Fluctuating atmospheric emission is a dominant source of noise for ground-based millimeter-wave observations of the cosmic microwave background (CMB) temperature anisotropy at angular scales ≳0.5°. We present a model of the atmosphere as a discrete set of emissive turbulent layers that move with respect to the observer with a horizontal wind velocity. After introducing a statistic derived from the time-lag dependent correlation function for detector pairs in an array, referred to as the pair-lag, we use this model to estimate the aggregate angular motion of the atmosphere derived from time-ordered data from the Atacama Cosmology Telescope (ACT). We find that estimates derived from ACT’s CMB observations alone agree with those derived from satellite weather data that additionally include a height-dependent horizontal wind velocity and water vapor density. We also explore the dependence of the measured atmospheric noise spectrum on the relative angle between the wind velocity and the telescope scan direction. In particular, we find that varying the scan velocity changes the noise spectrum in a predictable way. Computing the pair-lag statistic opens up new avenues for understanding how atmospheric fluctuations impact measurements of the CMB anisotropy.