Enhancing tropical cyclone intensity estimation via multi-scale spatial feature correlation
Year: 2026
Type: article
Abstract: Tropical cyclones pose significant threats to coastal regions, necessitating accurate intensity estimates for effective disaster management. Traditional methods that rely on manual feature extraction suffer from subjectivity and limited scalability. This study introduces a multi-scale spatial feature correlation network for tropical cyclone intensity estimation. The model combines a deformable input block, residual feature extraction, squeeze-and-excitation channel recalibration, and strip pooling to represent irregular cloud morphology, inter-channel dependencies, and long-range spatial context. On the TCIR test set, the proposed model achieved a Root Mean Square Error (RMSE) of 8.46 kt, compared with 8.68 kt without channel-specific smoothing. In the comparison experiments, SATCON and 3DAttentionTCNet yielded RMSEs of 9.21 and 9.48 kt, respectively.
Source: Scientific Reports
Authors: Jin Han, ZhengKang Zhu
Language: English
Subfield: Atmospheric Science
Field: Earth and Planetary Sciences
Domain: Physical Sciences
SDG: Climate action
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W4417048309
