Abstract: Due to the poor performance of the conventional propagator method (PM) algorithm based direction of arrival (DOA) estimation algorithm of hydrophone array in the case of low signal to noise ratio (SNR) or small number of snapshots, an improvement of this method is proposed in this paper. In this improved method, the covariance matrix is expanded and reconstructed by using the rotation invariance characteristics of the signal subspace, and the propagator matrix is obtained by the submatrices of the partitioned covariance matrix. The extended noise subspace is constructed by the propagator matrix, and then the spatial spectrum is estimated by using the orthogonality between the signal subspace and the noise subspace. The simulation and lake experimental results show that, the proposed method has better robust estimation performance in the cases of low SNR and small number of snapshots. Compared with conventional PM algorithm, the root mean square error (RMSE) of the proposed method is reduced by 0.6° when the SNR is 0 dB, and the RMSE of the proposed method is reduced by 0.1° when the number of snapshots is 150.