COLA Report 27

The Effect of the Cumulus Convection on the Climate of the Cola General Circulation Model

David G. Dewit


May 1996

Abstract

The effect of the cumulus convection parameterization on the simulated climate of the Center for Ocean-Land-Atmosphere Studies (COLA) atmospheric general circulation model (GCM) is evaluated. One of the major goals of this study is to evaluate the possible improvement to the coupled model system by a change in convection scheme. Three different convection schemes are tested in the model: Ku , Bets-Miller (BM) and relaxed Arakawa-Schubert (RAS). The results from the model integrations with the different convection schemes are compared with observational data to evaluate the schemes. Dynamical fields such as winds and temperature are compared with the European Center for Medium Range Weather Forecasts analysis. Precipitation is evaluated using a data set that is derived by merging Microwave Sounding Unit (MSU) precipitation data over the ocean with station data from the Climate Anomaly Monitoring system over land. The near equatorial wind stress is evaluated by direct comparison with the subjectively analyzed Florida State University wind stress product. In another study, the wind stress from the different integrations was used to force the Geophysical Fluid Dynamics Laboratory ocean model to evaluate the errors in the repective wind stress fields.

The standard COLA GCM uses the Kuo convection scheme. This study shows that some subtle yet significant improvements may be expected by the change to the RAS convection scheme. Ocean model simulations using the wind stress from the integration done with RAS were shown previously to produce a better annual cycle and interannual variability of the sea surface temperature and heat content along the equator than obtained when the wind stress from the Kuo or Betts Miller schemes were used. Results in this paper show that there is a strong correlation between the east-west precipitation gradient and the zonal wind stress along the equator. The improved wind stress with improved RAS is seen to be primarily due to the modest amount of precipitation along the equator east of the dateline in agreement with the observations.

The difference in the boreal winter mid-latitude dynamical response to the warm and the cold phases of ENSO is shown to be best simulated using the RAS scheme. In addition, the boreal winter climatological stationary waves, especially in the PNA region, are best simulated with the RAS scheme. The simulation done with RAS shows considerable reduction in the model root mean square error for the upper level geopotential and winds during boreal winter. There is also a significant reduction in the geopotential error for the summer; however there is a modest increase in the boreal summer wind error.

The mean precipitation for the different seasons in the integrations done with RAS are shown to better approximate the precipitation gradients in the tropics than when the Kuo scheme is used. The precipitation and circulation associated with the Indian summer monsoon are also shown to be improved when the RAS scheme is used. Little difference in the ability to reproduce Madden-Julian like transient variability in the tropics is found among the convection schemes. The results with the BM scheme are, in general, not seen to be a significant improvement over those with the Kuo scheme.

This report serve to document the latest version of the COLA model as the RAS convection scheme has been adopted in teh standard model.

di dnitoring
last update: 1 July 1996
comments to: www@grads.iges.org