Abstract: Spartina alterniflora, as a major invasive vegetation in coastal wetlands, has a serious impact on the ecosystem structure and function of coastal wetland. Therefore, timely and accurate monitoring of the growth status of S. alterniflora can provide a basis for the restoration of coastal wetland ecosystems. Based on Sentinel-2 multispectral satellite images, this study extracted three types of characteristic variables: original bands, vegetation index, and biophysical parameters, and constructed an optimal biomass estimation model for S. alterniflora by using multiple stepwise regression methods, and then performed spatial mapping of biomass. The results showed that among the constructed models, the estimation accuracy of the stepwise regression model based on combining three types of characteristic variables had the highest estimation accuracy, with a determination coefficient of 0.879, a root mean square error of 255.5 g/m² and a mean relative error of 10.63%. The spatial distribution of aboveground biomass on S. alterniflora was characterized by low biomass near land and high biomass near the ocean side. In summary, it is feasible and effective to use Sentinel-2 data to invert the aboveground biomass on the S. alterniflora with high accuracy.