Weather Prediction and Climate Simulation: a Meeting of the Models
Xie, S., Lawrence Livermore National Laboratory
General Circulation and Single Column Models/Parameterizations
Cloud Modeling
Phillips, T. J. G.L. Potter, D.L. Williamson, R.T. Cederwall, J.S. Boyle, M. Fiorino, J.J. Hnilo, J.G. Olson, S. Xie, J.J. Yio, Evaluating Parameterizations in General Circulation Models: Climate Simulation Meets Weather Prediction, Bulletin of the American Meteorological Society, accepted May 2004.
Xie. S., M. Zhang, J.S. Boyle, R. T. Cederwall, G.L. Potter, and W. Lin, (2004), Impact of a revised convective triggering mechanism on CAM2 Model Simulations: Results from Short-Range Weather Forecasts, J. Geophys Res., 109, D14102, doi: 10.1029/2004JD004692.
Distribution of a 20-day mean precipitation forecast throughout the continental United States shows much better agreement from the original climate model (CAM2O) to the adjusted CAM2M model.
Developed through a joint venture between the DOE's Climate Change Prediction Program (CCCP) and Atmospheric Radiation Measurement (ARM) Program, the CCCP-ARM Parameterization Testbed, or CAPT, is a diagnostic tool for evaluating general circulation models (GCMs) using weather prediction techniques. Results of research reported in two separate journals describe how CAPT is contributing to model improvements for both weather and climate. An article accepted by the Bulletin of the American Meteorological Society describes a method for using numerical weather prediction methods within the CAPT framework for improving parameterizations in general circulation models. In the Journal of Geophysical Research, the performance of both weather and climate models within the CAPT framework is evaluated against observations from ground-based instruments at ARM's Southern Great Plains site, observations of rainfall and cloud cover from satellites, and analyses of temperature, wind and humidity fields from the European Center for Medium Range Weather Forecasting.
Predicting the weather relies on knowing the current atmospheric state (temperature, winds and humidity) and then using an atmospheric model to extrapolate that state forward in time. This can be done accurately out to about a week. Simulating climate, on the other hand, requires running atmospheric models for much longer timescales forced only by boundary conditions at the top of atmosphere. Climate models are intended to simulate the statistics of weather, rather than the actual weather on any specific day. So while weather models can be evaluated daily on the basis of how well they predict the actual weather, climate models can only be evaluated on how well they predict climate statistics over periods of years.
Most short term (weather scale) forecast errors of clouds and precipitation are due to errors in the model treatment of these quantities (called parameterizations), rather than to errors in the model forecast of atmospheric fields of temperature and humidity. The purpose of the CAPT is to use a climate model as a weather forecasting model and therefore be able to evaluate it over shorter timescales against observations of real weather.
As discussed in the articles, ARM researchers used the CAPT framework as an effective means to identify parameterization deficiencies in the Community Atmosphere Model, or CAM2O, of the National Center for Atmospheric Research. In these simulations, the model produced too frequent convective precipitation (rain) during the day in summertime; much more so than actually occurred, based on measurements from the ARM site. Because the sun strongly heats the earth's surface during summertime, convection (overturning of the atmosphere by near-surface heating) occurs in the atmosphere nearly every day over land in the summer. The problem in the model was that the convection produced rain every day, which doesn't occur in the real world. The researchers introduced a modified convective initialization—or "triggering"—scheme that produced fewer, but more intense, rain events. The resulting model (CAM2M) showed a significant reduction in convective events and much better agreement with ARM and satellite observations of rainfall.
The CAPT, based at the Program for Climate Model Diagnosis and Intercomparison at the Lawrence Livermore National Laboratory, is a useful framework for improving model parameterizations and reducing systematic errors in short-range forecasting by linking model deficiencies directly with atmospheric process. This results in more realistic simulations of important atmospheric fields, such a temperature, moisture, clouds, radiation, surface temperature, and surface sensible and latent heat fluxes.