RiSIM: River surface image monitoring software for quantifying floating macroplastic transport.

Area of Science:

• Environmental Science
• Remote Sensing
• Artificial Intelligence

Background:

• Continuous plastic monitoring in rivers is crucial for understanding plastic pollution.
• Image-based monitoring with deep learning offers an effective strategy for observing floating plastic transport.

Purpose of the Study:

• To develop and validate River Surface Image Monitoring Software (RiSIM) for quantifying floating macroplastic transport.
• To assess the accuracy of RiSIM under various river conditions, including floods.

Main Methods:

• Utilized template matching for frame identification and deep learning models for plastic detection, classification, and tracking.
• Employed mark-release-recapture experiments and in-situ visual observations for validation.
• Analyzed plastic transport rates in terms of quantity and mass, correlating them with river discharge.

Main Results:

• RiSIM demonstrated high agreement with ground truth data for plastic transport rates (r = 0.91 for quantity, r = 0.80 for mass).
• The software accurately captured temporal variations in plastic transport, even during flood events (r = 0.87).
• Significant relationships were found between daily plastic transport rates and river discharge during floods (r² = 0.36 for quantity, r² = 0.27 for mass).

Conclusions:

• RiSIM is a powerful near-field remote sensing tool for quantifying riverine plastic transport.
• The software provides valuable data for managing mismanaged plastic waste in aquatic environments.
• RiSIM's ability to assess flood-induced plastic transport aids in targeted mitigation efforts.