Geographia Technica, Vol 21(2), Special Issue: Artificial Intelligence Applications in Geography, 2026, pp. 75-96
RANDOM FOREST–BASED MAPPING OF RIVERINE PLASTIC POLLUTION FROM SENTINEL-2 IN GOOGLE EARTH ENGINE: LULC AND RAINFALL CONTROLS IN KENDAL REGENCY, INDONESIA
Ananto A. AJI 
, Syaiful Muflichin PURNAMA , Vina Nurul HUSNA
ABSTRACT: Plastic pollution is a major environmental issue, especially in riverine and urban systems. Understanding its spatial distribution relative to land use/land cover (LULC) and precipitation is crucial. This study examined plastic waste distribution in Kendal Regency using a Plastic Index (PI) derived from remote sensing, Sentinel-2-based LULC classification, and precipitation data. Statistical analyses included boxplots, swarmplots, and correlation tests. PI values differed across LULC types, with higher values in settlements and industrial areas, and lower or negative values in forests and plantations. Irrigated paddies and water bodies showed high variability. In contrast, precipitation showed weak, inconsistent, and non-significant correlations with PI. Plastic accumulation is strongly linked to anthropogenic land cover rather than rainfall. The results highlight urbanization as a key driver of plastic pollution and provide insights for sustainable waste management strategies.
Keywords: Plastic Index (PI); Land Use/Land Cover (LULC); Precipitation; Urbanization; Remote Sensing.
Full article here