Abstract: Time synchronization is a key factor of synergistic observation in underwater acoustic wireless network. Clock skew of underwater sensor nodes may change with the environment. To deal with this issue, a clock skew tracking method in underwater acoustic network is proposed. This method uses multiple models to describe the variable clock skew. The existing time synchronization protocols, such as time synchronization for high latency (TSHL) acoustic networks and Tri-message, are first used to calculate the clock skew data, and then the interactive multi-model Kalman filter algorithm is put forward to estimate and track the variable clock skew. At the same time, Sage-Husa adaptive method is also used to adjust the filter parameters dynamically in the filtering process to improve the accuracy of the algorithm. Simulation results show that the proposed tracking method can reduce the mean square error from 3.0×10^-9 to 5×10^-10, and the performance of the algorithm is better than that of the existing time synchronization methods.