Two mathematical representations of noise due to atmospheric turbulence are presented. These representations are derived and used in computer simulations of the Bartlett Estimate implementation of beamforming. Beamforming is an array processing technique employing an array of acoustic sensors used to determine the bearing of an acoustic source. Atmospheric wind conditions introduce noise into the beamformer output. Consequently, the accuracy of the process is degraded and the bearing of the acoustic source is falsely indicated or impossible to determine. The two representations of noise presented here are intended to quantify the effects of mean wind passing over the array of sensors and to correct for these effects. The first noise model is an idealized case. The effect of the mean wind is incorporated as a change in the propagation velocity of the acoustic wave. This yields an effective phase shift applied to each term of the spatial correlation matrix in the Bartlett Estimate. The resultant error caused by this model can be corrected in closed form in the beamforming algorithm. The second noise model acts to change the true direction of propagation at the beginning of the beamforming process. A closed form correction for this model is not available. Efforts to derive effective means to reduce the contributions of the noise have not been successful. In either case, the maximum error introduced by the wind is a beam shift of approximately three degrees. That is, the bearing of the acoustic source is indicated at a point a few degrees from the true bearing location. These effects are not quite as pronounced as those seen in experimental results. Sidelobes are false indications of acoustic sources in the beamformer output away from the true bearing angle. The sidelobes that are observed in experimental results are not caused by these noise models. The effects of mean wind passing over the sensor array as modeled here do not alter the beamformer output as significantl...