Over the last fifteen years, the NCAR Climate and Global Dynamics (CGD) Division has provided a comprehensive, three-dimensional global atmospheric model to university and NCAR scientists for use in the analysis and understanding of global climate. Because of its widespread use, the model was designated a community tool and given the name Community Climate Model (CCM). The original versions of the NCAR Community Climate Model, CCM0A  and CCM0B , were based on the Australian spectral model [24,126] and an adiabatic, inviscid version of the ECMWF spectral model . The CCM0B implementation was constructed so that its simulated climate would match the earlier CCM0A model to within natural variability (e.g. incorporated the same set of physical parameterizations and numerical approximations), but also provided a more flexible infrastructure for conducting medium- and long-range global forecast studies. The major strength of this latter effort was that all aspects of the model were described in a series of technical notes, which included a Users' Guide , a subroutine guide which provided a detailed description of the code  a detailed description of the algorithms , and a compilation of the simulated circulation statistics . This development activity firmly established NCAR's commitment to provide a versatile, modular, and well-documented atmospheric general circulation model that would be suitable for climate and forecast studies by NCAR and university scientists. A more detailed discussion of the early history and philosophy of the Community Climate Model can be found in Anthes .
The second generation community model, CCM1, was introduced in July of 1987, and included a number of significant changes to the model formulation which were manifested in changes to the simulated climate. Principal changes to the model included major modifications to the parameterization of radiation, a revised vertical finite-differencing technique for the dynamical core, modifications to vertical and horizontal diffusion processes, and modifications to the formulation of surface energy exchange. A number of new modeling capabilities were also introduced, including a seasonal mode in which the specified surface conditions vary with time, and an optional interactive surface hydrology that followed the formulation presented by Manabe . A detailed series of technical documentation was also made available for this version [190,184,65,15] and more completely describe this version of the CCM.
The most ambitious set of model improvements occurred with the introduction of the third generation of the Community Climate Model, CCM2, which was released in October of 1992. This version was the product of a major effort to improve the physical representation of a wide range of key climate processes, including clouds and radiation, moist convection, the planetary boundary layer, and transport. The introduction of this model also marked a new philosophy with respect to implementation. The CCM2 code was entirely restructured so as to satisfy three major objectives: much greater ease of use, which included portability across a wide range of computational platforms; conformance to a plug-compatible physics interface standard; and the incorporation of single-job multitasking capabilities.
The standard CCM2 model configuration was significantly different from its predecessor in almost every way, starting with resolution where the CCM2 employed a horizontal T42 spectral resolution (approximately 2.8 x 2.8 degree transform grid), with 18 vertical levels and a rigid lid at 2.917 mb. Principal algorithmic approaches shared with CCM1 were the use of a semi-implicit, leap frog time integration scheme; the use of the spectral transform method for treating the dry dynamics; and the use of a bi-harmonic horizontal diffusion operator. Major changes to the dynamical formalism included the use of a terrain-following hybrid vertical coordinate, and the incorporation of a shape-preserving semi-Lagrangian transport scheme  for advecting water vapor, as well as an arbitrary number of other scalar fields (e.g. cloud water variables, chemical constituents, etc.). Principal changes to the physics included the use of a -Eddington approximation to calculate solar absorption ; the use of a Voigt line shape to more accurately treat infrared radiative cooling in the stratosphere; the inclusion of a diurnal cycle to properly account for the interactions between the radiative effects of the diurnal cycle and the surface fluxes of sensible and latent heat; the incorporation of a finite heat capacity soil/sea ice model; a more sophisticated cloud fraction parameterization and treatment of cloud optical properties ; the incorporation of a sophisticated non-local treatment of boundary-layer processes ; the use of a simple mass flux representation of moist convection , and the optional incorporation of the Biosphere-Atmosphere Transfer Scheme (BATS) of Dickinson et al. . As with previous versions of the model, a User's Guide  and model description  were provided to completely document the model formalism and implementation. Control simulation data sets were documented in Williamson .
The CCM3 was the fourth generation in the series of NCAR's Community Climate Model. Many aspects of the model formulation and implementation were identical to the CCM2, although there were a number of important changes that were incorporated into the collection of parameterized physics, along with some modest changes to the dynamical formalism. Modifications to the physical representation of specific climate processes in the CCM3 were motivated by the need to address the more serious systematic errors apparent in CCM2 simulations, as well as to make the atmospheric model more suitable for coupling to land, ocean, and sea-ice component models. Thus, an important aspect of the changes to the model atmosphere was that they address well known systematic biases in the top-of-atmosphere and surface (to the extent that they were known) energy budgets. When compared to the CCM2, changes to the model formulation fell into five major categories: modifications to the representation of radiative transfer through both clear and cloudy atmospheric columns, modifications to hydrological processes (i.e., in the form of changes to the atmospheric boundary layer, moist convection, and surface energy exchange), the incorporation of a sophisticated land surface model, the incorporation of an optional slab mixed-layer ocean/thermodynamic sea-ice component, and a collection of other changes to the formalism which did not introduce significant changes to the model climate.
Changes to the clear-sky radiation formalism included the incorporation of minor CO bands trace gases (, , , ) in the longwave parameterization, and the incorporation of a background aerosol (0.14 optical depth) in the shortwave parameterization. All-sky changes included improvements to the way in which cloud optical properties (effective radius and liquid water path) were diagnosed, the incorporation of the radiative properties of ice clouds, and a number of minor modifications to the diagnosis of convective and layered cloud amount. Collectively these modification substantially reduced systematic biases in the global annually averaged clear-sky and all-sky outgoing longwave radiation and absorbed solar radiation to well within observational uncertainty, while maintaining very good agreement with global observational estimates of cloud forcing. Additionally, the large warm bias in simulated July surface temperature over the Northern Hemisphere, the systematic over-prediction of precipitation over warm land areas, and a large component of the stationary-wave error in CCM2, were also reduced as a result of cloud-radiation improvements.
Modifications to hydrological processes included revisions to the major contributing parameterizations. The formulation of the atmospheric boundary layer parameterization was revised (in collaboration with Dr. A. A. M. Holtslag of KNMI), resulting in significantly improved estimates of boundary layer height, and a substantial reduction in the overall magnitude of the hydrological cycle. Parameterized convection was also modified where this process was represented using the deep moist convection formalism of Zhang and McFarlane  in conjunction with the scheme developed by Hack  for CCM2. This change resulted in an additional reduction in the magnitude of the hydrological cycle and a smoother distribution of tropical precipitation. Surface roughness over oceans was also diagnosed as a function of surface wind speed and stability, resulting in more realistic surface flux estimates for low wind speed conditions. The combination of these changes to hydrological components resulted in a 13% reduction in the annually averaged global latent heat flux and the associated precipitation rate. It should be pointed out that the improvements in the radiative and hydrological cycle characteristics of the model climate were achieved without compromising the quality of the simulated equilibrium thermodynamic structures (one of the major strengths of the CCM2) thanks in part to the incorporation of a Sundqvist  style evaporation of stratiform precipitation.
The CCM3 incorporated version 1 of the Land Surface Model (LSM) developed by Bonan  which provided for the comprehensive treatment of land surface processes. This was a one-dimensional model of energy, momentum, water, and CO exchange between the atmosphere and land, accounting for ecological differences among vegetation types, hydraulic and thermal differences among soil types, and allowing for multiple surface types including lakes and wetlands within a grid cell. LSM replaced the prescribed surface wetness, prescribed snow cover, and prescribed surface albedos in CCM2. It also replaced the land surface fluxes in CCM2, using instead flux parameterizations that included hydrological and ecological processes (e.g. soil water, phenology, stomatal physiology, interception of water by plants).
The fourth class of changes to the CCM2 included the option to run CCM3 with a simple slab ocean-thermodynamic sea ice model. The model employs a spatially and temporally prescribed ocean heat flux and mixed layer depth, which ensures replication of realistic sea surface temperatures and ice distributions for the present climate. The model allowed for the simplest interactive surface for the ocean and sea ice components of the climate system.
The final class of model modifications included a change to the form of the hydrostatic matrix which ensures consistency between and the discrete continuity equation, and a more generalized form of the gravity wave drag parameterization. In the latter case, the parameterization was configured to behave in the same way as the CCM2 parameterization of wave drag, but included the capability to exploit more sophisticated descriptions of this process.
One of the more significant implementation differences with the earlier model was that CCM3 included an optional message-passing configuration, allowing the model to be executed as a parallel task in distributed-memory environments. This was an example of how the Climate and Global Dynamics Division continued to invest in technical improvements to the CCM in the interest of making it easier to acquire and use in evolving computational environments. As was the case for CCM2, the code was internally documented, obviating the need for a separate technical note that describes each subroutine and common block in the model library. Thus, the Users' Guide, the land surface technical note, the CCM3 technical note , the actual code and a series of reviewed scientific publications (including a special issue of the Journal of Climate, Volume 11, Number 6) were designed to completely document CCM3.