Combining numerical modeling and machine learning for microplastic monitoring

Millions of tons of plastic waste enter the environment every year, wreaking havoc on the natural world as the material breaks down to form smaller particles called microplastics that end up in the air, land, and sea. The global COVID-19 pandemic has not helped the situation considering it increased demand for masks, gloves, antigen test kits, and other medical supplies that consist of plastic components. Clearly, efficient microplastic monitoring is crucial for tracking particle movement in the environment. Knowledge of particle dynamics could help to manage or even mitigate the effects of microplastic pollution.

Phan and Luscombe outline the features of available tools including numerical modeling and machine learning and discuss how they could help uncover improved methods to examine the physics of microplastic transport.

“While numerical models, which take into account physical parameters such as size, shape, density, and identity of microplastics, can help us understand various dynamics around the movement of microplastics, there currently are not a lot of large datasets needed to verify predictions,” said author Samantha Phan. “Therefore, the development of effective new monitoring techniques is imperative.”

Citing the use of machine learning and computer vision in applications such as autonomous vehicles and facial recognition, Phan and Luscombe suggest recent studies point to the possibility of an interdisciplinary field of applied physics in which image-based machine learning can help reveal the physics around microplastic transport. Machine learning can collect the physical parameters of microplastics for use in numerical models to increase the accuracy of predictions regarding microplastic transport.

“A better understanding of microplastic transport ultimately aids in microplastic monitoring efforts and will be necessary for better managing the progression of microplastic pollution,” said Phan.

Source: “Recent trends in marine microplastic modeling and machine learning tools: Potential for long-term microplastic monitoring,” by Samantha Phan and Christine K. Luscombe, Journal of Applied Physics (2023). The article can be accessed at http://doi.org/10.1063/5.0126358 .

This paper is part of the Special Collection Recognizing Women in Applied Physics; learn more here .