<![CDATA[This study investigated land use and land cover (LULC) changes in Obi Subdistrict, Indonesia, from 2010 to 2015, driven by the expansion of the nickel mining industry. Using Landsat 7 and Landsat 8 imagery, Random Forest classification and change detection were conducted to evaluate annual LULC dynamics. Preprocessing included cloud masking and the calculation of NDVI, NDBI, and NDWI to enhance class separability. Four land cover classes were defined: dense vegetation, sparse vegetation, bare soil, and urban areas. The results showed a significant increase in urban/built-up area from 2,052 ha (2010) to 4,843 ha (2015), alongside a decrease in sparse vegetation from 92,770 ha to 84,848 ha. Dense vegetation increased to 10,236 ha in 2015, suggesting potential regrowth. Chord diagrams and pixel-based change maps reveal that transitions from sparse vegetation to urban and dense vegetation dominate the landscape change. Accuracy assessment indicates classification reliability improved from Landsat 7 to Landsat 8, with dense vegetation F1-score increasing from 0.21 to 0.81. This study demonstrated the utility of spectral indices and machine learning in early-stage LULC detection. It recommends future improvements using object-based classification, ground-truth validation, and deep learning for more robust environmental monitoring in resource-rich areas. This study contributes an early-stage LULC assessment framework for mining zones in Indonesia, which can inform future land governance and remote sensing policy applications.]]>
Copyrights © 2025
