npj Climate and Atmospheric Science (Jan 2024)

Historical footprints and future projections of global dust burden from bias-corrected CMIP6 models

  • Jun Liu,
  • Xiaofan Wang,
  • Dongyou Wu,
  • Hailun Wei,
  • Yu Li,
  • Mingxia Ji

DOI
https://doi.org/10.1038/s41612-023-00550-9
Journal volume & issue
Vol. 7, no. 1
pp. 1 – 12

Abstract

Read online

Abstract Dust aerosols significantly affect the Earth’s climate, not only as a source of radiation, but also as ice nuclei, cloud condensation nuclei and thus affect CO2 exchange between the atmosphere and the ocean. However, there are large deviations in dust model simulations due to limited observations on a global scale. Based on ten initial Climate Models Intercomparison Project Phase Six (CMIP6) models, the multi-model ensemble (MME) approximately underestimates future changes in global dust mass loading (DML) by 7–21%, under four scenarios of shared socioeconomic pathways (SSPs). Therefore, this study primarily constrains the CMIP6 simulations under various emission scenarios by applying an equidistant cumulative distribution function (EDCDF) method combined with the Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA2) datasets based on observation assimilation. We find that the results (19.0–26.1 Tg) for 2000–2014 are closer to MERRA2 (20.0–24.8 Tg) than the initial simulations (4.4–37.5 Tg), with model deviation reduced by up to 75.6%. We emphasize that the DML during 2081–2100 is expected to increase significantly by 0.023 g m–2 in North Africa and the Atlantic region, while decreasing by 0.006 g m–2 in the Middle East and East Asia. In comparison with internal variability and scenario uncertainty, model uncertainty accounts for more than 70% of total uncertainty. When bias correction is applied, model uncertainty significantly decreases by 65% to 90%, resulting in a similar variance contribution to internal variability.