Modeling convection in the Greenland Sea

Abstract

A detailed examination of the development of a deep convection event observed in the Greenland Sea in 1988-89 is carried out through a combination of modeling, scale estimates, and data analysis. We develop a prognostic one-dimensional mixed layer model which is coupled to a thermodynamic ice model. Our model contains a representation of the lowest order boundary layer dynamics and adjustable coupling strengths between the mixed layer, ice, and atmosphere. We find that the model evolution is not very sensitive to the strength of the coupling between the ice and the mixed layer sufficiently far away from the limits of zero and infinite coupling; we interpret this result in physical terms. Further, we derive an analytical expression which provides a scale estimate of the rate of salinification of the mixed layer during the ice-covered preconditioning period as a function of the rate of ice advection. We also derive an estimate for the rate of the mixed layer deepening which includes ice effects. Based on these scale estimates and model simulations, we confirm that brine rejection and advection of ice out of the convection area were essential ingredients during the preconditioning process. We also demonstrate that an observed rise in the air temperature starting in late December 1988 followed by a period of moderately cold ≈-10°C temperatures was key to the development of the observed convection event. Finally, we show that haline driven deep convection underneath an ice cover is possible, but unlikely to occur in the Greenland Sea. On the basis of these results, we develop a coherent picture of the evolution of the convection process which is more detailed than that presented in any previous work. We also comment on the likelihood that deep convection occurred in the Greenland Sea in the past two decades from an examination of historical data, and relate these findings to what is known about the inter-annual variability of convective activity in the Greenland Sea.