Application of Remote Sensing and GIS in Monitoring Forest Cover Changes in Vietnam Based on Natural Zoning

Abstract

Forest cover changes monitoring in Vietnam has been conducted using remote sensing (RS) and geographic information systems (GIS). Given Vietnam’s diverse climate, this study focused on the Thanh Hoa, Kon Tum, and Dong Nai provinces due to their distinct natural conditions and forest structures. Land cover was classified into five categories: broadleaf forests, mixed forests, shrubland/grassland/agricultural land, non-forested areas, and water bodies. RS data processing was performed using Google Earth Engine (GEE), with land cover classification conducted via the Random Forest algorithm. The findings revealed significant changes in forest cover between 2010 and 2020. In Thanh Hoa, broadleaf forests expanded by 51.15% (91,159 ha), while mixed forests declined by 19.68% (105,445 ha). Kon Tum experienced reductions in both broadleaf forests (20.05%, 26,685 ha) and mixed forests (4.06%, 20,501 ha). Meanwhile, Dong Nai recorded increases in broadleaf forests (29.15%, 23,263 ha) and mixed forests (12.17%, 20,632 ha). The study’s reliability was confirmed by a Kappa coefficient of 0.81–0.89. To predict forest cover changes, two methods—the CA-Markov model and the MOLUSCE module—were compared. Results demonstrated that the MOLUSCE module achieved higher accuracy, with deviations from actual data of 1.61, 1.14, and 1.80 for Thanh Hoa, Kon Tum, and Dong Nai, respectively, whereas the CA-Markov model yielded larger deviations (8.79, 6.29, and 5.03). Future forest cover projections for 2030, generated using MOLUSCE, suggest significant impacts from agricultural expansion, deforestation, and restoration efforts on forest area. This study highlights the advantages of remote sensing and GIS for complex forest monitoring and sustainable management in Vietnam.