[U-04] Geospatial Applications for Societal Benefits

[U04-04] Detection of distribution of debris in the coastal environment

*Ye Min Htay¹, Ahmad Shaqeer Mohamed Thaheer¹, San Lin Phyo¹, Yukihiro Takahashi¹ (1.Hokkaido University)

Keywords:plastics, coastal, hyperspectral, classification

The pollution of plastics has emerged as a significant stressor on coastal environments, with plastic debris accumulating on beaches and degrading over time, leading to the formation of microplastics. This study focuses on the classification of both natural and man-made debris, including vegetation, rocks, woods, nylon, PET bottles, and plastic bags, to better understand the distribution of debris along coastal areas. Spectral characterization of each material is conducted using an InGaAs hyperspectral sensor, covering a range from 400nm to 1600nm. Laboratory measurements establish the unique spectral signatures of different debris types, facilitating their differentiation in remote sensing data. To streamline the classification process and reduce computational costs, a band selection method is applied to identify the most informative bands from the hyperspectral data. By selecting the best four bands, a simpler model is constructed, offering a more cost-effective alternative to full hyperspectral analysis while still maintaining classification accuracy. It offers an understanding of the distribution of plastics in coastal environments. Additionally, the methodology presented here demonstrates a practical approach for leveraging hyperspectral remote sensing data in environmental monitoring and management.