Machine learning-supported quantification and characterization of sediment plastic debris in an Anthropized Mfoundi River in Cameroon: Implications for the incidence of flood events

The emergence of plastic pollution as a global threat to both terrestrial and aquatic organisms has seen efforts geared toward minimizing its production and monitoring widespread distribution within the ecosystem. Though with distinct characteristics and sources, plastic debris and sediments often interact in the natural environment through a complex relationship, forming "sediment plastic debris" (SPD) which leaks into the riverine system through human actions or elements of nature. Quantifying SPD in riverine environments is therefore essential for understanding the extent of pollution and its ecological impacts. This study presents a practical approach to analyzing SPD production, transportation, and accumulation in developing countries' terrestrial and aquatic environments where plastic waste management strategies and infrastructure remain insufficient. Surface and riverine SPD hotspots were mapped and quantified and intercepting grids (13 mm grid size) were installed in five locations within the channel to capture leaked SPDs. The Random Forest and XGBoost were also employed to develop SPD susceptibility maps for the subbasin. Our findings revealed a remarkable proliferation of SPD hotspots with considerable risk of leaking into the riverine system. A significant >95 % of hotspots were located between 0 and 300 m from the road network while 50 % were located at distances <300 m from the river network or entirely on the flood plain and river channel. Our findings indicate that approximately 100 tons of plastics and 700 tons of putrescible debris are transported annually in low-flow conditions. These amounts are expected to increase due to contributions from surface runoff and remobilization from upstream. The two models used for SPD susceptibility showed results with remarkable AUC of 93 % and 92 %, predicting significant 30 % and 36 %, respectively, of the basin's area to be highly susceptible to SPD accumulation. These results underscore the urgent need for improved plastic waste management.