Prediction in Terrestrial and Aquatic Ecosystems Using Sentinel-2, Landsat 8, and MODIS Data: An Integrated Approach through Google Earth Engine

In this study, we analyzed drought and vegetation dynamics in Taiwan by combining satellite observations from Sentinel-2, Landsat 8, and MODIS. The focus was on both terrestrial and aquatic ecosystems, as well as agricultural areas, for the period 2017 to 2023. Using the Google Earth Engine (GEE), we calculated key indices including NDVI, VCI, NDWI, and derived vegetation condition metrics to assess drought severity. VCI was computed monthly based on historical NDVI values and categorized into five drought severity classes to capture relative vegetation stress. Each satellite contributed uniquely to the analysis: Sentinel-2 provided high-resolution data for fine-scale monitoring, Landsat 8 offered continuity over the medium term, and MODIS supplied long-term and frequent observations, allowing both local and regional perspectives on vegetation and drought patterns. Limited ground-based measurements confirmed that satellite-derived results were consistent with precipitation records from CHIRPS, supporting the reliability of the data. The results were presented through maps, time series plots, and dynamic graphs, highlighting both seasonal fluctuations and interannual variability in drought and vegetation conditions. By combining spatially and temporally diverse datasets, the better coverage and continuity of monitoring enhanced the capability to accurately monitor ecosystem health and vegetation dynamics. In general, this method provides useful information on the benefits of integrating multi-source satellite observations for sustainable ecosystem management. Our research offers scientific support for early warning system creation and pertinent ecosystem protection and resource management by way of a thorough understanding of both the drought dynamics/synchrony and vegetative responses.