Correction to: Nature Communications, published online 13 March 2023 The original version of this Article contained an error in Fig. 2, in which c was incorrectly described as “middle scenario” where it should have been described as “low scenario”. The correct version of Fig. 2 is: (Figure presented.) which replaces the previous incorrect version: (Figure presented.) This has been corrected in both the PDF and HTML versions of the Article.

Marine plastic pollution poses a potential threat to the ecosystem, but the sources and their magnitudes remain largely unclear. Existing bottom-up emission inventories vary among studies for two to three orders of magnitudes (OMs). Here, we adopt a top-down approach that uses observed dataset of sea surface plastic concentrations and an ensemble of ocean transport models to reduce the uncertainty of global plastic discharge. The optimal estimation of plastic emissions in this study varies about 1.5 OMs: 0.70 (0.13–3.8 as a 95% confidence interval) million metric tons yr⁻¹ at the present day. We find that the variability of surface plastic abundance caused by different emission inventories is higher than that caused by model parameters. We suggest that more accurate emission inventories, more data for the abundance in the seawater and other compartments, and more accurate model parameters are required to further reduce the uncertainty of our estimate.