Improvement of Convective Ice Parameterization for the NASA GISS GCM and Impacts on Cloud Ice Simulation

Poster PDF

Description

Detrainment of ice from convective updrafts can affect GCM climate sensitivity (Zhao 2014, J. Clim.). This is a challenge for GCMs since detrainment is determined by both the vigor of convective updrafts and the microphysics in the updraft. The CMIP5 GISS Model E2 GCM simulated a mid- and high-latitude cloud ice water path (IWP) ~50% less than that for CMIP3 (Jiang et al. 2012, JGR). Tropical IWP increased by ~15% in CMIP5. While the tropical IWP was still within the published upper-bounds of IWP uncertainty, the 200 mb ice water content (IWC) was about twice the published upper-bound, largely due to IWC in deep-convecting regions. Recent advances in the Model E2 cumulus parameterization have had a substantial impact on tropical IWC. These include a cold pool parameterization to differentiate strong from weak entrainment situations and regulate updraft vertical velocity (Del Genio et al. 2015, J. Clim.) and improvements to the treatment of ice in the convective updraft. In this poster we describe an improved parameterization of convective cloud ice developed using in situ observations from the MC3E, SPartICus, AMMA, and TC4 field campaigns. High-resolution satellite IR and microwave brightness temperatures are mapped to campaign aircraft flight paths, and joint brightness temperature thresholds are used to sub-sample in situ particle size distribution (PSD) estimates associated with deep convective outflow. We also utilize new ice particle terminal velocity formulations derived from multiple field campaigns (Heymsfield et al. 2013, JAS). The PSD and fall speeds are combined with the GCM’s parameterized updraft vertical velocity to partition convective condensate into precipitating and detrained components. The improved parameterization predicts an ice mass weighted-average particle diameter and a particle cross sectional area weighted-average size diameter as a function of temperature and IWC. Assuming a gamma-distribution form for the PSD, these two diameter estimates are all that are needed to predict the concentration of ice particles as a function of particle diameter in the convective plume. We first test the convective ice parameterization in the GISS SCM at times in which convection is locally forced at the SGP. We then evaluate GCM simulations with the improved parameterization and find a ~50% decrease in upper tropospheric IWC, bringing the tropical and global mean IWP climatologies into closer agreement with satellite best estimates.