Machine learning-based prediction for settling velocity of microplastics with various shapes

Area of Science:

• Environmental Science
• Fluid Dynamics
• Computational Science

Background:

• Microplastics contaminate aquatic ecosystems, impacting water body transport and distribution.
• Accurate prediction of microplastic terminal settling velocity is crucial for understanding their environmental fate.
• Existing models struggle to predict settling velocity for microplastics with diverse shapes.

Purpose of the Study:

• To develop high-performance machine learning models for predicting microplastic terminal settling velocity.
• To identify the key feature parameters influencing microplastic settling velocity.
• To analyze the importance and effect of each parameter on prediction accuracy.

Main Methods:

• Classified microplastic shapes (fiber, film, fragment) based on principal dimensions.
• Identified optimal shape parameters (Corey shape factor, flatness, elongation, sphericity) for each category.
• Utilized machine learning models incorporating dimensionless diameter, relative density, and optimal shape parameters.

Main Results:

• Machine learning models achieved R² > 0.867 in predicting terminal settling velocity for various microplastic shapes.
• Models significantly outperformed existing theoretical and regression models.
• Microplastic size was identified as the most important factor, with shape effects becoming significant for larger microplastics (D* > 65).

Conclusions:

• Machine learning provides a robust framework for predicting microplastic terminal settling velocity.
• Accurate prediction is essential for managing microplastic pollution in aquatic environments.
• Understanding the influence of size and shape parameters enhances environmental transport models.