Tidal data reflects the changes of coastal sea level, and plays an important role in many fields. Tidal data missing brings inconvenience to data usage. Based on the tidal data of Longhai and Xiamen Ocean Stations in 2020, this paper proposed a filling method based on Gated Recurrent Neural Network (GRU). Compared with traditional interpolation methods such as linear interpolation and spline interpolation, the GRU method is stable, accurate and convenient to use. Especially when the length of missing data is large, GRU method is obviously superior to the traditional interpolation method. At the same time, this method is also suitable for filling other missing time series data, such as water temperature.