Abstract: In the direction of arrival (DOA) estimation of coherent signals, when the coherent signals received by the array are at a low signal noise ratio, the DOA estimation performance will be significantly reduced. To address this problem, an enhanced spatio-temporal smoothing (ESTS) algorithm is proposed, which improves on the spatio-temporal smoothing (STS) technique using the spatio-temporal correlation matrix to reconstruct the received data matrix. First, the subarray spatio-temporal correlation matrix was squared preprocessed, and then by fully exploiting the covariance and mutual covariance information of the subarray spatio-temporal correlation matrix to decoherence, the resolution of the coherent signals as well as the robustness to the noise perturbation was enhanced. Both theoretical analysis and statistical results demonstrate that compared with other spatial smoothing class decoherence methods, the proposed method improves the decorrelation performance of DOA estimation of coherent signals with low signal noise ratio, fewer number of snapshots, and small angular separation.