Water vapor column irregularities within the lower troposphere's convective boundary layer constitute a nearly passive tracer of turbulent air motions and eddy convection. Remote sensing of turbulence in this layer from the ground is done with microwave radars and with lidars. Microwave interferometry is also sensitive to vapor column fluctuations via the known relative refractive delay of water vapor compared to dry air. However, interferometric measurements are intrinsically nonlocal, yielding only the baseline difference (the difference of measurements at two ends of an interferometer baseline) of vapor column, and thus are not amenable to classical data-reduction methods for fluid turbulence, other than the equal-time structure function. In order to remove this limitation of interferometry we have developed a method to estimate the time-lagged cross covariance of vapor-column fluctuations over two sites using interferometry data. This effectively circumvents the limitation to baseline-differenced vapor column data at least in statistical second moments. The method is illustrated with 15-GHz data taken by the very large array.