Scenario for abrupt events

Our scenario for the abrupt DO events is therefore as follows: a small disturbance in the freshwater balance of the Nordic Seas suddenly caused warm Atlantic water to flow past Iceland into the Nordic Seas, within a period of a few years.

This warm water caused the sea ice to melt and the temperature of the entire region to rise. The current then gradually weakened over the centuries, until it dropped below a critical point where the warm current ceased again.

Fig. 3 shows the temperature evolution for this scenario, which can explain the three characteristic phases of a DO event. The spatial pattern of the warming and the delayed reaction in Antarctica in our model also correlate well with the actual data.

In search of the trigger

What this theory lacks is a trigger: what caused such disruptions in the Nordic Seas? The Greenland ice core data indicate that a mysterious cycle lasting 1,470 years underlies these events, which was discovered by Gerard Bond and is also found in other climate data.

The interval between successive DO events is very often exactly 1,470 years, sometimes also two or three times that value, as if there were some kind of regular oscillation which triggers a DO event sometimes, but not every time.

Our model calculations show how the instability of Atlantic currents can act as a huge non-linear amplifier, transforming an originally weak cycle into a dramatic and abrupt climate change. The irregular sequence of DO events can be reproduced well in the model if it is assumed to be triggered by a weak 1,470-year oscillation in combination with random fluctuations (e.g. weather variability).

1.470 year cycle

In that case, the climate changes are triggered by a phenomenon which physicists call "stochastic resonance". The only problem is that there is no known cycle of such duration which could act as a trigger. But perhaps we are seeing a superposition of cycles: the two well-known cycles of solar activity, the Gleissberg cycle (period: 87 years) and the De-Vries cycle (period: 210 years), just happen to have a period of 1,470 years as their lowest common multiple.

In our climate model, DO events can indeed be triggered at this interval by combining these two cycles. Further research is needed to substantiate or refute these still speculative theories. DO events are not the only abrupt climate changes found in recent climate history.

So-called Heinrich events (Fig. 4) occurred at irregular intervals of several thousand years during the last ice age. These events can be seen in deepsea sediments from the North Atlantic, where each such event left an up to one metre thick layer of small stones instead of the usual soft sediment.

These stones are too heavy to have been transported by the wind or ocean currents — they can only have dropped to the sea bed from melting icebergs. Great armadas of icebergs must have drifted across the Atlantic at certain times. These are assumed to have broken off the North American continental ice sheet and entered the ocean through Hudson Strait.

They were probably caused by an instability in the ice sheet, which was several thousand metres thick at the time. Snowfall caused it to grow continually until the slopes became unstable and slipped — rather like a mound of sand from which the sand slides down in avalanches as more sand is piled on top. Sediment data indicate that the formation of North Atlantic Deep Water (NADW) temporarily ceased completely as a result of the Heinrich events.

This is shown by the upper circulation mode in Fig. 2. Climate data reveal an associated, abrupt drop in temperature, particularly in the midlatitudes, e.g. the Mediterranean region. Greenland was affected to a lesser extent, probably because in glacial times the warm current did not reach far enough to the north to warm the climate at higher latitudes (except during DO events).

Climate in present interglacial period more stable than climate of the last ice age

One important question is why the climate in our present interglacial period (the Holocene) is evidently more stable than the climate of the last ice age. There have not been any DO events or Heinrich events during the Holocene, i.e. for more than 10,000 years. One final, but fairly weak, phase of abrupt cooling occurred 8,200 years ago (sometimes referred to as the 8k event — Fig. 1).

Data and simulations indicate that it resulted from the last inflow of meltwater at the end of the ice age. When the dam of ice holding back the huge meltwater lake known as Lake Agassiz broke, the freshwater poured into the Atlantic and temporarily disturbed the warm North Atlantic Current. Many researchers believe that it was the relatively stable climate of the Holocene that prompted man to start farming and settle some 10,000 years ago.

The reason why there have been no Heinrich events during the Holocene is self-evident: they can only occur during an ice age because they are due to instabilities in the continental ice sheets. The answer is more complex in the case of DO events.

If our theory of DO events as outlined above is correct, it would be the different ocean circulation mode prevailing in the Atlantic that makes the Holocene climate so stable. This circulation mode is not right near a threshold like the circulation mode during the ice age, and it cannot be disrupted by minor disturbances.

This also applies in the computerised climate model: the disturbances with which we triggered DO events under ice age conditions have no effect on the model climate under the conditions of the Holocene.