Abstract: The splash sound of target's water entry is an instantaneous signal, consisting of impact component and bubble components.It is difficult to detect and estimate the azimuth of this signal by traditional azimuth estimation methods using a single vector sensor,especially when the SNR is low.Empirical mode decomposition is a nonlinear analysis method which can emphasize signals' instantaneous characteristic.In this paper,the pressure and particle velocities are decomposed into different intrinsic mode functions (IMF),and the MAIA (mode acoustic intensity averager) method can realize the instantaneous signals' detection and azimuth estimation.The lake experiment and sea trial results show that this method can decompose water-entry sound and ambient interference into different IMFs,so as to detect the starting time of water-entry sound signal and estimate azimuth effectively.