Turbulent fluctuation of integrated water vapor in the troposphere is one of the major noise sources in radio interferometry. Processed integrated water vapor estimates from microwave radiometers colocated with interferometers have been used to set bounds on this uncertainty. The bound has been in the form of a calculated structure function, which is a measure of temporal or spatial decorrelation of fluctuations. In this paper a new model is presented for the estimation of the structure function in the absence of radiometer measurements. Using this model, the structure function can be estimated using measurements or estimates of a limited number of meteorological parameters. These parameters include boundary layer depth, surface heat and humidity fluxes, entrainment humidity flux, average virtual potential temperature in the boundary layer, and geostrophic wind speed. These parameters can be found or estimated from radiosonde and surface eddy correlation system data. The model is based on a framework of turbulence meteorology and provides excellent agreement when compared with state-of-the-art atmospheric turbulence simulations. Results of preliminary comparisons with ground truth show some excellent agreement, as well as some problems. The performance of the new model exceeds that of one current model. ¿ 1999 American Geophysical Union