Direct Comparisons of Ice Microphysical Properties Simulated by the Community Atmosphere Model CAM5.4 with ARM SPartICus Observations

Poster PDF

Description

Cirrus clouds in the upper troposphere play a key role in the Earth radiation budget, and their radiative forcing depends strongly on number concentration and size distribution of ice particles. In this study we evaluate the cloud microphysical properties simulated by the Community Atmosphere Model version 5.4 (CAM5.4) against the Small Particles in Cirrus (SPartICus) observations over the ARM South Great Plain (SGP) site in April 2010. Model simulation is performed using specific dynamics to preserve prognostic meteorology (U, V, and T) close to GEOS-5 analysis. Model results collocated with SPartICus flight tracks spatially and temporally are directly compared with the observations. We compare CAM5.4 simulated ice crystal number concentration (Ni), ice particle size distribution, ice water content (IWC), and Ni and IWC co-variances with temperature and vertical velocity with the statistics from SPartICus observations. All analyses are restricted to T ≤ -40°C. Model sensitivity tests are performed with different ice nucleation mechanisms (homogeneous versus heterogeneous nucleation) to reflect the uncertainties in cirrus parameterizations. In addition, the vertical velocity variability produced from orographic gravity waves (GWs) on ice microphysics is tested in the model. We find that (1) both modeled and observed Ni has a strong correlation with temperature, indicating a dominant role of homogeneous ice nucleation in the ice formation compared to the heterogeneous nucleation in the midlatitude cirrus over the SGP site, (2) GWs enhance the ice number concentrations significantly over the Rocky Mountain regions, and their effect should be included in the global model, and (3) the model tends to underestimate Ni for small ice particles (with diameter smaller than 50 µm). Moreover, the mean slope parameters (λ) of the gamma function for ice/snow crystal size distribution at a certain temperature as simulated are about 2-3 times as large as those derived from the observations. This indicates a distinctly smaller size for ice crystals by the model.