As climate models evolved through the 1990s, scientists
began to shift focus from reproducing general circulation to experimenting
with the feedbacks of climatic processes due to increasing greenhouse
gases, changing ocean currents and the way the model responds to
forced perturbations such as ENSO. As the next generation of models
come out improvements in the models make them more reliable for
global predictions and more capable of regional analyses. Here is
a simplified description of the anatomy of the latest version of
the Community Climate System Model: CCSM3.
At their core, all GCMs employ a specific set of primitive dynamic equations, which
allow the atmosphere to move in three dimensions, warm up and cool
down, and transport moisture, etc. These equations are solved over
and over again at specified locations in the model's three-dimensional
space. There are two main methods for establishing the horizontal
domain of a model. The simplest is to establish a grid along lines
of latitude and longitude. For example, the CCSM3 can be run on
a 2o x 2.5o grid. Another method is to treat atmospheric motion
as waves using Fast Fourier Transforms
(FFTs) to make the spectral conversion. The
resolution then is represented as the number of waves that can be
represented around the earth. The CCSM3 uses wave numbers of T31,
T42, and T85; where T represents the triangular truncation of the
Fourier transform. This resolution can be approximated to longitude/latitude
with a resolution of T31 and T85 is 3.75o
and 1.41o respectively.
The vertical domain in the CCSM3 is represented by 26 levels, but is complicated by the fact that the atmosphere is compressible and gets exponentially less dense as you move up in altitude. Therefore, the model levels are irregularly spaced so as to have the most levels in the troposphere where most of the weather and interaction between climatic processes occurs. In addition, topography on the Earth's surface creates difficulties with using pressure as the vertical coordinate because in many locations the ground intersects pressure levels. The CCSM3, as most models, uses a variation of the terrain following coordinate called sigma ( and is defined as:
σ = p/ps
where p = pressure and ps = pressure at the surface.
After the atmospheric core of the model has been constructed modelers must try to incorporate all of the other climatic processes and feedback mechanisms that influence climate so as to have an accurate, dynamic representation of the climate system. While the models of the past focused on the atmosphere and sometimes included the oceans, today's models contain separate modules for the land surface, oceans, and sea ice and sometimes include atmospheric chemistry and advanced treatment of aerosols.
The land surface component of the CCSM3 uses the same horizontal grid as the atmospheric component and has 10 subsurface layers to account for soil-atmosphere interactions. The land surface can also be classified as a variety of types including ice, water, urban, and vegetation. These distinctions are important for the radiation balance because the albedo of the land surface can change dramatically. For example, the albedo of urban black top is very close to 0, meaning it absorbs almost all radiation whereas the albedo of snow cover or white sand is closer to 1 meaning it reflects most radiation.
The ocean and sea ice modules of the CCSM3 use a slightly different horizontal grid from the atmosphere, although they have similar horizontal resolution. In addition, the ocean component uses either 25 or 40 vertical levels defined by depth, extending down to the ocean's deepest basins.
These components make up the newest version of the CCSM3 GCM and can be run in different configurations, primarily by varying the horizontal and vertical resolutions. It should be noted, however, that running the finest resolution configuration takes more than 1100 hours of computer time to simulate one year of the atmosphere. That is approximately 46 days and experiments looking at trends even 10 years into the future take significant time to complete. That is why, for the longer period experiments, scientists use the more coarse resolution.
GCMs have become integral for helping scientists study the Earth's large-scale circulations, forecasting interannual variability such as ENSO, and evaluating the possibility of climate change in the decades to come. GCMs differ from other models mostly in their spatial and temporal domains and the inclusion of many processes not needed for other models because of the longer time scales involved. Their spatial domain covers the whole globe as opposed to, for example, a numerical weather prediction (NWP) model which may cover only North America. On the temporal scale they attempt to simulate earth's atmosphere from periods of several months to several decades, whereas NWP models can forecast for periods as short as a few hours and up to several days relatively well.
Go To Climate Models in Action