Short-term prediction of tropical cyclone track and intensity via four mainstream deep learning techniques: Fid-Bau Portal

Short-term prediction of tropical cyclone track and intensity via four mainstream deep learning techniques

Gan, S.L. / Fu, J.Y. / Zhao, G.F. / Chan, P.W. / He, Y.C.

Abstract Tropical cyclone (TC) is a highly destructive natural disaster, whose impact is closely correlated to its track and intensity. Thus, it is crucial to obtain the information on TC track and intensity both timely and accurately for early warning and mitigation of related disasters. While there are numerous prediction methods, deep learning (DL) technology has demonstrated exceptional potential in recent years. However, studies on TC prediction via DL techniques are still insufficient, and in particular there is a lack of efforts toward the dependence of prediction results upon some key parameters involved in the prediction strategy. To this end, this paper present a comprehensive study on TC prediction via four mainstream DL techniques, i.e., Long Short-Term Memory (LSTM) network, Convolutional LSTM (CovnLSTM) network, Temporal Convolutional Network (TCN), and the recently developed Transformer model. The performance of different DL techniques is examined, with a highlight on the effects of three setting parameters involved in the prediction process on the model performance, i.e., length of input records N in, interval between input records ΔT, and lead time of the prediction N out. Results reveal that TCN performs best from the perspectives of prediction accuracy, running efficiency and stability, followed by ConvLSTM; whilst LSTM and Transformer fail to provide competitive predictions, although Transformer showcases notable training efficiency. In reference to the prediction strategy, increasing ΔT and N out can consistently enlarge prediction errors across all models, while the model performance can be optimized by setting the value of N in in a range of 3–6. It is interesting to find that the overall performance of all models can be improved noticeably by using records with a finer ΔT in the scenarios with the same lead time but with varied N out. Meanwhile, TCN can be further improved by extending N out from N out = m to N out = n (n > m) and using the m-th step prediction in the scenario with N out = n as the output in the scenario with N out = m. The presented results provide useful insights for reasonably utilizing DL models to deal with prediction issues, which are applicable not only for TCs but also for other objectives that are widely concerned in wind engineering, e.g., wind speed/pressure and structural responses.

Highlights • Explored 4 mainstream deep learning models for short-term forecasting of typhoon track and intensity. • Examined the dependence of model performance on 3 key parameters involved in prediction strategy. • TCN outperforms others from various perspectives, while predictions via Transformer and LSTM suffer from larger errors. • Model performance can be improved by using appropriate strategy even based on the same database.