Abstract: This paper presents a method combining B-splines and autoregressive moving averages for the joint correction of sound velocity disturbances, addressing the problem of existing parameterized models for sound velocity errors in underwater geodetic positioning without considering the temporal correlation of sound velocity disturbances. Initially, a quadratic polynomial and a cubic B-spline model are utilized to preliminarily correct the disturbed sound speed structure. Subsequently, considering temporal correlation variations, sound speed error processing is conducted based on the autoregressive moving average model. Finally, the method is validated using data from the 3000-meter sea trial in the South China Sea. The results indicate that, compared to the quadratic polynomial and cubic B-spline sound speed corrections, applying the autoregressive moving average model for successive sound speed correction reduces the root mean square error of time observation value residuals by 58% and 30%, respectively.