Designing future dark energy space missions

Context. Future dark energy space missions such as JDEM and EUCLID are being designed to survey the galaxy population to trace the geometry of the universe and the growth of structure, which both depend on the cosmological model. To reach the goal of high precision cosmology they need to evaluate the capabilities of different instrument designs based on realistic mock catalogs of the galaxy distribution. Aims. The aim of this paper is to construct realistic and flexible mock catalogs based on our knowledge of galaxy populations from current deep surveys. We explore two categories of mock catalogs: (i) based on luminosity functions that we fit to observations (GOODS, UDF, COSMOS, VVDS); (ii) based on the observed COSMOS galaxy distribution. Methods. The COSMOS mock catalog benefits from all the properties of the data-rich COSMOS survey and the highly accurate photometric redshift distribution based on 30-band photometry. Nevertheless this catalog is limited by the depth of the COSMOS survey. Thus, we also evaluate a mock galaxy catalog generated from luminosity functions using the Le Phare software. For these two catalogs, we have produced simulated number counts in several bands, color diagrams and redshift distributions for validation against real observational data. Results. Using these mock catalogs we derive some basic requirements to help design future Dark Energy missions in terms of the number of galaxies available for the weak-lensing analysis as a function of the PSF size and depth of the survey. We also compute the spectroscopic success rate for future spectroscopic redshift surveys (i) aiming at measuring BAO in the case of the wide field spectroscopic redshift survey, and (ii) for the photometric redshift calibration survey which is required to achieve weak lensing tomography with great accuracy. In particular, we demonstrate that for the photometric redshift calibration, using only NIR (1–1.7 μm) spectroscopy, we cannot achieve a complete spectroscopic survey down to the limit of the photometric survey ($I