Abrupt intensification and spatiotemporal dynamics of large-scale harmful algal blooms in China’s marginal seas mapped using multi-sensor satellite imagery and deep learning
This research aims to monitor and analyze the dynamics of harmful algal blooms (HABs) in China's marginal seas using advanced satellite imagery and deep learning techniques.
Used a Res-UNet deep learning model to analyze 74,882 satellite images from MODIS, VIIRS, and OLCI.
Integrated multi-sensor data to achieve improved monitoring of harmful algal blooms across multiple coastal regions.
Validated the model against Chinese marine disaster bulletins and high-resolution imagery, achieving an average F1-score of 0.85.
HAB intensity increased by over 51% in 2023-2024 compared to the 2018-2022 average, especially in the Bohai and South China Sea.
Bimodal peaks in HAB occurrences were observed in April and August in the Yellow Sea.
Seasonal shifts in HABs were noted, moving northward from the South China Sea in spring to the Bohai and East China Seas in summer.