Climate Modeling Section

The Climate Modeling Section's (CMS) research goal is to increase the understanding of the atmosphere and its role in the climate system through modeling and observational studies and to represent that understanding in atmospheric models. This involves studies of the physical mechanisms governing the global climate system and the numerical modeling techniques required to represent these mechanisms. CMS scientists compare observed atmospheric data with model output data to validate and to improve the models on a wide range of time scales. They have had strong input to the development of the most recent generation of the atmospheric model, Community Atmosphere Model (CAM2), and have been primarily responsible for the implementation and testing of the model. A CMS scientist serves as co-chair of the Communitity Climate System Model (CCSM) Atmosphere Model Working Group (James Hack), where all of the other CMS scientists are active participants in AMWG activities.

CCM/CCSM Development

William Collins (CMS) developed a new parameterization for the absorption and emission of longwave radiation by water vapor for use in NCAR's climate models.  Since water vapor contributes 60-70% of the natural greenhouse effect, it is important to treat the effects of water vapor on longwave radiation as accurately as possible.  This new treatment preserves the formulation of the radiative transfer equations using the absorptivity/emissivity method.  However, the components of the absorptivity and emissivity related to water vapor have been replaced with new terms calculated with the General Line-by-line Atmospheric Transmittance and Radiance Model (GENLN2).  In the current version of GENLN2, the parameters for H2O lines have been obtained from the high-resolution transmission molecular absorption (HITRAN-96) database, and the continuum is treated using the Clough, Kneizys, and Davies (CKD) model version 2.1.  The new treatment replaces the original parameterization developed for the Community Climate Model (CCM), although it retains the explicit dependence on emission and path temperature introduced in the original scheme for greater accuracy.  The mean absolute errors in the surface and top-of-atmosphere clear-sky longwave fluxes for standard atmospheres are reduced to less than 1 W/m2.  Mean absolute differences between the cooling rates from the original method and GENLN2 are typically 0.2 K/day.  These differences are reduced by at least a factor of 3 using the new parameterization. The new parameterization increases the longwave cooling at 300 mb by 0.4 to 0.7 K/day, and it decreases the cooling near 800 mb by 0.2 to 0.6 K/day.  The increased cooling is caused by line absorption and the foreign continuum in the rotation band, and the decreased cooling is caused by the self-continuum in the rotation band.  These changes in the vertical profile of longwave cooling interact strongly with the parameterization of convection.  The implications of the changes for climate sensitivity are under active investigation.

James Hack (CMS) has continued work on evaluating the interaction of convection and atmospheric boundary layer processes with the large-scale radiation field as simulated in the global model.  The incorporation of a more accurate treatment of longwave radiative transfer produced a substantial drying of the mid to low troposphere, largely in response to enhanced moist convective overturning accompanying the large increase in upper tropospheric radiative cooling.  Using equilibrium and transient single-column model simulations, Hack and  John Truesdale (CMS) incorporated a rainwater evaporation mechanism in the Zhang-McFarlane parameterization of moist convection.  This mechanism mitigates a severe dry bias in the global simulation, while making only very minor changes to the precipitation distribution.  It also has the effect of locally regulating moist convection in regions where it had been excessively active.  These local changes in diabatic heating have a significant and positive impact on the local dynamical circulation. As in the case of the updated radiative transfer parameterization, the implications of this precipitation evaporation process for climate sensitivity are being investigated.

Hack has also continued work on a modified Arakawa-Schubert treatment of convection.  The main emphasis for this implementation is to give shallow convection a larger role in the reduction of boundary-layer based Convective Available Potential Energy (CAPE), which will be dealt with using a linear optimization approach, and to develop a computational approach that allows the closure to be implemented as a function of the change in CAPE as opposed to a relaxation approach toward climatological values (a common implementation of the quasi-equilibrium closure).

David Williamson (CMS) and Jerry Olson (CMS) uncovered indications of the effect of time truncation errors in subgrid scale parameterizations in a series of aqua-planet simulations. These are simulations in which the earth is assumed to be covered with water with the sea surface temperature specified to have a simple geometry, such as zonal symmetry.  With a shorter time step, the latitudinal structure of the zonally averaged precipitation has a single precipitation peak centered on the equator. With a longer time step it has a double peak straddling the equator with a minimum centered on the equator. This error was exposed using simulations with the semi-Lagrangian version of CCM3, which allows a longer time step than the standard Eulerian version. The different structure is shown to be caused primarily by the different time step in the parameterizations rather than by different truncation errors introduced by the dynamical cores themselves when run with different time steps.  Different diffusive smoothing associated with different spectral resolutions introduces a secondary effect in determining the structure.

Williamson and Olson argue that the double structure forms in the CCM3 with the longer time step because more water is put into the atmosphere over a longer discrete time step, the evaporation rate being the same.   The additional water vapor in the region of equatorial moisture convergence results in more CAPE farther from the equator.  This allows convection to initiate farther from the equator. The resulting heating drives upward vertical motion and low level convergence away from the equator, resulting in much weaker upward motion at the equator.  This feedback eliminates the instability at the equator and shuts off convection there.  The parameterizations depend on the amount of water inserted into the troposphere during a discrete time step. This is as important as the rate of insertion. They carried out experiments that support this explanation.  Their study shows that the significance of the time truncation error of parameterizations deserves further consideration in complete atmospheric general circulation models. 

Williamson and Olson have begun exploring the benefits of using operational Numerical Weather Prediction (NWP) as a means for examining parameterization methods. The hope is that it will allow direct comparison of the parameterized variables (e.g., clouds, precipitation) with observations early in the forecast while the modeled state is still near that of the atmosphere, but after initial transient computational modes are damped.  Some forecast centers have found such an approach very useful in developing and evaluating parameterizations.  Climate modeling groups not associated with a NWP center generally have not been able to take advantage of such an approach because of the large amount of work involved in developing a data ingest and assimilation system.  Nevertheless, it may be feasible to study parameterizations in climate models in a forecast mode using an ensemble of historical analyses from very recent years.  Williamson and Olson have started to develop such a method with the CCM in conjunction with the European Centre for Medium-Range Weather Forecasts (ECMWF), who are providing the analyses and verification data.  While initializing atmospheric state variables seems reasonably straightforward, methods to initialize parameterization variables, such as cloud water, which carry time history, from analyses are less obvious, and methods to initialize land variables are problematical.  It is difficult to map discrete/discontinuous land variables between different grids.  The dominant land types on the climate model grid may differ from those on the analysis grid, and there is no uniform definition of land model state variables. Williamson and Olson have devised several methods to "spin up" land variables and parameterization variables to be consistent with atmospheric analyses. They are currently evaluating these methods.  If successful, they will then examine the behavior of the parameterizations in the CAM2 in a forecast mode compared to reanalyses in regions where the analyses can be shown to represent the atmosphere, i.e., regions with adequate data so the operational model forecast does not dominate the analysis.

As co-chair of the CCSM Atmosphere Model Working Group (AMWG), Hack has coordinated and participated in the evaluation of a large suite of proposed configurations for the new CCSM CAM.  All members of the Climate Modeling Section have actively participated in the definition of many of the AMWG experimental configurations and their detailed evaluation.   Much of this work, for example, was done collaboratively with university investigators, such as Marat Khairoutinov (Colorado State University), Minghua Zhang (State University of New York, Stony Brook), and S. J. Lin (NASA Data Assimilation Office).  Prototype configurations for the CAM have included a variety of convection schemes, several longwave radiative transfer schemes, generalizations to cloud overlap parameterization, changes to vertical resolution, and the incorporation of prognostic approaches to cloud water.  The outcome of this work was a new configuration for the atmosphere that includes major improvements to the CCM3 formulation, which produces a number of significant simulation improvements.

A very important part of the CAM development and evaluation effort has included a web-based diagnostic analysis capability developed by Mark Stevens (CMS) and Hack.  These efforts have helped to ensure that changes to the prototype model were evaluated by the community in a comprehensive, systematic, and timely way.  Work continues on expanding the scope and depth of the diagnostic analysis package. 

Climate and Chemistry Research

Collins and Phililp Rasch (CMS) participated in three field programs; ACE-Asia (Aerosol Characterization Experiment in Asia); CLAMS (Chesapeake Lighthouse Aerosol & Aircraft Measurements for Satellites); and MINOS (Mediterranean Intensive Oxidant Study) in which they produced aerosol forecasts using a version of MATCH (Model for Atmospheric Transport and Chemistry) that included a recently developed aerosol assimilation capability.  The aerosol forecasts were used in the real-time deployment of aircraft and ships for each of these field programs.  Collins and Rasch have participated  in post field program research, for both these programs and the Indian Ocean Experiment (INDOEX) field experiment to understand differences in the model forecasts atmospheric analyses in order to improve the modeling capability, to  better understand aerosol forcing in the atmosphere, and to examine  aerosol/climate feedbacks.

Rasch further developed and tested new versions of MATCH. Modifications to the model include: revisions to the dust mobilization and scavenging modules (in collaboration with Charles Zender (University of California, Irvine) and Natalie Mahowald (University of California, Santa Barbara)); modifications to the model to facilitate simulations at very high horizontal and vertical resolution (e.g., T170 and 42 Levels); and modification to the model and associated data retrieval components to facilitate real time aerosol forecasts for field programs.

Rasch worked with Mahowald and Robert Plumb (Massachusetts Institute of Technology) on transport in a CCM isentropic vertical coordinate framework.  He also developed a convective trigger formulation and evaluated its behavior in the coupled and uncoupled versions of CCSM in collaboration with Charlotte Demott (Colorado State University).  Another collaborative effort involved John Bergmann (NOAA) to understand the role of cloud overlap on the model simulation by examining the sensitivity of the model simulation to different assumptions about how clouds overlap as a function of a correlation length scale.


Climate Research

Jeffrey Kiehl (CMS) in collaboration with Hack and V. Ramanathan (University of California, San Diego/Scripps Institution of Oceanography) carried out a series of CCM3 simulations to study the effects of absorbing aerosols on the climate in the Indian Ocean region. The aerosol spatial and temporal distribution was constrained by observations obtained during the INDOEX. Simulations included prescribed sea surface temperatures and a slab ocean mixed layer model. The results indicate that the presence of the absorbing aerosol lead to a cooling of the land and ocean surface, not a warming as suggested by some recent publications. The hydrologic cycle also changed significantly in the climate simulations. For example, the convective precipitation located over the northern Indian Ocean shifts further northward when absorbing aerosols are included in the simulations. The perturbations to the ocean surface energy flux are quite large in both the Bay of Bengal and the Arabian Sea. In fact, the perturbation is as large as the prescribed implied ocean heat flux in these regions. Thus, a slab ocean model may limit the interpretation of the response in these regions.

To address the issue regarding ocean heat flux, Kiehl, Warren Washington (Climate Change Research Section (CCR)), and V. Ramanathan have carried out similar absorbing aerosol experiments with the fully coupled parallel climate model (PCM). The surface temperature response from the fully coupled model is quite similar to the response seen in the slab ocean model. To investigate changes in precipitation will require longer simulations, which are under way.

Kiehl has also been involved in several collaborations using the paleoclimate configuration of the CCSM model. In collaboration with Lucinda Shellito and Lisa Sloan (University of California, Santa Cruz), he carried out a series of simulations with the paleo CSM to look at the role of clouds in regulating tropical sea surface temperatures during the Eocene (~55 million years ago).  Simulations used elevated carbon dioxide levels constrained by paleoclimate data. These studies are looking at the hypothesis that cloud feedbacks in the tropics may have constrained tropical warming relative to extratropical warming.  He has also collaborated with Zender to implement a dust model into the paleo CSM to investigate the radiative role of wind blown dust during the Last Glacial Maximum (LGM). Preliminary tests have been carried out to ensure that the present day simulated dust distribution looks reasonable. In the near future, LGM boundary conditions will be applied (e.g., ice extent, CO2 levels) to see how radiative forcing by wind blown dust may have affected the LGM climate.



Byron Boville (CMS) and Fabrizio Sassi (CMS) continued their collaboration with Rolando Garcia (Atmospheric Chemistry Division (ACD)), Douglas Kinnison (ACD), and Raymond Roble (High Altitude Observatory (HAO)) on the development of a Whole Atmosphere CCM (WACCM).  The development of WACCM is motivated by an appreciation of the importance of coupling between atmospheric regions. Vertical propagation of atmospheric waves and the transport of minor species from the troposphere are known to play major roles in the dynamics and chemistry of the middle and upper atmosphere. There is also increasing awareness that changes in the propagation characteristics of planetary waves in the stratosphere (due to natural or anthropogenic factors) may play a role in tropospheric climate variability by influencing such phenomena as the Arctic Oscillation. WACCM will eventually represent the atmosphere from the surface to ~500 km, including interactive chemistry.

Work on WACCM began during FY 2000 and continues at present. The initial WACCM-01 is based on CCM3 that has been extended to the new upper boundary at 140 km by (1) incorporating non-Local Thermodynamic Equilibrium Infrared transfer above 60 km; (2) extending the gravity wave parameterization to include the forcing due to a spectrum of propagating gravity waves; (3) including molecular diffusion and diffusive separation effects above 90 km from the Thermosphere Ionosphere Mesosphere Electrodynamics General Circulation Model (TIME-GCM); and (4) incorporating shortwave heating for wavelengths between 120 and 200 nm from the TIME-GCM. WACCM-01 uses 2 time level semi-Lagrangian dynamics at T63 spectral truncation and 66 levels. WACCM is being updated to be based on the CCSM tropospheric model, CAM2, which already includes most of the required modifications. Current work is focussed on incorporating a version of the chemical scheme from ACD's Model for OZone And Related chemical Tracers (MOZART)-3 photochemistry model, which has been extended to include some thermospheric processes.

A 20 year control simulation has been performed with WACCM-01, forced with the observed ozone distribution. This simulation shows that the model can simulate realistic zonal mean fields, tidal motions, and other aspects of dynamics. Dynamical fields (3 hourly) from the WACCM-01 have been used to drive the extended MOZART-3 offline chemical transport model, which simulates realistic ozone and other constituents. This figure (37 K) shows December, January, and February (DJF) zonal means of zonal winds, temperatures, and gravity wave forcing for 10 years of the WACCM-01 control simulation. The 4th panel is the DJF ozone simulated by the MOZART-3 offline chemical transport model driven by WACCM-01 output.

The simulated zonal winds and temperatures reproduce the observed structure of the troposphere, stratosphere, and lower mesosphere quite well. The zonal wind reversals near 80 km and above are driven by the parameterized gravity drag, as can be clearly seen in the figure. The reversals in the zonal wind structure above 80 km are observed, however, the simulated winds are somewhat too weak. 

The cold summer (and warm winter) mesopause temperatures near 80 km are also observed features driven largely by gravity wave forcing. The observed summer mesopause temperature is actually even colder than the 170 K simulated in WACCM-01. This temperature is quite sensitive to the gravity wave forcing and to the way heat transport is parameterized in breaking gravity waves.

The ozone simulated by MOZART-3, using WACCM wind and temperature fields, reproduces the observed ozone reasonably well, although closer analysis would reveal significant discrepancies, as with the dynamics. The secondary maxima in the mesosphere and lower thermosphere are observed.