Advanced buoy phenological extraction methods for highly dynamic Zhejiang coastal waters

Accurate phenological information on phytoplankton is crucial for understanding energy flow, carbon cycling, and the growth and reproductive rhythms of key species within marine ecosystems. However, in dynamic and variable coastal waters, traditional methods struggle to detect critical phenological phases and trends precisely due to the high temporal and spatial variability. Furthermore, the relative contribution of interacting environmental factors to phenological shifts in coastal waters remains insufficiently understood. This study develops an integrated phenology extraction framework based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), designed to handle nonlinear and non-stationary phytoplankton time series in highly dynamic coastal environments. Using in-situ buoy data, we analyzed phytoplankton phenology characteristics and driving factors in the coastal waters of Zhejiang, China, from 2016 to 2020. The results indicate a significant advancement in bloom onset and peak timing across most areas, with temperature and turbidity identified as the factors most closely associated with these phenological shifts. While empirical rules are still applied for phenological definition, the CEEMDAN-based framework significantly reduces the reliance on fixed global concentration thresholds by providing a denoised, adaptive baseline. This research is the first to apply the EMD (Empirical Mode Decomposition) series of decomposition methods to phenological extraction, demonstrating its practicality in handling nonlinear, non-stationary chlorophyll-a time series from buoy data in highly dynamic, eutrophic coastal waters. By adaptively separating high-frequency disturbances, it reveals temperature- and turbidity-driven shifts in bloom timing, offering insights into phenological responses to environmental stressors like eutrophication and pollution, and informing strategies for pollution mitigation and coastal ecosystem management.