Impact of Small Ice Crystals on GCM Simulations

David L. Mitchell¹, Philip Rascht², Dorothea Ivanova³,
Greg McFarquhar⁴, and Timo Nousiainen⁵

1) Desert Research Institute, Reno, NV
2) National Center for Atmospheric Reseach, Boulder, Colorado
3) Embry-Riddle Aeronautical University, Prescott, Arizona
4) University of Illinois, Urbana, Illinois
5) University of Helsinki, Finland

This research was supported by the National Science Foundation (NSF),
Grant No. ATM-0413401.

Abstract

In the prediction of climate change, the greatest uncertainty lies in representing the clouds. Ice clouds are particularly challenging, and to date there is no accepted method for measuring the concentrations of small (D < 60 µm) ice crystals. This study examines the sensitivity of a global climate model to different assumptions regarding the number concentrations of small ice crystals when they are allowed to affect ice sedimentation rates. When their concentrations are relatively high, the GCM predicts a 12% increase in cloud ice amount and a 5.5% increase in cirrus cloud coverage globally. This produces a net cloud forcing of -5 W m^-2 in the tropics and warms the upper tropical troposphere over 3 °C. Ice crystal concentration differences assumed were modest in comparison to corresponding measurement uncertainties, revealing a potentially large source of uncertainty in the prediction of global climate.

The GCM Experiment:

The temperature dependence of the median mass-dimension (D_m) of the PSD is illustrated above for the PSD1 and PSD2 size distribution schemes. The ice fall speed for a given mode of the PSD is based on D_m and crystal shape. The PSD1 scheme assumes mostly bullet rosettes, while the PSD2 scheme assumes mostly planar polycrystals.

The overall ice fall speed for the PSD, V_t, is weighted by the ice water content of each mode of the PSD times the corresponding fall velocity. Note that V_t does not depend on the total PSD ice water content, IWC. The temperature dependence of V_t is shown above for PSD1 and PSD2.

Summary

1. Two temperature dependent parameterizations of ice particle size distributions (PSD) were incorporated into NCAR’s Community Atmosphere Model (CAM3), along with corresponding ice mass sedimentation rates and radiative properties. The main difference between PSD schemes was the concentration of small (D < 60 µm) ice crystals.
2. Higher small ice crystal concentrations resulted in lower ice fall speeds and greater cirrus IWP and coverage. This produced large changes in cloud forcing relative to the simulation with lower small crystal concentrations (PSD1), with a net cloud forcing in the tropics of -5 W m^-2. Temperatures in the upper tropical troposphere were over 3 °C warmer relative to PSD1.
3. The PSD schemes used here underestimate the observed uncertainty in small ice crystal concentrations from in situ measurements. As the cloud microphysics in GCMs becomes more realistic, these large uncertainties in small ice crystal concentrations may produce large uncertainties in surface warming in GCM CO₂ doubling experiments. A greater focus on this problem appears warranted.