The vertical distribution of black carbon in CMIP5models: Comparison to observations and the importance of convective transport

Large uncertainty in the direct radiative forcing of black carbon (BC) exists, with published estimates ranging from 0.25 to 0.9 W m(-2). A significant source of this uncertainty relates to the vertical distribution of BC, particularly relative to cloud layers. We first compare the vertical distribution of BC in Coupled Model Intercomparison Project Phase 5 (CMIP5) models to aircraft measurements and find that models tend to overestimate upper tropospheric/lower stratospheric (UT/LS) BC, particularly over the central Pacific from Hiaper Pole-to-Pole Observations Flight 1 (HIPPO1). However, CMIP5 generally underestimates Arctic BC from the Arctic Research of the Composition of the Troposphere from Aircraft and Satellites campaign, implying a geographically dependent bias. Factors controlling the vertical distribution of BC in CMIP5 models, such as wet and dry deposition, precipitation, and convective mass flux (MC), are subsequently investigated. We also perform a series of sensitivity experiments with the Community Atmosphere Model version 5, including prescribed meteorology, enhanced vertical resolution, and altered convective wet scavenging efficiency and deep convection. We find that convective mass flux has opposing effects on the amount of black carbon in the atmosphere. More MC is associated with more convective precipitation, enhanced wet removal, and less BC below 500 hPa. However, more MC, particularly above 500 hPa, yield more BC aloft due to enhanced convective lofting. These relationshipsparticularly MC versus BC below 500 hPa-are generally stronger in the tropics. Compared to the Modern-Era Retrospective Analysis for Research and Applications, most CMIP5 models overestimate MC, with all models overestimating MC above 500 hPa. Our results suggest that excessive convective transport is one of the reasons for CMIP5 overestimation of UT/LS BC.