UCAR > Communications > UCAR Quarterly > Winter 2001 Search


Winter 2001

New CCSM is nearly ready for prime time

by Bob Henson

Developers are getting ready to freeze and introduce the new Community Climate System Model for testing and community use. Dubbed CCSM 2.0, it's actually the successor to the Climate System Model (CSM). The new version has physical improvements to each component of the fully coupled system. Along with a number of fixes that improve its performance, the CCSM includes a broader set of components from across the Earth system.

Combining water vapor and precipitation, this image was extracted from a high-resolution control run of the Community Climate Model, the atmospheric component of the CCSM.

A CSM-based simulation of 20th-century climate enhances the performance of the control run (dotted) by adding the impact from volcanic aerosols (solid black). The wide green band shows the observed temperature and its range of uncertainty in degrees Celsius. The dark green line shows an experiment forced with solar irradiance, volcanoes, tropospheric sulfates, and greenhouse gases. (Illustrations courtesy NCAR/CGD and NCAR/SCD Visualization Lab.)

When the CSM debuted in 1996, it was the first fully coupled model that produced a stable simulation of surface temperature without the need for flux corrections, which are ad hoc adjustments to energy exchange between the atmosphere and ocean. The model also employed a flux coupler to exchange information across the components. The CSM took several years of intense, concerted effort within NCAR's Climate and Global Dynamics Division. In contrast to that centralized push, the CCSM has been a more widely distributed project. Each component model was developed under the leadership of scientists both at NCAR and across the UCAR community.

"I don't know of any other modeling effort in the world that involves such a large, interconnected effort," says Jeffrey Kiehl (NCAR), chair of the model's scientific steering committee. "The original CSM involved some people from the community, but we weren't ready logistically at that point to involve the broader community in all aspects of development. We couldn't go out and say, Let's build a community. You have to have something to build a community around. It's taken some time to build that infrastructure, but I think we now have it."

NCAR's James Hack, co-chair of the atmospheric component, thinks that "some of the simulation characteristics are going to be radically improved" in CCSM 2.0. Vertical resolution in the atmosphere has increased to 26 levels from the former 18, says Kiehl. "A lot of those new levels were put in around the tropopause, where we wanted to get the exchange of air masses between the troposphere and stratosphere a little more accurately." Tropical climate will be more realistic, with a better distribution of rainfall and water vapor and more realistic wind fields in the upper troposphere.

While the atmosphere's horizontal resolution is unchanged (T42, or 2.8° latitude by 2.8° longitude), the ocean model now has more than a twofold increase in resolution, allowing a more realistic depiction of processes such as equatorial ocean currents. Some biogeochemical exchanges will also be included in CCSM 2.0, with more on the way in future versions.

The big picture

Measuring the progress of any global model is a unique challenge. Although satellites now assess rainfall and water vapor over parts of the ocean, there is no fully global data set showing how water in the atmosphere varies horizontally and vertically over time.

One common index of model performance—average global temperature—is of keen interest to both modelers and policy makers. Among the most solid evidence of success for CCSM is a series of 20th-century runs using the PaleoCSM, a version used for studying past climatic eras. With the addition of four known forcing factors—volcanoes, aerosols, greenhouse gases, and the Sun—the PaleoCSM caught most of the major peaks and valleys in average global temperature over the past century. Kiehl also cites the fully interactive sulfur chemistry and the realistic depiction of the El Niño/Southern Oscillation as two key strengths that have helped make the model state-of-the-art.

NCAR's recent hardware expansion (see IBM) should allow for more frequent CCSM runs, and some long climate simulations are being carried out at Lawrence Berkeley National Laboratory and other U.S. Department of Energy sites as well. NCAR and DOE will continue to work closely together as the DOE-funded Parallel Climate Model, developed by NCAR scientists Warren Washington and Gerald Meehl and colleagues, is merged with the CCSM over the next year. A new software engineering group within CGD is heavily involved in tuning the CCSM so that it runs as efficiently as possible on multiple platforms.

University scientists have been active in developing aspects of the model components (e.g., atmosphere, land, and sea ice) and the experiments to be run with the new model. CCSM output will also be openly available to the community. For instance, says Kiehl, "A university might downscale a CCSM run and use it to force their regional climate model." Also, some model components are being tested and refined by university researchers. Leo Donner (Princeton University/Geophysical Fluid Dynamics Laboratory) is working on a convective parameterization scheme, as is a group at Colorado State University headed by David Randall. Christopher Bretherton (University of Washington) is developing a new treatment of the boundary layer.

The agenda for 2002

Plans are to freeze the CCSM in January, then carry out several multicentury simulations sufficient for assessing the model's skill. "We also need to document everything we've done," Kiehl explains. He adds that, although the temptation to tinker with the CCSM is a strong one, every model has to stand still for awhile. "It's the old tug between continued development and the desire to have a stable model for scientific research. It's a difficult issue, but I've been personally committed to getting CCSM 2.0 out the door."

A good place for newcomers to get up to speed is the annual CCSM workshop, held in Breckenridge over three days each June. It's been growing by leaps and bounds, says Kiehl. "Six years ago, we had 100 people, about half from NCAR. Last summer we had 250 participants, most of them from the community." Details on the 2002 workshop can be found on the CCSM Web site below. Those interested in the CCSM are encouraged to visit the site for more information on the various working groups and their activities.

On the Web:
Community Climate System Model
CCSM Working Group cochairs

Some of the progress in CCSM model components

Atmosphere

Improvements in radiative transfer through both clear air and clouds. Improved distribution of precipitable water. More accurate surface solar radiation budget in eastern oceans and wind stress across eastern Pacific. Highly flexible framework for the continued exploration of alternative configurations, including boundary layer, dynamical cores, stratiform clouds, and parameterizations of radiation and moist convection.

Ocean

Based on the Parallel Ocean Program (POP) code, designed to run efficiently on massively parallel machines. North Pole displaced into Greenland to eliminate the need to filter Arctic Ocean simulations. Improved horizontal resolution, now averaging less than 1°. Anisotropic horizontal viscosity, including a flow-dependent form that allows realistic equatorial currents in the PaleoCSM. More accurate numerical implementation of Gent-McWilliams eddy parameterization using a constant coefficient. Effects of river runoff now included directly on oceans and through marginal seas. Substantially improved equation of state for seawater.

Land

Completely new biogeophysical and hydrological parameterizations, obtained through comparisons among several land models. Major reductions in cold bias of summer land-

surface temperature across Northern Hemisphere. Improved simulation of snow cover and seasonality of runoff. Transport of river runoff to the oceans at 0.5° resolution.

Sea ice

Completely new dynamical model of sea ice to better represent the evolution of its extent and thickness. New thermodynamic model that allows for a distribution of sea ice thicknesses and roughnesses.


In this issue... Other issues of UCAR Quarterly
UCARNCARUOP

UCAR > Communications > UCAR Quarterly > Winter 2001 Search

Edited by Bob Henson, bhenson@ucar.edu
Prepared for the Web by Jacque Marshall
Last revised: Thu Dec 20 16:42:17 MST 2001