| Up till now, the evolution of the Meridional Overturning Circulation (MOC) in climate models has hardly been validated in a systematic way because the state of the MOC is still poorly known from observations. This proposal suggests a new strategy to validate MOC/climate interactions and to assess possible biases therein. Recently, a new theory on the control of the sinking branch of the MOC was developed. With the use of specific scaling relations and a budget analysis, it is now possible to assess whether models give a physically correct description of the descending branch of the MOC. Because boundary currents appear essential in this process, it is expected that ocean resolution will be crucial for modeling the sinking of deep water. In low resolution models part of the sinking will be driven by numerical friction/diffusion. We propose to evaluate three different model configurations to elucidate the role of ocean resolution and air/sea interaction. The main objective is to arrive at quantitative estimates of MOC/climate feedbacks by calculating lead/lag relations between the MOC and oceanic and atmospheric fields, in particular the heat and freshwater flux through the ocean/atmosphere interface, and the wind field. The physical feedback is then the (lagged) response in sea surface temperature and sea-ice coverage to variations in sinking and upwelling, and the direct response of sinking to atmospheric changes. The response of the ocean's surface to sinking variations is passed to the atmosphere. The theoretical analysis will serve as an interpretative framework to explain the model differences. Also, the model results will be validated with hydrographic data and reanalysis data from the ECMWF, where the MOC will be approximated by its signature in sea surface temperature. This comparison enables us to quantify the model bias in MOC/climate feedbacks and the role of specific physical processes (boundary physics in the ocean, entrainment in overflows, air/sea interaction) in constituting these biases in coupled and uncoupled models. |
