Author/Authors :
Backhaus، نويسنده , , Jan O and Kنmpf، نويسنده , , Jochen، نويسنده ,
Abstract :
A high-resolution, rotational non-hydrostatic coupled ice–ocean convection model, defined in a vertical ocean slice, was applied to simulations of oceanic convection and its interaction with an ice cover. The model experiments exemplify typical mixed layer situations observed in the stratified upper Greenland Sea. Results are discussed in relation to observed hydrography, and ice conditions. The focus is placed on the initial penetrative phase of convection that covers small (sub-meso) spatial and temporal scales. Short episodes of strong atmospheric forcing (outbreaks of cold polar air) are considered. Simulations of convective erosion of both a shallow and a deep cold, low-saline surface `freshwater’ layer of the Greenland Sea highlight the thermal feedback on ice growth and surface buoyancy flux caused by an entrainment of warmer underlying waters. The entrainment, and an erosion of the stratification well below the actual penetration depth of convection, is caused by penetrative convective plumes. Plumes transmit fluctuations of non-hydrostatic pressure across isopycnals. The model predicted a transition from haline to predominantly thermal convection whenever ice formation was involved. After a cold-air outbreak the freshwater layer is largely re-established, but has increased salinity, and temperatures well above the freezing point. A thin ice cover (<10 cm) with a coverage between 60 and 90% remains at the sea surface, thus impeding a deep penetration of convection for subsequent cold air outbreaks. With a moderate cooling ice formation and melting are in balance. The net buoyancy flux of the ice–ocean system attains zero values, and convection is shut off. Ice cover and ocean approach a state of stagnation which has often been observed. An attempt was made to hindcast an observed event of localised deep convection in the absence of sea ice. Forced by an oceanic heat loss of about 1000 W/m2, which lasted for 140 h, the simulation reproduced the observed mixed-layer deepening from an initial 500 to 1200 m. These results and observations obtained with moored instruments suggest that deep reaching, penetrative convection in the Greenland Sea is the result of thermal convection, favoured by an ice-free ocean surface, rather than of haline convection. The increased frequency of sightings of a locally confined deep penetration of convection, in conjunction with the model hindcast, provide support for a hypothesised pre-conditioning of deep convection due to the mesoscale eigendynamics of the freshwater layer.