Data collected using instrumentation mounted on 10 m micrometeorological towers on the roofs of buildings distributed across Washington, DC, and data derived from a dense surface network operated commercially (by AWS Convergence Technologies Inc.) indicate good agreement between the turbulence measured by the rooftop sonic anemometers and the total wind variance computed from a subnetwork of simpler surface stations located within a 5 km radius. This is even though the individual surface stations are not located such that their data can be considered representative, in the manner usually associated with the needs of numerical models. Limited data collected in New York City yield similar results. For Washington, DC, and New York City, the standard deviations of wind components in the roughness sublayer are apportioned (approximately equally) between temporal and spatial components in a coherent way that could enable ready exploitation of surface network data in dispersion models. The micrometeorological roof-top stations detect a total turbulence level that is close to the sum of the time and space components derived from the surface stations. The coupling of turbulence levels (as measured by standard deviations) between the local surface network and the wind aloft is far stronger than for the velocities themselves, for which ratios of average surface vector wind speed to the wind aloft average about 0.3 for Washington, DC, and about 0.4 for the two locations in New York City.