Forecasting the accuracy of velocity-field reconstruction

Joint analyses of the large-scale distribution of galaxies, and their motions under the gravitational influence of this density field, allow powerful tests of the cosmological model, including measurement of the growth rate of cosmic structure. In this paper we perform a statistical comparison between two important classes of method for performing these tests. In the first method we measure the 2-point power spectra between the velocity and density tracers, and jointly fit these statistics using theoretical models. In the second method we use the density tracers to reconstruct a model velocity field through space, which we compare with the measured galaxy velocities on a point-by-point basis. By generating an ensemble of numerical simulations in a simplified test scenario, we show that the error in the growth rate inferred by the "reconstruction and scaling" method may be under-estimated, unless the full covariances of the underlying and reconstructed velocity fields are included in the analysis. In this case the inferred growth rate errors agree with both the power spectrum method and a Fisher matrix forecast. We provide a roadmap for evaluating these covariances, considering reconstruction performed using both a Fourier basis within a cuboid, and a Spherical Fourier-Bessel basis within a curved-sky observational volume.