Getting the Ice Right in Global Models

Aircraft data show that ice particles are smaller and fall faster than models had assumed.

The Science

Depending on their height and thickness, ice clouds have the potential to either warm or cool Earth’s surface. Therefore, getting the details of these clouds right in global climate model (GCM) simulations is an important step toward increasing the accuracy of future climate projections. Aircraft observations from multiple field campaigns sponsored by the Department of Energy’s Atmospheric Radiation Measurement (ARM) Climate Research Facility and National Aeronautics and Space Administration (NASA) show that ice particles detrained from deep convective clouds (thunderstorms) are smaller and fall faster than previously assumed. Scientists used this new knowledge to improve the representation of ice clouds in the NASA Global Institute for Space Studies (GISS) GCM, thereby also improving GCM simulation of ice clouds in and near regions of active convection.

The Impact

The GISS GCM previously produced too much cloud ice, especially in convective regions in the tropics and midlatitudes where deep, raining clouds are found. Using a new ice cloud formulation based on in situ aircraft observations, model simulations show a 30% to 50% decrease in the upper tropospheric deep convective ice water content, bringing the model results into better agreement with global satellite observations.


Recent studies showed that the GISS GCM produced upper tropospheric ice water contents that exceeded an estimated upper bound by a factor of 2. Scientists traced this issue to the approach used in the GCM for partitioning ice formed in deep convective updrafts into falling (i.e., snow) and lofted/detrained (i.e., cloud) components. They analyzed aircraft observations of ice clouds adjacent to deep convective cloud cores to develop new observational benchmarks for ice particle sizes and fall speeds. Observations used in the study include data from the ARM-NASA Midlatitude Continental Convective Clouds Experiment (MC3E) and ARM Small Particles in Cirrus (SPARTICUS) campaign.
Based on the aircraft observations, researchers determined that the convective ice particles in the model were often too large and fell too slowly. To correct this issue, the researchers developed new empirical relationships for the sizes and fall speeds of ice particles near active convection and implemented those relationships into the GCM convective parameterization. Because ice particles in deep clouds are smaller, but fall faster, there is an overall decrease in cloud ice water content in deep convective regions. The new cloud ice simulation agrees better with global satellite retrievals. The study highlights the value of using multiple field campaign and satellite observations in both the GCM development step and subsequent GCM evaluation step.

Principal Investigator(s)

Gregory S. Elsaesser
Columbia University/Goddard Institute for Space Studies


This research was funded by the National Aeronautics and Space Adminstration (NASA) Modeling and Analysis Program, Precipitation Measurement Missions, CloudSat / CALIPSO Mission; and Department of Energy, Office of Science, Office of Biological and Environmental Research, Atmospheric System Research program. Computing resources were provided by the NASA High-End Computing Program through the NASA Center for Climate Simulation at the Goddard Space Flight Center, with additional support from the NASA Global Institute for Space Studies and Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA.


G. S. Elsaesser, A. D. Del Genio, J. H. Jiang, and M. van Lier-Walqui, “An Improved Convective Ice Parameterization for the NASA GISS Global Climate Model and Impacts on Cloud Ice Simulation.” Journal of Climate 30, 317-336 (2017). DOI: 10.1175/JCLI-D-16-0346.1.