Abstract-Depth-averaged ocean current plays a significant role in marine scientific research, in particular, which is valuable for the navigation of underwater gliders. In this paper, we study the estimation and forecast of depth-averaged ocean current using underwater gliders. By considering three factors: the seawater density difference, the pressure hull compression deformation and the unstable depth intervals of a profile, we build a model for rapid calculation of underwater gliders' horizontal speed in order to improve the accuracy of the estimated depthaveraged ocean current. Then we adopt a novel machine learning model to forecast the depth-averaged ocean current of the next profile. Compared with two rough forecasting methods commonly used in engineering, our novel forecasting model has a better performance.