). By sampling along the major, minor, and one intermediate axis, GHOSTS observations provide:

  
Figure 1: The spatial distribution of red giant branch (RGB) stars around M31 with an inset image of M31 to scale (Ferguson et al. 2002). Substantial substructure is observed and the halo is found to be quite flattened, quite unlike the textbook picture of a galaxy halo. The ellipse semi-major axis is 55 kpc. For reference we show our proposed aperture positions for NGC5907 when scaled to the distance of M31.

This survey thus represents the largest comprehensive study of resolved stellar populations in the outskirts of disk galaxies to date. The combination of our observations with high-resolution N-body simulations will undoubtedly yield significant advances in the study of disk galaxy assembly and evolution.