Abstract: A real-time reconstruction method of sound velocity profiles based on mean sound velocity measurement is proposed, as the traditional reconstruction methods of sound velocity profiles are usually not real-time. Firstly, eigen decomposition is performed on historical data, and the modes of empirical orthogonal function (EOF) are chosen. Secondly, the EOF reconstruction coefficient is obtained by solving the minimum norm solution of linear equations in two unknowns based on the measured sound velocity at a certain depth under water and the mean sound velocity from this depth to the sea surface. Finally, sound velocity profiles of full ocean depth are reconstructed combining with historical sound velocity profiles. The method is simulated and verified with the reanalysis data of the western Pacific Ocean, it shows that the reconstructed sound velocity profiles are in good agreement with the real sound velocity profiles, and the mean root mean square errors is about 1.31 m·s⁻¹. The proposed method provides technical support for undersea vehicle′s real-time reconstruction of sound velocity profiles.