Aerosol direct radiative forcing spread in CMIP6 models

Abstract:

The radiative forcing from aerosols remains one of the most uncertain forcing on climate. We calculate aerosol forcing of 10 climate models that are participating in CMIP6 by using rapid radiative transfer model (RRTM). The global mean aerosol direct radiative forcing (DRF) is estimated as -1.138 (externally mixed state) and -0.872 (internally mixed state) from 1850 to 2014. The forcing spread between models is noteworthy and the estimations of aerosol loading and vertical distribution by models also differ significantly. Through analyzing the impact of aerosol parameters on DRF, we find the spread of aerosol optical depth (AOD) and single scattering albedo (SSA) causes larger aerosol direct forcing uncertainty. The increase of AOD strengthens the DRF and the decrease of AOD weakens DRF. Negative DRF is enhanced by increasing SSA, and the positive DRF is enhanced by decreasing SSA. Asymmetric factors have the opposite effect on DRF than SSA. In a word, the simulation error of aerosol loading is still the largest source of DRF spread. Besides, the fraction of absorbing and scattering aerosols also has significant effect on forcing estimation.

The aerosol height also plays an important role in its direct radiative forcing. We examine the influence of vertical distribution on DRF estimation mainly by standardize the aerosol mass for 10 models to achieve the same total aerosol mass. It indicates the DRF spread between models related to vertical distribution accounts for about 10% of the total spread. As for each aerosol type, there are larger forcing errors of black carbon, organic aerosol and sulfate aerosol, while DRF spread is most sensitive to black carbon. The largest DRF and its spread is in visible and near infrared bands. Aerosol vertical distribution affects the DRF estimation of each aerosol type. CMIP6 models overestimate the height of sea aerosol and underestimates the height of aerosol on land compared with CALIPSO aerosol products. DRF of CMIP6 data are much stronger than that of CALIPSO.