Spartina alterniflora, a typical invasive species in China's coastal wetlands, has become a focal point of research due to its rapid expansion mechanisms and ecological impacts on tidal flat ecosystem security. This study utilized high-resolution UAV remote sensing to construct a multi-temporal dataset from 2013 to 2022 and developed a U-Net-ResNet34 deep learning model specifically designed for complex intertidal environments. By optimizing feature extraction through residual connections, the model significantly enhanced the classification accuracy of easily confused land cover types, such as mangroves and Spartina alterniflora. Building upon existing landscape pattern indices, strongly correlated indices were grouped through correlation analysis, followed by principal component analysis to construct two synthesized indicators: the Spatial expansion metric (SEM) and the Aggregation-edge trade-off index (AETI). Combined with landscape expansion modes and standard deviation ellipse models, the spatial diffusion characteristics were quantitatively revealed. The results indicated that: (1) The average overall classification accuracy of the improved U-Net-ResNet34 model is 96.73%, and the Kappa coefficient was 0.97, which effectively solved the spectral confusion problem of mangrove-Spartina alterniflora; (2) The landscape dynamics exhibited a three-phase characteristic of "diffusion-colonization-competition". During the diffusion phase, the patch density increased at an annual rate of 42.3%; during the colonization phase, the Spatial expansion metric (SEM) grew at an average annual rate of 18.7%; and during the competition phase, the Aggregation-edge trade-off index (AETI) decreased by 64.2%; (3) Expansion modes shifted from early-stage outlying (62.5%) to mature-phase infilling-dominated (95.03%); (4) The center of mass in space shifted southward by 1.01 km, and the azimuth of the main axis was deflected by 30.43°, and the directionality was significantly enhanced. The study elucidated the ecological mechanisms by which Spartina alterniflora achieved rapid invasion through clonal propagation and niche adaptation, providing a scientific basis for controlling biological invasions in tidal flats.