Green tide cover area monitoring and prediction based on multi-source remote sensing fusion

Green tides are recurring ecological disasters in the Yellow Sea region. In this paper, a Fully Automated Green Tide Extraction Method (FAGTE) is proposed to extract Yellow Sea green tide data from multi-source satellite remote sensing (RS) images with resolutions of 16–250 m obtained during 2021–2024. The average accuracy of green tide extractions exceeds 91 %, and the extraction of small green tide patches from high-resolution satellite images was significantly superior to that from low-resolution images. Additionally, a novel method for fusing the extracted green tide cover areas from satellite images of varying resolutions is proposed. In 2023, the maximum post-fusion green tide cover area was 2262.12 km². Gompertz and Logistic growth curve models were used to monitor and predict regional green tide growth trends. The fitted growth curves exhibited $R^2$ values >96 %, while curves fused from multiple sources demonstrated $R^2$ values >99 %, indicating high accuracy. The predicted green tide start and end dates roughly matched the actual dates, with the highest accuracy in 2023 (relative errors: 1.63 % and 0.81 %, respectively). Both the growth curve fitting effect and the relative errors of predicted start and end dates were related to the green tide development mode. This study provides a scientific basis for monitoring and predicting green tides in the Yellow Sea.