Dual-Branched Spatio-Temporal Fusion Network for Multihorizon Tropical Cyclone Track Forecast

• Published in: IEEE journal of selected topics in applied earth observations and remote sensing Vol. 15; pp. 3842 - 3852
• Main Authors: Liu, Zili, Hao, Kun, Geng, Xiaoyi, Zou, Zhengxia, Shi, Zhenwei
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Casualties; Computer applications; Cyclones; Data models; Deep learning; Disasters; Feature extraction; Forecasting; History; Hurricanes; Logic gates; Methods; Multimodal data fusion; Predictive models; Pressure field; spatio-temporal data; State vectors; Temporal variations; Three-dimensional displays; tracking forecast; Trajectory; Tropical climate; Tropical cyclones; tropical cyclones (TCs); Vectors
• ISSN: 1939-1404; 2151-1535
• DOI: 10.1109/JSTARS.2022.3170299

A tropical cyclone (TC) is a typical extreme tropical weather system, which could cause serious disasters in transit areas. Accurate TC track forecasting is the key to reducing casualties and damages, however, long-term forecasting of TCs is a challenging problem due to their extremely high dynamics and uncertainty. Existing TC track forecasting methods mainly focus on utilizing a single modality of source data, meanwhile, suffer from limited long-term forecasting capability and high computational complexity. In this article, we propose to address the abovementioned challenges from a new perspective-by utilizing large-scale spatio-temporal multimodal historical data and advanced deep learning techniques. A novel multihorizon TC track forecasting model named dual-branched spatio-temporal fusion network (DBF-Net) is proposed and evaluated. DBF-Net contains a TC features branch that extracts temporal features from 2-D state vectors and a pressure field branch that extracts spatio-temporal features from reanalysis 3-D pressure field. We show that with the abovementioned design, DBF-Net can fully exploit the implicit associations of multimodal data, achieving advantages that unimodal data-based method does not have. Extensive experiments on 39 years of historical TCs track data in the Northwest Pacific show that our DBF-Net achieves significant accuracy improvement compared with previous TCs track forecast methods.