Modeling spatiotemporal patterns of microplastic pollution in the lupit river using multilinear regression

Despite increasing awareness of microplastics as contaminants, their sources and abundance factors remain poorly understood. This study found widespread microplastic pollution in the Lupit River. Microplastic concentrations were determined from surface water samples once per season. Samples were analyzed across rural, residential, informal settlement, and commercial zones during both seasons. Microplastic concentration (particles/m²) was modeled using multiple linear regression with four predictors: population, seasonality, macroplastic frequency (items/m-hr), and volumetric flow rate (m²/s). To enhance model stability, multicollinear variables (velocity, width, and depth) were removed. Concentrations were lower in the wet season due to dilution and flushing and higher in the dry season likely from accumulation and weathering. The final model showed strong explanatory power (R² = 0.690, adjusted R² = 0.643), with population and seasonality as significant predictors (p < .001). Seasonality had a negative effect (β = - 621.90) due to dilution. Surprisingly, population correlated negatively (β = -0.0217), suggesting better waste infrastructure in denser areas. Macroplastic abundance and flow rate were not statistically significant due to slow weathering and local microplastic accumulation. The model's standard error (301.6 particles/m²) accounted for 16.9% of the mean concentration (1,780 particles/m²), signifying good prediction accuracy. The developed predictive model provides a low-cost tool for estimating microplastic levels in data-scarce areas which can assist agencies in tracking pollution trends over time and to assess the effectiveness of waste management policies. Future research should validate the model across different catchments and temporal scales to enhance its utility in long-term monitoring.