[M-AG32] Satellite Land Physical Processes Monitoring at Medium/High/Very High Resolution

[MAG32-P03] Regional-Waste Dumping Susceptibility Analysis

*Ya-Chen Yang¹、Chi-Farn Chen¹、Yi-Jeng Lin¹ (1.Center for Space and Remote Sensing Research, National Central University, Taiwan)

キーワード：Waste dumping, Land monitoring, Machine learning

In recent years, incidents of illegal waste dumping have become increasingly frequent in Taiwan, leading to environmental pollution and posing significant threats to public safety and agricultural production. The statistical data from the government reveals a consistent annual rise in the number of cases violating the Waste Disposal Act. In 2023, the number has increased by 1.5 times compared to the year 2017. To understand the association of the location with the waste dumping, this study attempts to develop a waste dumping susceptibility prediction model. The model uses the random forest algorithm of machine learning to relate the locations of the waste dumping to the environmental factors. The data of waste dumping are collected from 2016 to 2023, with the first seven years dedicated to model development and the final year allocated for model validation. The environmental factors include 12 GIS layers: demographic data, land use maps, designated soil and water conservation maps, and distribution map of enterprises and organizations regulated by the Ministry of Environment. The test sites focus on Pingtung County and Taoyuan City in Taiwan. The analysis results indicate model accuracies of 94% and 97% for Pingtung County and Taoyuan City, respectively. According to the model calculations, three probability maps can be generated for low susceptibility, moderate susceptibility, and high susceptibility areas. In Pingtung County, high susceptibility areas account for 6.4% of the entire county. Analysis of environmental layer characteristics indicates that areas near vacant land, reservoirs, and estuaries are more susceptible to becoming high susceptibility areas for illegal waste dumping in Pingtung County. Similarly, in Taoyuan City, high susceptibility areas constitute 0.8% of the entire city. The analysis of environmental layer characteristics suggests that locations removed from landslides and near vacant land are more prone to becoming high susceptibility areas for illegal waste dumping in Taoyuan City. Maps highlighting the high-risk areas for waste dumping have been created for further application. In conclusion, this study can serve as a reference for volunteers to enhance monitoring and for local
governments aiming to develop more effective land management policies.