Application of Google Earth Engine to detect land - cover changes in Ben Luc commune, Tay Ninh province, period 2020 - 2025

Abstract

This study applies the Google Earth Engine (GEE) platform combined with Sentinel-2 imagery to derive land-cover maps and land-cover change maps for Ben Luc commune, Tay Ninh province, in the period 2020-2025. Two dry-season composites for 2020 and 2025 were generated from Sentinel-2 Level-2A data, pre-processed on GEE by cloud masking, mosaicking and clipping to the administrative boundary. The images were then classified using the Random Forest (RF) algorithm into four major land-cover classes: bare land, built-up land, water bodies and vegetation. Training and validation samples were derived from high-resolution imagery and field verification. Classification accuracy was assessed using confusion matrices, overall accuracy and the Kappa coefficient. The results show an overall accuracy of 91.05% with a Kappa coefficient of 0.88 for 2020, and 87.58% with a Kappa coefficient of 0.83 for 2025, indicating a high reliability of the classification model. In 2020, bare land and vegetation together accounted for nearly 70% of the commune’s area, while built-up land and water covered smaller proportions. In 2025, bare land decreased by about 657.77 ha, whereas vegetation increased by around 402.50 ha, built-up land expanded by over 124.33 ha and water bodies increased by roughly 130.95 ha. These changes reflect a reduction in unused land, enhanced green coverage and the expansion of built-up space and surface water systems. Based on the two land-cover maps, a land-cover change map for 2020-2025 was generated using overlay analysis in ArcGIS. The resulting layers were stored in a PostgreSQL/PostGIS database and published via GeoServer to build a thematic WebGIS application. The WebGIS enables users to query change types, areas and locations directly on an interactive online map. The findings demonstrate the effectiveness of GEE and Sentinel-2 data for local-scale land cover monitoring, as well as the potential of integrating WebGIS to support land resource management and spatial planning.