Abstract: Affected by reclamation and development, biological invasion, and periodic inundation, remote sensing-based vegetation extraction in the coastal zone of Jiangsu Province faces challenges such as inaccurate spatial delineation and unclear spatiotemporal evolution characteristics. Using the Google Earth Engine (GEE) cloud platform and long-term Landsat time series, this study develops a remote-sensing framework for vegetation extraction and spatiotemporal evolution analysis, and elucidates the four-decade dynamics of vegetation cover in the Jiangsu coastal zone. The results show that: (1) by explicitly accounting for tidal inundation, land-sea interactions, and seasonal variability, the proposed model markedly improves the accuracy of vegetation extraction and the robustness of change monitoring in highly dynamic coastal environments; (2) from 1985 to 2024, the total vegetation-covered area in the Jiangsu coastal zone exhibited a slight overall contraction, decreasing from 90,678.40 km² to 88,599.71 km², whereas vegetation density increased, with the mean NDVI rising from 0.64 to 0.69; (3) NDVI transition analysis indicates that the largest positive shift occurred from the relatively high vegetation cover class to the high vegetation cover class, totaling 29,769.76 km², indicating an optimized evolution toward aggregation at higher vegetation levels. These findings provide technical support for coastal ecological monitoring, vegetation restoration assessment, and regional ecological security management, and offer scientific evidence for promoting ecological restoration and achieving carbon peaking and carbon neutrality targets.

Key words: vegetation cover, NDVI, spatiotemporal pattern, coastal zone