Applications of remote sensing and GIS techniques for identifying of the plastic waste from space: Evidence from Khulna city corporation in Bangladesh

• Leverages Sentinel-2 imagery and supervised classification algorithms to identify plastic waste on land. • Support vector classification achieving the highest accuracy. • Bands 6, 7, and 8 of Sentinel-2 imagery show higher spectral reflectance for plastic. • Plastic waste identification provides valuable insights in managing plastic waste sustainably. Plastic waste poses a significant threat to the environment, public health, and aquatic life. Several methods are now under development in many studies to monitor plastic waste through earth observation satellites. These methods were successfully applied to monitor plastic litter and debris. Due to the special optical signature of plastic, it is easy to identify it in aquatic environments. But in the case of identifying plastic waste on land or in a terrestrial environment, it is very difficult as different types of land cover have their own special optical signature. Conducting field surveys could be a possible solution for monitoring plastic waste on land, but it’s costly and time-consuming. To tackle this problem, remote sensing-based observation can make a sustainable contribution. This study aims to identify plastic waste on land by the combination of Sentinel-2 imagery and two supervised classification algorithms: (1) maximum likelihood and (2) support vector classification. Two locations where plastic waste was recycled were considered for conducting this study by field observations. A total of 60 samples have been taken in this study, out of which 80% (48) have been taken as training samples and the remaining 20% (12) have been taken as testing samples, and the entire process was done using ArcGIS 10.8. This analysis revealed that algorithms used in this study successfully identify plastic waste on land, and between two algorithms, support vector classification achieves the highest accuracy (93%). Bands 6, 7, and 8 show higher spectral reflectance for plastic. The finding suggests that supervised algorithms can be used to identify plastic waste on land. Other algorithms, high-resolution satellite imagery, and a larger dataset are necessary to identify smaller plastic waste on land. This study will help policymakers and decision-makers at national and local levels to identify and management of plastic waste in a sustainable way.