Prediction of Land Use and Land Cover (LULC) Changes for 2030 Using Time Series Data and Random Forest Algorithm

This study predicts land use and land cover (LULC) changes in Karo District in 2030 using Sentinel-2 data and the Random Forest algorithm. LULC data from 2017 to 2023 was used to train the model. Results show a significant decrease in forest area from 127,903 ha in 2017 to 85,291 ha in 2030 and a sharp increase in urban area from 2,287 ha to 51,158 ha. The water area is predicted to increase from 3,714 ha in 2017 to 29,479 ha in 2030. Bareland is expected to decrease from 8,784 ha in 2017 to 2,961 ha in 2030, while cropland will decrease from 74,369 ha in 2017 to 48,169 ha in 2030. Although the model showed high accuracy, manual observations revealed errors, such as misclassifying blue-roofed residential areas as water and predicting water areas in cloudcovered forests. This research highlights the importance of remote sensing and machine learning in monitoring and predicting LULC changes, providing a solid basis for improved land management and regional planning.