AMOC sensitivity to iceberg forcing

The last glacial cycle was characterized by a number of abrupt cooling events in the North Atlantic known as Heinrich events. Heinrich events are associated with massive ice sheet surges from the former Laurentide ice sheet and subsequent predominantly eastward transport of icebergs across the North Atlantic. Paleo records also indicate that ice discharge events coincided with a weakening of the Atlantic Meridional Overturning Circulation (AMOC). Typically, in model studies, ice mass discharge is treated as a direct freshwater injection either at release sites or uniformly distributed within a latitude belt. These experiments are known as freshwater hosing experiments and have been performed using many different Earth System Models (ESMs). Hosing experiments have shown that the AMOC is highly sensitive to the amount and the location of the freshwater injection. However, this approach ignores iceberg physics and thus potentially affects the AMOC response. Therefore, including interactively coupled icebergs in an ESM better represents the iceberg-ocean interaction. Hitherto, all existing iceberg modules are implemented within a Lagrangian framework and require model adjustment to make it possible to run it on millennial time-scales. Here, a new Eulerian iceberg module for long-term climate studies is presented, and the effect of iceberg hosing instead of direct freshwater injection is simulated for different climate states.

Experiments set-up

  • MPI-ESM-CR (ECHAM6-T31L31 and MPI-OM GR30L40)
  • Interactive Eulerian iceberg module
    • 5 iceberg size classes
  • Prescribed ice sheets
  • Prescribed iceberg/ freshwater forcing
    • Increasing/ decreasing at a rate of 0.1 Sv in 1000 years (Fig. 1 (left))
  • Pre-industrial (PI) and Last Glacial Maximum (LGM) experiments with different type of hosing:
    • Iceberg (IB and lgmIB; Fig. 1 (right), star)
    • Freshwater point sourse at the same location as IB and lgmIB (FWPS and lgmFWPS; Fig. 1 (right), star)
    • Freshwater North Atlantic within 50-70°N (FWNA and lgmFWNA; Fig. 1 (right), green area)
    • Control (CTRL and lgmCTRL; no forcing)


Fig. 1 Input rate (Sv; left) and hosing locations (right).


The simulated ocean climate at the LGM and the PI are very different with much colder and more saline conditions at the LGM. The key difference between PI and LGM is the location and the mechanisms involved in deep water formation. In PI, the main North Atlantic deep water is formed in the Nordic Seas where salty originally subtropical water has been cooled sufficiently much to permit deep water formation. In LGM, North Atlantic deep water formation occurs in the Arctic and is driven by a net negative freshwater budget due to strong sea ice export.


Fig. 2 Iceberg meltwater flux (mm yr-1) in PI (left) and LGM (right) experiments at an input rate of 0.04 Sv (year 400).

Fig. 3 Ocean response to hosing in PI (top) and LGM (bottom). 100-years mean of the AMOC (Sv, left) and sea surface temperature (°C, SST) in North Atlantic (right).


  • The Eulerian iceberg module is coupled to MPI-ESM and is applicable for long-term simulations.
  • Iceberg hosing shows a different AMOC sensitivity than freshwater hosing.
  • Iceberg meltwater flux strongly depends on the background climate.
  • The location, and mechanisms involved in deep water formation depend on the background climate that, in turn, determines the efficiency of the system response to hosing.