Microplastics have become a major concern for scientists and policymakers around the world as more and more plastic waste is discharged into the oceans. The significant spatial heterogeneity of ocean microplastic abundance, however, makes its main drivers still unclear. We propose a deep learning model based on a few-shot strategy (CGMAT) to comprehensively explore the coupling mechanisms of multidimensional drivers under global warming and their impacts on microplastic distribution patterns in the coming decade. Socioeconomic data, meteorological and hydrological data of 23 municipal administrative units around the Taiwan Strait, as well as monthly mean temperature data from the BCC-CSM2-MR (China) model, are included in the SSP5-8.5 scenario of the sixth phase of the Coupled Model Intercomparison Project (CMIP6). The simulation results were consistent with the measured values (R² ≥ 0.94), and the results indicated that the synergistic coupling effect of global warming and anthropogenic factors led to a significant increase (50.334 ± 80.321 particles/m³) in microplastic abundance in tropical and subtropical coastal areas in the range of 0.1°-1°C global temperature increase. In view of the amplifying effect of global warming, more attention must be paid to the impact of anthropogenic emissions on the distribution of marine microplastics.
 

Microplastics,Gobal Warming,Human Activities,Coupling Mechanisms,deepl Learning model,Few-shot Learning