Forecasting Coastal Morphodynamics and Urban Expansion at Sungai Karang, Malaysia: A Machine Learning Approach within Google Earth Engine toward 2030

ABSTRACT This study examines the predictive modeling of beach morphology and land use changes at Sungai Karang, Kuantan, Malaysia, utilizing machine learning techniques with a Random Forest model developed within the Google Earth Engine (GEE) framework. The model, trained on historical land use and morphological data from 2008 and 2013, predicts significant land use modifications by 2030. The model achieved an accuracy of 98.71% and a Kappa coefficient of 0.9850, indicating strong agreement between predicted and actual classifications. Key findings reveal an exponential increase in urbanization from 141.49 ha in 2008 to 8421.23 ha by 2030, signifying rapid urban growth. At the same time, natural ecosystems, including forests and marshlands, face substantial decline, with forests nearly vanishing by 2030 (from 895.35 to 1.88 ha). Mangrove forests, which fluctuate in earlier periods, are projected to decrease to 2055.98 ha by 2030. Other notable changes encompass reductions in water bodies and recreational areas, while infrastructure such as roads and railways expands, reflecting broader urban development. The model also predicts a significant increase in quarrying activities, indicating higher risks of land degradation. These results highlight the urgent need for sustainable land use planning, especially in protecting vulnerable ecosystems and mitigating the effects of urbanization and industrialization. The study provides valuable insights for policymakers, stressing the importance of integrating climate resilience and environmental conservation into development strategies. This predictive approach offers an essential tool for informed decision‐making, promoting more balanced and sustainable regional growth.