Abstract: Reed wetlands are crucial carbon sink ecosystems.Accurately assessing their carbon sequestration capacity is vital for understanding the carbon cycle and ensuring ecological security.This study,taking the Dongting Lake reed wetland as an example,proposes a net primary productivity(NPP) estimation method integrating multi-source remote sensing,and explores the driving mechanism of phenological changes on carbon sink capacity.This study employed piecewise function regression.It synergistically utilized Sentinel-1 SAR and DEM data to invert canopy height.Landsat 9 OLI and Sentinel-2 optical data were fused to construct separate NPP prediction models for the initial growth stage (April) and the senescence period (November), based on vegetation indices and canopy height.The research revealed a significant positive correlation between canopy height and NPP in April.In contrast,during the November senescence period,although height was maintained,NPP decreased significantly and showed a negative correlation.The piecewise model effectively captured carbon sink differences across distinct phenological stages.Validation with field measurements from 2023 demonstrated that the method significantly improved estimation accuracy and reliability.This study validates that integrating multi-source remote sensing with piecewise modeling enhances the precision and timeliness of dynamic carbon sink monitoring in wetlands.It provides a scientific basis for precise wetland management.

Key words: phragmites wetlands, carbon sink accounting model, net primary productivity, SAR-DEM synergistic inversion, vegetation indices, segmented function regression model