Different methods for estimating signal coherence from underwater acoustic data taken in shallow water are compared. Specifically, this study details methods to estimate coherence from narrow-band, high-frequency, short pulse length, cw signals that typify the signals used by mine-hunting sonars in shallow-water environments. Due to the short record length of the received pulses, the spectral frequency resolution was poor using classical Fourier techniques. This motivated alternate approaches for the spectral estimation and coherence estimates. The other methods compared are an autoregressive parametric based approach, a harmonic wavelet approach, and concatenation of successive pulses followed by the classical method.