Design and analysis of redshift surveys

In this paper we consider methods of analysis and optimal design of redshift surveys. In the first part, we develop a formalism for analysing galaxy redshift surveys which are essentially two-dimensional, such as thin declination slices. The formalism is a power spectrum method, using spherical coordinates, allowing the distorting effects of galaxy peculiar velocities to be calculated to linear order on the assumption of statistical isotropy but without further approximation. In this paper, we calculate the measured two-dimensional power for a constant declination strip, widely used in redshift surveys. We present a likelihood method for estimating the three-dimensional real-space power spectrum and the redshift distortion simultaneously, and show that for thin surveys of reasonable depth, the large-scale 3D power cannot be measured with high accuracy. The redshift distortion may be estimated successfully, and with higher accuracy if the 3D power spectrum can be measured independently, for example from a large-scale sky-projected catalogue. In the second part, we show how a 3D survey design can be optimized to measure the power spectrum, considering whether areal coverage is more important than depth, and whether the survey should be sampled sparsely or not. We show quite generally that width is better than depth, and show how the optimal sparse-sampling fraction f depends on the power P to be measured. For a Schechter luminosity function, a simple optimization fP ≃ 500h Mpc is found.