Climate change systems and simulations
The dramatic Dansgaard-Oeschger events were indeed genuine and also largescale climate changes, not limited only to Greenland. Since then, these events have been further confirmed by data from more than 170 locations around the planet, including New Zealand and the Antarctic, but the cause remained a mystery at first.
One thing was very clear from the deep-sea data: each climate change in Greenland must have been associated with distinct changes in ocean currents.
Michael Sarnthein, a marine geologist from Kiel, identified three different circulation modes from those data: in one, the warm North Atlantic Current (the continuation of the Gulf Stream) continued up to the Scandinavian coast, more or less as is the case today.
In the second, however, it only reached to somewhere south of Iceland. In the third, it had evidently ceased altogether (cf. Fig. 2). In order to understand such climate changes, research teams throughout the world are performing computer simulations of the climate system.
They strive to calculate the most important aspects of climate — ocean currents and winds, air and water temperatures, cloud and ice cover, etc. — for the entire earth with the aid of basic equations of thermodynamics and hydrodynamics, as well as from empirical relations.
Easier to compute climate than to predict the weather
This will never be entirely successful, but at least it is easier for climatologists to compute the climate than for meteorologists to predict the weather: while weather is dominated by chaos or at least by stochastic processes and therefore can only be predicted to a very limited extent, this fortunately does not apply to the average properties of climate. (Mathematically speaking, weather forecasting is an initialvalue problem; with marginally different initial conditions, the weather will develop along totally different lines after a few days. Simulating climate, on the other hand, is a boundary-condition problem, as the earth's energy balance determines the mean climate conditions.)
Despite their limitations and shortcomings, computer models of the climate are already very useful tools for simulating certain situations — such as how continental ice sheets or changes in atmospheric carbon dioxide concentration affect the large-scale temperature distribution and other climate parameters. Experiments can be carried out with the computer climate that would be impossible with the real planet, for instance in order to find out how stable or unstable the climate is in a given period.
Six years ago, our team was able to present the first successful simulation of the climate prevailing at the peak of the last great ice age (around 20,000 years ago), including the ocean circulation. Other international teams followed shortly afterwards, using different models.
Comparison of simulation models
Comparing the result of such a simulation with all the available climate data is an important test of the model's quality. It was found at the time that changes in Atlantic currents in our model played an amplifying role in cooling the northern hemisphere. Since then, we have systematically studied the behaviour of the ocean currents under ice age conditions in numerous further experiments.
On this basis we have developed a theory which might explain the mechanism underlying the abrupt changes in climate. The three circulation modes already described by Sarnthein for the Atlantic currents (Fig. 2) were also found in our computer model. Only one proved stable under ice age conditions, namely the middle mode in which the warm current ceased south of Iceland.
The other two — i.e. the one corresponding to today's Atlantic currents and the one without any warm current — could be initiated by introducing specific perturbations in the model, but the Atlantic automatically reverted to its only stable mode after a few centuries. In a warm climate, such as that prevailing today, the situation is exactly the opposite: according to our model, the two modes which are unstable under ice age conditions are now stable. The stable ice age mode, on the other hand, is not found.
Trigger of unstable circulations
What kind of perturbation is needed to trigger one of the unstable circulation modes? In this context, it is important to know that the ocean current depends strongly on the inflow of freshwater into the North Atlantic, i.e. the total precipitation plus river runoff and meltwater minus evaporation.
The inflow of freshwater determines the salinity of the seawater — and the salinity, in turn, determines the density of the water: the sinking of high-density water drives the ocean currents. To change the current, we need only change the inflow of freshwater.
Since the current also transports salt, this produces a reinforcing feedback effect leading to the unique, non-linear behaviour of the Atlantic Ocean. Model calculations indicate that a very precarious balance prevailed in the Atlantic Ocean during the ice age.
Minor disruptions in the inflow of freshwater (the Achilles heel is located in the Nordic Seas, where the system is particularly susceptible to perturbations) caused the Atlantic to shift temporarily from its stable, cold circulation mode to a different mode more akin to today's climate.