Accurately predicting the long-term trajectory of a surface drifting buoy (SDB) is challenging. This paper proposes a promising solution to the SDB trajectory prediction based on artificial intelligence (AI) technologies. Initially, a scalable mathematical model for trajectory prediction is developed, transforming the challenge of predicting trajectory points into predicting velocities in eastward and northward directions. Subsequently, a four-layer trajectory prediction calculation framework (FLTPCF) is established, outlining a complete workflow for the real-time online training of marine environment data and SDBs’ trajectory prediction. Thirdly, for facilitating accurate long-term trajectory prediction, a hybrid artificial neural network trajectory prediction model, named CNN–BiGRU–Attention, integrates a Convolutional Neural Network (CNN), Bidirectional Gated Recurrent Unit (BiGRU), and Attention mechanism (AM), tuned for spatiotemporal feature extraction and extended time-series reasoning. Extensive experiments, including ablation studies, comparative analyses with state-of-the-art models like BiLSTM and Transformer, evaluations against numerical methods, and adaptability tests, were conducted for justifying the CNN–BiGRU–Attention model. The results highlight the CNN–BiGRU–Attention model’s excellent convergence, accuracy, and generalization capabilities in predicting 24, 48, and 72 h trajectories for SDBs with varying drogue statuses and under different sea conditions. This work has great potential to promote the intelligent degree of marine environmental monitoring.

Ocean moored buoys are essential ocean monitoring devices that are permanently moored in the sea to collect real-time hydrological and meteorological data. In response to the anomalies and missing data in datasets collected from ocean moored buoys, this paper innovatively established an intelligent quality control Transformer-Encoder-BiLSTM model. This model can impute missing data and identify anomalies in buoy datasets. The model first uses the multi-head attention mechanism of the Transformer Encoder to extract global features from time-series data of buoy observations. Subsequently, it utilizes the BiLSTM network for temporal reasoning training to capture dynamic changes within the time series, predicted data. Finally, using the predicted data as a benchmark, the model conducts anomaly detection, fills in missing values, and rectifies stuck values. We conducted a series of comprehensive experiments, with the data from Buoy No. 0199 in Qingdao, China as an illustrative example. The experimental results indicate that the performance indicator R² of the model is above 0.9, the accuracy of quality control is above 97%, while both precision and recall are above 84%. The F1 scores range between 81.61% and 90.09%. These experiments demonstrate that this method exhibits high accuracy and efficiency in filling in missing data, rectifying stuck values and identifying anomalous data, showing broad application potential.