Coastal ecosystems are declining due to natural and human influences. This study analyzes the temporal patterns of seagrass meadows around Pramuka Island, Jakarta, using Sentinel-2 imagery from 2016 to 2023. The data were processed with the dark spectrum fitting (DSF) algorithm for atmospheric correction, followed by sun glint correction. The Depth-Invariant Index (DII) and various spectral indices were used as inputs for classification with a random forest model. The annual variation in detected dense seagrass meadows showed a decreasing trend at a rate of 1.64 ha per year, while sparse and dense seagrasses together accounted for approximately 20% of the shallow water areas. Seagrass, particularly denser types, could be distinguished by its reflectance in the red edge (704 nm) band. Despite fluctuations of both gains and losses, the net result over the observed period was a total loss of 52.9 ha. The relationship between sea surface temperature (SST) and seagrass extent had a correlation of − 0.43, generally showing an inverse pattern. However, this relationship varied over time, with SST and seagrass extent occasionally fluctuating together in certain years.