Mimicking Mother Nature
	To better understand what causes these long-term trends, scientists use oceanic and atmospheric computer models. These models are complex mathematical simulations of the natural systems that drive the Earth's climate. Different models focus on different aspects of climate change. Some simulate the effects of carbon dioxide on the atmosphere, others attempt to emulate airflow over land, and still others mimic the interaction between the sea and the air. While such models can be used in a number of applications, they are especially handy for testing hypotheses and obtaining a better understanding of climate phenomena. They allow scientists to test their climate theories and predictions relatively easily in a controlled fashion without all the unexpected variables mother nature can throw them. As an example, suppose a model was built that simulated the effects of water pollution on global climate and rainfall. Not long afterwards, a group of researchers comes up with the hypothesis that sewage run-off in New York City affects rainfall in China, which in turn causes fluctuations in tea prices. The team could test their theory on the model by simply feeding in sewage data from New York into the model and observing the resulting rainfall patterns produced by the computer. If their hypothesis and the model were correct, then the fluctuations in the sewage would simply correlate with the rainfall and the prices of tea in China. They could then go out and collect real-world data to verify their results. If there is no correlation, the model could also help them discern where their theory breaks down. Though the atmospheric and oceanographic models NAO researchers use can be much more complex than the one in this fictitious illustration, the manner in which they test their theories is the same. Their goal is to use the models to link the long-term decadal trends in the NAO to ocean currents, ice flows, and whatever else may be influencing the climate anomaly. As of now, they are still in the initial stages of the process. The scientists are merely attempting to establish that a basic connection exists between the NAO and the ocean.
	For several years Mehta and his team used an atmospheric model to study this connection. They made a big breakthrough last year by establishing a crucial link between sea surface temperature and the decadal variations in the NAO. "If the NAO does interact with the sea, then the interaction would be primarily with the temperature at the surface," said Mehta. The model they used essentially takes in sea surface temperatures over the planet for a given year and then calculates a number of variables including temperature, pressure, and rainfall. With this information, Mehta says it isn’t too difficult to figure out the average monthly and yearly pressure readings for the two Atlantic winter pressure systems or the amount of rainfall over Europe and the United States. So far the Goddard team has published one set of results. As an input to the model, the researchers compiled records of monthly sea surface temperatures taken across the globe from the years 1949 to 1996. Weather stations as well as Navy vessels and commercial ships made these readings on their voyages. Mehta explained that the dates were chosen because both maritime traffic and efforts to record sea surface temperatures increased dramatically after World War II. On their own without any further modifications, this initial set of readings would at best return undependable values on air pressure and rainfall after being fed into the computer. "This is due to a manifestation of chaos. A small difference in initial conditions of the readings can make a big difference in the final results that emerge from the model," he said. To eliminate this chaos and obtain accurate values for the NAO, the Goddard team fed the sea surface temperature data into the model sixteen different times. Each time they put in these data, they used a different set of initial atmospheric conditions–those atmospheric values that the computer sees as the starting point for the model run (Mehta et al., 1999). After averaging each of the 48 years of pressure, temperature, and other readings that the sixteen computer model runs produced, the scientists found that the computer readout matched up nicely to historical records. From these results, Mehta then calculated the yearly averages of the NAO and the corresponding changes in rainfall and temperature. "On the whole our model was a success. We were able to simulate the NAO itself and it's impact on Europe and the Northeastern U.S.," says Mehta. "The only years that did not correspond to actual readings were from 1949 to 1959. We have yet to explain the discrepancy here." Mehta warns that his analysis cannot make predictions, but can only show the difference in rainfall and pressure in and around the Atlantic for the years when sea surface temperatures of the ocean are given. The real significance in these results is that they demonstrate that the low-frequency, long-term changes in the NAO can be attributed primarily to sea surface temperatures. "We would only be able to forecast the phenomenon if we first could predict specific global sea surface temperatures," said Mehta. But even now other teams in both Europe and the United States are working with other ocean models to link currents, sea ice, and even the NAO itself to sea surface temperatures. Their hope is that if they nail down all the variables influencing the NAO as well as it’s reciprocal effects on the ocean, they may be able to predict the sign the anomaly will take each year. Then they may even be able to determine what is causing the multidecadal global warming trend. "However, this will probably not happen right away. It will be some time before these models are built, and the interactions involved here are very complex," says Mehta. References Mehta, V.M., M.J. Suarez, J. Manganello, and T.L. Delworth, 2000: Oceanic influence on the North Atlantic Oscillation and associated Northern hemisphere climate variations. Geophysical Research Letters, 27, 121-124. National Oceanic and Atmospheric Administration, 1999: North Atlantic Oscillation, Silver Spring, Maryland. World Climate Research Program, 1997: CLIVAR International Implementation Plan, Southampton, United Kingdom, D1. Links North Atlantic Oscillation at the Lamont-Doherty Earth Observatory  Relying on the Ocean’s Long Term Memory		The above maps show the relationship between the North Atlantic Oscillation Index and sea level pressure. The observed data (top) show that when the NAO index is high, the atmospheric pressure over Greenland and Iceland is lower than normal, and the pressure west of the Iberian Peninsula (Spain and Portugal) is higher than normal. Model results based on global sea surface temperatures (bottom) show a similar pattern. (Images by Vikram Mehta, NASA Goddard Space Flight Center)

	Subscribe to the Earth Observatory About the Earth Observatory Please send comments or questions to: eo-contact@eodomo.gsfc.nasa.gov Program Manager: David Herring Responsible NASA official: Yoram Kaufman NASA/GSFC Security and Privacy Statement