An analysis of parameterization interactions and sensitivity of single‐column model simulations to convection schemes in CAM4 and CAM5

Abstract

This paper uses different deep convection triggering functions and closure assumptions in two versions of the Community Atmospheric Model (CAM4 and CAM5) to investigate the interactions of parameterization components and the sensitivity of single‐column model simulations of tropical convection. The schemes include those used in the standard CAM4 and CAM5 as well as two variants used in other global models. Large‐scale forcing and verification data are from the Global Atmospheric Research Program's Atlantic Tropical Experiment and Kwajalein Experiment field campaigns during which many deep and shallow convective events were observed. Results show that in both CAM4 and CAM5, when the intensity of deep convection decreases as a result of parameterization change, the intensity of shallow convection increases, leading to little change in the total simulated precipitation but very different changes in precipitation types. The different precipitation types manifest themselves in other measures of model performances including temperature and humidity. The standard CAM4 and CAM5 both simulated warm bias in the middle troposphere and dry bias in the lower troposphere. In experiments where deep convection is reduced, these biases are also reduced, but the associated changes in shallow convection create new patterns of model biases. Results imply the need to use multiple independent observations simultaneously to constrain models so that the degrees of model freedom are reduced and to treat model physical parameterizations as an integrated system rather than individual components.