The research team led by Professor Zhang Changjiang from the School of Electronics and Information Engineering (Big Data Science) at TU, in collaboration with the Shanghai Typhoon Institute of the China Meteorological Administration, has recently made significant progress in estimating the structural scale of tropical cyclones. The related work, titled “Estimation of Tropical Cyclone Size by Combining Sequential Infrared Satellite Images with Multitask Deep Learning,” was published in IEEE Transactions on Geoscience and Remote Sensing (TGRS).

IEEE Transactions on Geoscience and Remote Sensing is a flagship journal of the IEEE Geoscience and Remote Sensing Society (GRSS), ranking among the top five international journals in geoscience with an impact factor of 7.5 (JCR Q1). Professor Zhang is the first author and corresponding author of the paper. Geng Bingfan from the same school is a co-first author. Collaborators include Researcher Ma Leiming and Researcher Lu Xiaoqin from the Shanghai Typhoon Institute. TU is listed as the first and sole corresponding affiliation.

Schematic diagram of the multi-task deep learning model integrating physical auxiliary tasks for tropical cyclone structural scale estimation

Current methods for estimating tropical cyclone structural scales using satellite data rely heavily on subjective experience. Objective estimation remains in its infancy, often using single-frame infrared satellite images without considering temporal evolution characteristics, leading to limited accuracy. This study proposed a deep learning model combining multi-task learning and physical auxiliary tasks. It extracted spatiotemporal features from sequential infrared satellite images (current and historical) and constrained the model training with physical factors closely related to tropical cyclone structural scales, achieving objective estimation. The model was validated in the Northwest Pacific basin, demonstrating superior performance compared to existing methods. This work not only provided a novel approach for objective estimation but also offered insights for typhoon disaster prevention and mitigation.

This research was supported by the National Natural Science Foundation of China (No. 42075140) and the Zhejiang Provincial Natural Science Foundation (No. TGG24D050001), both chaired by Professor Zhang Changjiang.

Link to the paper:

https://ieeexplore.ieee.org/document/10815990