Geographical contributions to global climate sensitivity in a General Circulation Model

Abstract

This paper addresses two questions: how do radiative contributions to global climate model feedbacks vary geographically, and hence which regions and physical processes are most important in determining final climate sensitivity in a General Circulation Model (GCM)? Offline radiation calculations were used to evaluate in detail the strength and spatial distribution of top of atmosphere (TOA) radiative perturbations for the BMRC GCM under a doubling of CO 2. The long wave and short wave radiative perturbations were considered separately. The net global effect of these radiative perturbations determines the strength of the global model feedbacks, and hence the climate sensitivity. The geographical distribution of the radiative perturbations on the other hand helps to identify the model processes that are most important for determining the strength of model feedbacks. This study found that globally, the dominant positive feedbacks were for (a) water vapour amount and height, and cloud height for long wave radiation and (b) albedo and cloud amount for short wave radiation. The dominant negative feedback (apart from the surface temperature term itself) occurred for cloud amount changes in the long wave. Geographically, the contributions to the water vapour feedback strength varied markedly by location, with subtropical, and upper tropospheric regions contributing relatively strongly to the net global feedback. The water vapour height component correlated quite strongly with convective changes while the amount term was correlated with fractional precipitable water changes. The contribution of different precipitable water regimes in the tropics to global water vapour feedback was assessed—relatively dry areas contributed disproportionately, but did not dominate the overall feedback because of their small areas. Contributions to cloud amount and height feedbacks in the long wave were found to be uncorrelated in space, indicating their different controlling processes. The long wave amount component correlated strongly with upper cloud, and convective changes. The short wave component of the cloud feedback depended on cloud changes at all levels, producing its strongest contribution to feedbacks in mid latitudes, where the sign of cloud changes agreed at different heights. The effect of cloud cover on non-cloud feedbacks was also investigated. This study found that clouds weaken all feedbacks, except that of lapse rate, with the greatest impact being on the surface albedo and water vapour amount. The radiative perturbation analysis method presented here proves to be a powerful tool for identifying important physical processes that determine final climate model sensitivity.