Origins of a relatively tight lower bound on anthropogenic aerosol radiative forcing from Bayesian analysis of historical observations (2021)

Anna Lea Albright, Cristian Proistosescu, and Peter Huybers

Data availability: Historical observed global-mean surface temperatures are from Cowtan and Way, 2014, version 2.0. IPCC AR5 effective radiative forcing data, Table AII.1.2 are from Myhre et al, 2013. Transient effective aerosol forcing data from the RFMIP experiment in CMIP6 were accessed at ESGF. Sulfur dioxide, black carbon, and organic carbon data are from Hoesly et al, 2018.

Abstract: A variety of empirical estimates have been published for the lower bounds on aerosol radiative forcing, clustered around -1.0 Wm^-2 or -2.0 Wm^-2. The reasons for obtaining such different constraints are not well understood. In this study, we explore bounds on aerosol radiative forcing using a Bayesian model of aerosol forcing and Earth's multi-timescale temperature response to radiative forcing. A joint inference of climate sensitivity and effective aerosol radiative forcing from historical surface temperatures since 1850 yields a maximum likelihood estimate for equilibrium climate sensitivity of 3.7°C (5-95% c.i. 1.8 to 6.7°C) and an effective anthropogenic aerosol radiative forcing of -0.70 Wm^-2 (5-95% c.i. -1.2 to -0.33 Wm^-2). The wide range of equilibrium climate sensitivity reflects difficulty in empirically constraining multi-decadal and longer-term responses using the instrumental record of surface temperature, as has been noted previously. A relatively tight bound on aerosol forcing is nonetheless obtained from the structure of temperature and aerosol precursor emissions, principally the rapid growth in precursor emissions between 1950--1980. Uncertainty in effective volcanic forcing contributes little uncertainty to effective anthropogenic aerosol radiative forcing because their time-variable contributions weakly co-vary. Obtaining a -2 Wm^-2 5-percentile aerosol forcing lower bound would require prescribing internal climate variance that is a factor of four larger than found in simulations and a comparable variance in global temperature observations. Ocean heat uptake observations may further constrain aerosol radiative forcing but require a better understanding of the relationship between time-variable radiative feedbacks and net radiative forcing.