Analisis Perubahan Tutupan Lahan Permukiman Berbasis Sentinel-2 dan SVM di Kecamatan Tapin Utara

Rojaa Firoous Suri
Nasruddin, Nasruddin
Nurul Astuti

Jurnal Sains dan Teknologi
Vol. 5 No. 1 (2026): Februari 2026

Residential land-cover change is an important indicator of spatial transformation and regional development dynamics. This study analyzes changes in residential land cover in North Tapin District, Tapin Regency, during 2016–2024 using Sentinel-2 Level-2A imagery integrated with Geographic Information Systems (GIS). A supervised classification approach based on the Support Vector Machine (SVM) algorithm with a Radial Basis Function kernel was applied to distinguish residential and non-residential areas. Several spectral indices were incorporated to improve class separability prior to classification. Accuracy assessment using a confusion matrix with 50 validation points indicates high reliability, with Overall Accuracy values of 98% (κ=0.91) in 2016 and 96% (κ=0.83) in 2024. Post-classification comparison reveals an increase of 181.85 ha in residential land over eight years, mainly resulting from the conversion of non-residential areas. Spatially, expansion patterns tend to follow major road corridors and highly accessible zones. These findings demonstrate that the integration of Sentinel-2 imagery and SVM provides an effective framework for multitemporal residential land-cover monitoring and supports evidence-based regional planning.