Autonomous underwater vehicles (AUVs) are mobile platforms with excellent maneuverability, which are preferred for micro scale and meso scale ocean feature tracking in a near real time mode. With the feature extract analysis of ocean phenomena, autonomous control and tracking strategy of multi AUVs are designed for ocean feature tracking. Firstly, formation control is designed coordinately with the estimation of the ocean feature, and EKF (extended Klaman filter) is applied for ocean feature scalar estimation from the sample process noise and sample instrument noise. Secondly, the virtual leader-follow mode is utilized for multi AUVs formation control, and formation rotational is designed to adapt the variation of ocean phenomena. Lastly, the simulation platform is established for multi vehicles ocean feature tracking, and the simulation results are selected to verify the effectiveness of autonomous tracking strategy.