Water mass transformation rates as a climate model evaluation methodology

Water mass formation is a result of complex processes such as air-sea buoyancy fluxes, winter convection, heat and freshwater transports, thus closely related to the ocean overturning and circulation. Using the water mass transformation function introduced by Speer and Tziperman (1992) we can evaluate the amount of water formed at the sea surface in particular classes of density. In North Atlantic, there are two main mode waters: the sub-tropical mode water (STMW, σ_θ=26), also known as 18deg water, forming south of the Gulf Stream in the north-western corner of the sub-tropical gyre, and the sub-polar mode water (SPMW, σ_θ=26.9-27.75) forming along the periphery of the sub-polar gyre and northern part of the sub-tropical gyre (Hanawa & Talley, 2001). Being an integrated measure of air-sea interaction and deep-water formation, the water mass transformation function can serve as a climate model evaluation metric for both validation of the air-sea fluxes and the water mass formation rates (WMTR). It requires input for sea surface density and heat and freshwater fluxes. The in-situ based estimates such as NOC1.1a are limited by spatial and temporal coverage. Recently, more observation-based data sets have become available including satellite and reanalysis products. Here we are going to compare estimates for North Atlantic WMTR derived from: 1) in-situ data such as Levitus climatology for SSS and SST and NOC1.1a fluxes; 2) satellite observations utilizing the Aquarius satellite product for density and HOAPS satellite based heat and freshwater fluxes; 3) re-analyses e.g. CORE2 (based on NCEP reanalysis) and OAFlux. Using these observational constrains we will evaluate the air-sea interaction and water mass formation in the Norwegian Earth System Model. Analysis of various types of numerical experiments e.g. AMIP (atmosphere only), OMIP (ocean only) and CMIP (fully coupled) conducted via utilizing the different observational estimates will help to identify the origin of the model biases (atmosphere, ocean processes and/or numerics) and assess the importance of different climate feedbacks. The overall goal of our study is to build a methodology for using the WMTR metrics in climate model evaluation.