85th AMS Annual Meeting

Wednesday, 12 January 2005: 5:15 PM
Effects of Convection Parameterization Closure on the Simulation of Tropical Climate in the NCAR CCM3 GCM
Guang J. Zhang, SIO/Univ. of California, La Jolla, CA; and M. Mu
A new closure based on the observations has been incorporated into the Zhang-McFarlane convection scheme. Multi-year simulations using the revised scheme in the NCAR CCM3 have been analyzed. This paper presents the results from these simulations. We will show that the modifications to the convection scheme have improved the simulation of the tropical precipitation in several important aspects. The serious negative precipitation bias in the northwestern tropical Pacific monsoon region that appears in the NCAR models has been largely eliminated. The positive precipitation bias in summer in the Arabian Peninsula is also eliminated. The double ITCZ in the tropical Pacific is alleviated considerably. The associated cloud and water vapor fields in the tropical upper troposphere are also improved. Budget analysis in selected regions is conducted to understand how the improved simulation of the tropical precipitation distribution is related to the changes in the convection scheme. For example, in the Arabian Peninsula, when the amount of convective instability in the atmosphere is used for parameterization closure, convection develops in the early summer when the land surface is heated up. High surface evaporation ensues due to high soil moisture resulting from convective precipitation. This leads to a more unstable atmosphere thus more convection. Such a positive feedback is the main cause of the fictitious precipitation in the NCAR model over the Arabian Peninsula. When the new closure is used, this positive feedback is completely eliminated in this region, giving a much realistic simulation of summertime precipitation.

Supplementary URL: