The Program for Climate Model Diagnosis and Intercomparison (PCMDI) was established in 1989 at the Lawrence Livermore National Laboratory (LLNL) in the San Francisco Bay area. Our staff includes research scientists, computer scientists, and support personnel. There are several means of communicating with PCMDI.
The need for standards in both modeling and analysis has become increasingly apparent as more complex models are developed, while the disagreement among models, and between models and observations, remains significant and poorly understood. The nature and causes of these disagreements must be accounted for in a systematic fashion before models can be confidently used for climate sensitivity and predictability studies in support of global change research.
As a preliminary contribution to model diagnosis and intercomparison under standardized conditions, PCMDI is coordinating the Atmospheric Model Intercomparison Project (AMIP) on behalf of the Working Group on Numerical Experimentation (WGNE) of the World Climate Research Programme (WCRP).
In this project, some 30 AMIP modeling groups are simulating the climate of the decade 1979-1988 using the observed monthly sea-surface temperature and sea ice, as well as common values of solar luminosity (1365 W/m^2) and atmospheric carbon dioxide concentration (345 ppm). The participating models' principal numerical algorithms and dynamical/physical parameterizations are summarized in AMIP model documentation.
A set of monthly-mean standard output data from each AMIP model is archived at PCMDI, and some AMIP modeling groups have also provided history data at six-hourly intervals. Analysis of the models' performance in comparison with observational data reveals their systematic errors, while comparison of the AMIP simulations with each other provides a measure of current modeling uncertainties.
More focussed analyses of the models' ability to simulate specific processes and phenomena are being carried out by more than two dozen AMIP diagnostic subprojects in which PCMDI is playing a supporting role. The results of this diagnosis and intercomparison of the AMIP model simulations were extensively discussed at an international AMIP Scientific Conference in May of 1995.
PCMDI scientists are conducting AMIP studies, both independently and in association with AMIP diagnostic subprojects. We are testing regional and hemispheric validation techniques spanning synoptic to interannual time scales in an effort to evaluate the fidelity with which models can simulate observations.
PCMDI has extensively studied the effects of spatial and temporal resolution on climate simulations performed with the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric model and the Semtner-Chervin ocean model. In addition, we are examining the natural variability of simulated seasonal and interannual climate in studies of ensemble integrations of the ECMWF model from different initial conditions.
PCMDI has conducted several exploratory studies of the use of nonlinear and common principal components for model/data and model/model intercomparisons. We also are developing statistical techniques for detection of climate change (by external forcing such as greenhouse warming) against the background of internal variability of the atmosphere and ocean.
PCMDI scientists are collaborating with a number of AMIP diagnostic subprojects. Some of this work is coordinated with other WCRP/WGNE initiatives, such as the Monsoon Numerical Experimentation Group (MONEG) and the GEWEX-sponsored Project for Intercomparison of Land-Surface Parameterization Schemes (PILPS).
PCMDI also is supporting the Feedback Analysis for GCM Intercomparison and Observations (FANGIO) project through the development of prototype diagnostics to identify important radiative feedbacks. In addition, we are collaborating with the Paleoclimatic Model Intercomparison Project (PMIP), for which PCMDI is providing data storage and diagnostic software.
PCMDI computer scientists also have worked with colleagues at the National Energy Research Supercomputer Center (NERSC) and in the OPTIMASS Project at the Lawrence Berkeley National Laboratory (LBNL) to develop innovative data access, translation, and storage algorithms and software.
A comprehensive climate database management system (CDMS) is currently under development at PCMDI. This tool will provide efficient access of information from both our model and observational data archives.
During the next few years, PCMDI will continue its efforts to develop a climate modeling infrastructure (consisting of standards and software for documentation, data management, diagnosis, visualization, validation, intercomparison, and exchange) in cooperation with the climate community. Much of this effort will be directed to supporting the next phase of the AMIP (designated as AMIP II).
In the longer term, we will progressively shift our emphasis from atmospheric models to models of the coupled ocean-atmosphere system, and will support the development of a Coupled Model Intercomparison Project (CMIP) in cooperation with the CLIVAR initiative of the World Climate Research Programme.
PCMDI also will address specific questions that are important model-related elements of global change research. These include consideration of how to most effectively control the unrealistic climate drift that is typical of coupled systems, and how simulated natural variability may be used in the study of climate predictability and the detection of climate change on global and regional scales.
This research, supported by a climate modeling infrastructure, will provide the basis for the systematic application and improvement of models and for the comprehensive evaluation of their ability to characterize future climate changes as a result of anthropogenic activities.
Last update November 20, 1996. For questions, contact
Jerry Potter (potter@oliver.llnl.gov)
UCRL-MI-119847