This report describes an efficient, autocorrelation based algorithm for estimating low altitude radial winds using signals from the two receiving beams of an airport surveillance radar (ASR). The approach seeks to achieve the accuracy demonstrated previously for spectral domain dual beam velocity estimators with significantly reduced computational requirements. Fundamental to the technique is the assumption that the power spectrum measured with an airport surveillance radar's broad elevation beam can be fitted by a two component Gaussian model. The parameters of this model are estimated using measured low-order autocorrelation lags from the low and high beam received signals. The desired near surface radial velocity estimate is obtained directly as one of these parameters -- the center frequency of the "low altitude" Gaussian spectrum component. Simualted data and field measurements from Lincoln Laboratory's experimental ASR-8 in Huntsville, Alabama were used to evaluate the accuracy of the autocorrelation based velocity estimates. Monte Carlo simulations indicate that biases relative to the near surface outflow velocity in a microburst would be less than 2.5 m/s unless the microburst were distant (range > 12 km) or very shallow (depth of maximum wind speed layer < 50 m). Estimate standard deviations averaged 0.5 m/s after the spatial filtering employed in our processing sequence. The algorithm's velocity estimate accuracy was sufficient to allow for automatic detection of measured microbursts during 1988 with a detection probability exceeding 0.9 and a false alarm probability less than 0.05. Our analyses indicates that the dual-beam autocorrelation based velocity estimator should support ASR with shear detection at approximately the same level of confidence as the low-high beam spectral differencing algorithm evaluated by Weber and Noyes (1988).