It performs an invaluable service for our climate: The ocean can absorb and store huge amounts of heat. At the same time, ocean currents act like giant conveyor belts, transporting warm water from the Equator toward the poles; in this way, just like the atmosphere, they ensure that heat is redistributed around the globe. For example the North Atlantic Current system, which also includes the Gulf Stream, moves warm water from the Gulf of Mexico to Europe’s North Sea and into the Arctic – maintaining a mild climate in Northern Europe.
As such, the Gulf Stream is part of a giant current referred to as the Atlantic Meridional Overturning Circulation, which is above all driven by temperature and salinity differences in the water, and by the wind: As it flows to the north, surface water becomes increasingly colder and heavier, causing it to sink. When ice forms, the salt is left behind in the water, making it even heavier; then the cold, high-saline water flows back to the south, far below the ocean surface.
At the KlimaCampus, my colleagues and I are analyzing to what extent the variability of this circulation, and of the resultant heat transport, can be predicted: With the help of a numerical model, we can determine when there are fluctuations in the Overturning Circulation, making it stronger or weaker. In this context, we are especially interested in the variations from year to year within ten-year model simulations, which places our study between short-term weather forecasts on the one hand, and long-term climate prognoses for the next 100 years on the other. For long-term predictions, framework conditions like the rising CO2 concentration in the atmosphere are essential, whereas the initial conditions at the starting point of the simulation are crucial for short-term forecasts. Our middle-term simulations have to satisfy both criteria.
In order to estimate how well simulations can predict variations in the circulation, we need a reference value – ideally in the form of observational data. So we’re not actually making predictions in the litaral sense (forecasts). Instead, we test our simulations against reference data from the past (“hindcasts”) to determine whether or not they can provide accurate estimates. Since taking measurements in the ocean is an expensive and laborious undertaking, in many areas we still have only fragmentary data. Therefore we use an additional numerical modeling, which combines observational data with numerical simulations to realistically represent the ocean, as the reference for our hindcasts. A further advantage: We can assess the accuracy for all latitudes of the North Atlantic, as the analysis is not limited by a lack of observational data.
And we don’t compare the reference to just one, but all in all to roughly 200 hindcasts, each of which shows one possible development of the ocean. The more closely they correlate with the reference, the more accurate they are. Our analysis has shown that we can essentially make accurate prognoses for the next two to five years; this varies depending on the latitudes we are researching. We can make the longest-term predictions between the subtropical and subpolar gyres at 40 degrees North – the line of latitude that is home to New York and Madrid.
News from climate science: All Articles at a glance
Guest articles since 2010 have been published as KlimaCampus booklets
Yearly cycle of Atlantic currents: Climate Visualization Laboratory