Matched filter optimization of kSZ measurements with a reconstructed cosmological flow field

We develop and test a new statistical method to measure the kinematic Sunyaev-Zel'dovich (kSZ) effect. A sample of independently detected clusters is combined with the cosmic flow field predicted from a galaxy redshift survey in order to derive a matched filter that optimally weights the kSZ signal for the sample as a whole given the noise involved in the problem. We apply this formalism to realistic mock microwave skies based on cosmological $N$-body simulations, and demonstrate its robustness and performance. In particular, we carefully assess the various sources of uncertainty, cosmic microwave background primary fluctuations, instrumental noise, uncertainties in the determination of the velocity field, and effects introduced by miscentring of clusters and by uncertainties of the mass-observable relation (normalization and scatter). We show that available data (\plk\ maps and the MaxBCG catalogue) should deliver a $7.7\sigma$ detection of the kSZ. A similar cluster catalogue with broader sky coverage should increase the detection significance to $\sim 13\sigma$. We point out that such measurements could be binned in order to study the properties of the cosmic gas and velocity fields, or combined into a single measurement to constrain cosmological parameters or deviations of the law of gravity from General Relativity.