A Machine-Learning Model for Investigating Microplastics Source-Receptor Relationships in Aquatic Environments

Due to its durability, low manufacturing cost, and versatility, plastic is a highly desirable product, and its production has been increasing over time. As such, the presence of plastic pollution in our environment also continues to increase. Plastic pollutants are categorized by size, and fragments that are smaller than 5 mm in size are considered microplastics. Microplastics are known to negatively affect wildlife and their ecosystems, and as the presence of microplastics increases in our environment, so too does the risk of humans ingesting or inhaling these particles. Thus, studying microplastics and their environmental effects is essential to mitigate the harm they cause to organisms, their habitats, and to humans. Specifically, modeling the movement of microplastics in waterways is a significant area of study that allows for informed decision-making in terms of cleanup efforts and policymaking. To study where a microplastic particles might accumulate, or where they might come from, researchers typically develop probabilistic models using satellite observations, field observations, and numerical models. These approaches are limited, however, because the resulting models cannot predict source-receptor relationships for future climate scenarios or microplastic conditions that have not been simulated in the building of the model. Recent advancements in computational technology have promoted the integration of automation into new applications, such as machine learning. The current research presents a machine-learning technique that was applied to a real-world case study to investigate microplastic source-receptor relationships in the St. John River Estuary in New Brunswick, Canada. The model was able to estimate the likelihood of microplastic particles accumulating in seven regions of the river in a timely manner was predicted, for microplastic particles that remained in the river domain, thereby supporting targeted cleanup operations and informed decision-making to manage microplastic exposure risk in the Saint John River Estuary.