Aircraft trajectory prediction is a critical enabler for modern air traffic management, offering accurate estimations of future aircraft positions to enhance safety, efficiency, and predictability and supporting intelligent airspace operations. In this study, we propose a hybrid deep learning model that integrates convolutional neural networks (CNNs) and long short‐term memory (LSTM) networks to jointly capture spatial patterns and temporal dependencies from multivariate flight data. The CNN module extracts local spatial features from each historical time step, while the LSTM module models sequential dynamics to forecast the next 3D position in latitude, longitude, and altitude. Evaluated on the real‐world UST air traffic dataset comprising over 6000 inbound flights to Hong Kong International Airport, our method consistently outperforms competitive baselines across multiple error metrics and visualizations. The model achieves high prediction accuracy while maintaining a compact architecture suitable for real‐time applications, demonstrating the effectiveness of combining convolutional and recurrent structures for trajectory forecasting in structured airspace environments.
Song et al. (2026) studied this question.