Theoretical performance capabilities of the optimum filter for maximization of signal‐to‐noise ratio (SNR) and a suboptimum approximation, easy to realize in practice, are compared to the conventional matched filter (designed for white noise only) in the case of low Doppler signals masked by surface reverberation. Spectrum models for surface backscattering as a function of local environmental parameters are based on recent results describing scattering from a subsurface bubble layer. Significant improvements in SNR in the low Doppler region as well as a dramatic dependence of performance on windspeed and direction, and the ratio of reverberation to uncorrelated noise power are predicted. Results serve as average performance limits to be compared to those obtained by algorithmic implementations of optimum and adaptive filters.