Abstract: Spartina alterniflora is the most harmful invasive plant in coastal wetlands in China. It is of great significance to use remote sensing technology to clarify its spatial distribution and monitor its clearance dynamics for coastal Spartina alterniflora control and wetland resource protection. This study proposed a dynamic monitoring framework for Spartina alterniflora that integrated multi-source time-series remote sensing data to address the issues of severe cloud interference, vegetation spectral confusion, and feature redundancy in coastal wetland vegetation classification. Based on Sentinel-1, Sentinel-2, and Landsat8 data from Yancheng,Jiangsu Province,covering the period from 2022-2023, the optimal time window for extracting Spartina alterniflora was determined through NDVI temporal phenology analysis. Combined with Random Forest, a highly discriminative feature set was selected from multiple sources of features (spectrum, texture, radar, terrain). Finally, a U-Net deep learning model was used to achieve high-precision extraction and dynamic monitoring of Spartina alterniflora. The results showed that the mid growth season (June to August) was the optimal time window for extracting Spartina alterniflora. Using images from this period to extract the wetland in Yancheng in 2022, the overall classification accuracy was 94.34%, with a Kappa coefficient of 0.873 4. The total invasive area of Spartina alterniflora was 119.08 km2, mainly distributed in Dafeng, Dongtai, Sheyang. During the process of Spartina alterniflora management in 2023, a total area of 80.49 km2 was cleared. Among them, Dafeng had the widest cleared area, with a total of 46.46 km2. This research provided a feasible plan for long-term and large-scale monitoring of the spread of Spartina alterniflora in coastal wetlands, and provided important reference for large-scale invasive plant control and wetland protection.