Inventory of EU Marine Climate Change Research

New search

CONVECTION - Greenland Sea Convection Mechanisms and their Climatic Implications

Summary information

Funding:FP5 - Research project
Total cost:3649465
Ec contribution:2500000
Start date:2000-11-01
End date:2003-11-01
Duration:36 months
Coordinator:Peter Wadhams (
Organisation:Polar Research Institute, University of Cambridge - United Kingdom
Themes:Sea level rise; ice melting; ocean current changes; deep circulation changes
Project name:CONVECTION - Greenland Sea Convection Mechanisms and their Climatic Implications
Project summary:Abstract
CONVECTION aims to assess open ocean deep-water production in the Greenland Sea by a combination of operational remote sensing, modelling, and field requirements. We seek to understand the physics underlying convection and how this process links with global climatic factors. The field measurements comprise work in two winters and three summer cruises. Winter convection will be mapped by an acoustic shadowgraph array, moored yo-yo CTDs and transects by an AUV, while ice production and movement will be mapped by in site measurements and buoy deployments, with support from ice tank experiments. Passive microwave, SAR, wind scatterometer and airborne data will be combined with in situ data to feed a model that calculates salt flux distribution over the region. When combined with a large-scale ice-ocean model and a small-scale convection model the final package will explain the convection process and its variability under extremes of forcing.
Project outputs:Scientific Achievements
Previous studies and research programmes in the central Greenland Sea led to a consensus about how and why the Greenland Sea is such a favourable location for deep convection. This was based on a combination of hydrographic properties and ice formation. The upward doming of the deep waters in the centre of the gyre, which brings the deep waters close to the surface and makes them more easily accessible to surface forcing, has been the main hydrographic feature of the region. The local formation and advection of ice, with net brine release and the establishment of descending plumes, is then believed to lead to deep convective events. Salt flux modelling, backed up by experimental studies of pancake ice formation and brine drainage, supports the concept of ice formation as an important source of negative buoyancy in winter. The field studies planned and executed within the project reflected this basic view by the inclusion of an acoustic experiment for the detection of active plumes and an AUV survey underneath the ice cover. Mesoscale non-hydrostatic modelling studies demonstrate the mechanism by which such plumes can be produced and shows how they could develop. The experimental results from the project, however, have caused this general picture to be revised through two discoveries.

Firstly, the doming in the Greenland Gyre has been superseded in recent years by a pronounced and persistent two-layer structure with a density-salinity step at a depth which has steadily increased to 1.600-1.800 m. The upper part seems to be completely ventilated each winter, regardless of ice formation or its lack. This leads to the expectation that there might be convection processes different from plume convection, and within the project another main ventilation type has indeed been identified which leads to different, often contrasting effects. A warming of the ventilated layer by winter convection, for instance, was not thought possible before the project, but the new field data show that this can result from a mixed layer-like ventilation mechanism. This ventilation type, which is independent of ice formation, dominated from the late 1990s onward. It is not confined to shallow depth levels but also proceeds to the medium depth density step which permanently limits deeper convection. Consequently, there is no single main driving mechanism for deep convection, and ice formation is not a necessary prerequisite for it.

The second discovery was of a deep convective chimney near the gyre centre at 75°N 0°W, extending to a depth of 2.500 m and thus penetrating through the density step described above. The chimney was discovered in March 2001 and was subsequently revisited and remapped by successive CONVECTION cruises through the summer of 2001, winter and summer of 2002, and spring of 2003, with persuasive experimental evidence that the same chimney persisted, making this the longest-lived such feature yet observed in the ocean. A second chimney was discovered during the spring 2003 survey, which was comprehensive enough to demonstrate that two chimneys are likely to be the total quantity of such features currently existing, in contrast to a larger number which may have existed in the late 1990s as suggested by the motion patterns of neutrally buoyant floats. The 75/0 chimney had a diameter of 10-20 km; was observed to be in anticyclonic rotation at a rate of f/2 out to a radius of 10 km then at a slower rate; became capped in summer by a fresh 50 m surface layer and an intrusion of Atlantic water down to a depth of 500 m leaving a deeper core untouched; and opened up again to the surface each winter. The chimney was also remarkably immobile, moving only a few km between measurements, although in spring 2003 it began a faster movement to the NW. These remarkable discoveries still remain to be embedded in a full synthesis of the convection process. Until we can carry out further experimental observation and theoretical modelling of chimney structure we cannot be sure whether they are playing an active role in the overall convection process, whether they provide a means for deep water formation, or indeed how they are created and maintain such longevity. They come in as a new and unexpected factor in the overall picture of the Greenland Sea convection process.

A question addressed by the project is whether it is possible to identify conditions which are especially supportive of deep convection. Possible local candidates would be strong winds, cold winters, ice formation, or a low inflow of Atlantic Water (AW). Investigations with 1-d model (which is small enough to be run repeatedly with different forcings and initial conditions) revealed that the vertical density structure of the upper water column is the most discriminating factor. The heat content of the AW represents no hindrance, ice formation is helpful but not essential, and moderate heat losses in winter can suffice for deep convection. The vertical density structure of the upper layers is determined not solely by the lateral fresh water input (low saline Polar Waters in the upper few tens of metres) but also by the convection history. This stems from the fact that plume convection usually results in an overall stability increase because of the varying final depths reached by the individual plumes. After winters with plume convection it is usually markedly more difficult to ventilate the affected layer again. A mixed layer-like convection type has the opposite effect by mixing efficiently through large depth intervals. Thereafter, very low stabilities in, say, the upper 1.300 to 1.800 m facilitate a reventilation during the next winter. A switch back to plume convection, which needs a fresher surface layer, can be caused by a fresh water (Polar Water, ice melt) input. This, again, is naturally sensitive to meteorological forcing.

The larger scale picture of how convection fits into, and is affected by, the larger scale pattern of oceanic and atmospheric circulation in the Greenland Sea – Arctic Ocean system, was addressed by the AWI modelling group. They were assisted by the enormous mass of historical oceanographic and sea ice data collected as another aspect of the project, which is included in this report on a CD-ROM. This provided data to test hindcasts of ice extent, ocean structure and convection back through the century. Remote sensing data, interpreted using innovative algorithms for young ice types and for wave energies, gave new insight into the behaviour and movement of ice within the whole experimental region.

The project has taken our understanding of Greenland Sea convection far beyond the level attained at the end of the previous EU-supported project in this field, ESOP-2. The discovery and mapping of longlived chimneys, the investigation of how two modes of convection may prevail in different years, the innovative research in acoustic mapping of plumes, in AUV mapping and tank studies of ice, and in large-scale and mesoscale modelling, have enriched our understanding of local and large scale processes. The synthesis of these discoveries into a new way of thinking about the Greenland Sea is now in fertile progress, and is already insistently leading to a need for new studies. This is a critical region for the control of the Atlantic Thermohaline Circulation and hence the climate of NW Europe, and this project has revealed a new richness and complexity about the processes which go on here.

Socio-economic relevance and policy implications
Much publicity has become attached to the hypothesis that the shut-off of convection in the Greenland Sea will lead to a decline in the vigour of the Gulf Stream and Atlantic Thermohaline Circulation (THC) and a consequent cooling to the climate of NW Europe, estimated in models to take over from global warming as an absolute declining trend by 2100. From the discoveries in CONVECTION it appears that convection in the Greenland Sea is not shutting off, but is taking place in a variety of forms which offer some resistance to a warming trend. This has major implications for modifying our view of how European (and global) climate is set to change over the next few decades, suggesting that the expected shut-off of the THC and cooling of European climate may well not occur and that the models predicting them are too simplistic.

The improved understanding of the physics of convection means that we can return to the models of carbon expert and air-sea CO2 fluxes developed during the earlier EU ESOP project and derive improved estimates of the role of the central Greenland Sea gyre in CO2 sequestration (moderating global warming) and in the carbon budget of the Nordic Seas, of vital importance for fisheries.

The success of the AWI large-scale ice-ocean model of the Greenland Sea – Arctic Ocean system in hindcasting sea ice extent over the past century implies that improved forecasting of sea ice extent in the Nordic Seas over the forthcoming years is now possible. This will allow improved planning for fisheries expansion and for the growth of new trade and oil/or exporting routes from stretches of the Greenland and Svalbard coasts currently inaccessible through sea ice.