Aerosol Indirect Effects on Cirrus Clouds based on NASA Flight Campaigns and Global Climate Models

Cirrus clouds have a large coverage of Earth’s surface up to 30% to 40%. Their radiative effects are significantly influenced by their microphysical properties, such as ice water content, ice crystal number concentrations and mean diameter. In this work, we examine aerosol indirect effects on cirrus microphysical properties based on global-scale airborne in-situ observations and climate model simulations. The in-situ observations include seven NSF and five NASA flight campaigns. Simulations of two global climate models are examined, including NASA GEOS-5 and NCAR CESM2/CAM6 models. Cirrus clouds are separated into five evolution phases, from clear-sky ice supersaturated regions, to nucleation, early growth, later growth, and sedimentation phases. Varying aerosol indirect effects are quantified throughout the lifetime of cirrus clouds. A machine learning method is used to contrast influences of various factors on cirrus microphysical properties, including aerosol number concentrations, temperature, relative humidity, and vertical velocity. Aerosol indirect effects in clean and polluted conditions are contrasted, using chemical tracers such as carbon monoxide, ozone, and black carbon.

Large aerosols (> 500 nm) were found to have stronger aerosol indirect effects compared with small aerosols during cirrus nucleation phase. As cirrus evolves into early growth and later growth phases, small aerosols (> 100 nm) show similar effects to large aerosols. Furthermore, aerosol indirect effects of large aerosols were found to be stronger at clean conditions compared with polluted conditions. These results indicate that heterogeneous freezing is more impactful at the nucleation phase especially under clean conditions with relatively lower carbon monoxide concentrations. On the other hand, homogeneous freezing becomes almost equally important at later evolution phases. Both GEOS-5 and CAM6 models underestimate aerosol indirect effects on cirrus clouds. The model simulations show larger underestimations of heterogeneous freezing than homogeneous freezing, demonstrating the necessity of more investigation to diagnose specific reasons behind these model biases.