Monitoring thirty years of LULC change (1990 - 2021) from Landsat images of JATABEK, Indonesia

Abstract

Monitoring land use/land cover (LULC) change is urgently needed in a sustainable environmental management process. Many management aspects require accurate information about past, present, and future projections to develop sustainable management strategies. Jakarta and its surrounding cities (Tangerang and Bekasi) called JATABEK is one area frequently flooded. This research aimed to derive the LULC map and its change by integrating remote sensing and Geographic Information Systems. Multi-year Landsat images conducted mapping using machine learning and Random Forest (RF) classifier processed through Google Earth Engine (GEE). The results showed that the two main classes that experienced drastic changes in JATABEK were built-up land and paddy fields which increased and decreased by 244% and 69%, respectively. In addition, RF and GEE are very robust in complex areas such as JATABEK. This is proved by the overall accuracy and Kappa coefficient above 80%. Planners and managers used a holistic picture of LULC in JATABEK to establish flood management strategies.