Until now only a few rare studies have been carried out on the impact that the NAO could have on the Gulf Stream.  However, their results were contradictory.   “One group of scientists stated that the NAO index is negative (when the dipole is less marked), and so the ocean current would be weaker and vice versa.  A second group said exactly the opposite!  Finally, a third group estimated that a positive NAO would imply a Gulf Stream further north and vice versa”  

In order to get a clearer picture, Sylvain Watelet undertook a more precise study.  Unlike his predecessors, who concentrated on very localised data and on quite short periods of time, the PhD student opted instead to analyse a much wider area of the Gulf Stream (between 50 and 70 degrees west of longitude and between 35 and 47 degrees north of latitude, in other words more or less the whole “horizontal” part of the current) and over a longer period, from 1900 to 2012.

“That’s where the Diva software comes in, he explains.  I reconstruct the ranges of temperature (of the surface for now) on the Gulf Stream area since 1900, thanks to DIVA (Data Interpolating Variational Analysis).  Then, with these ranges, I calculate the location of the Gulf Stream (that is where the north-south temperature gradient is highest).  Finally, from the power of the gradient, I draw an intensity index of the Gulf Stream.” The first step was to gather all the statistics, by searching through data bases available on the web, mainly those that had measurements of surface water temperature.  In fact, the researcher from Liege analysed temperature and salinity for 15 different depths (from 0 to 3000 m), but he has not yet made use of these results.  The data comes from fixed and floating buoys, cruises, etc.  Not as easy a task as it may seem, between the duplications, inaccuracies, misleading rounding, deliberate falsifications from expeditions (military, for instance) when people did not wish their exact itinerary to be retraced. Finally, the PhD student created a homogeneous data base with no fewer than 40 million entries.  

Correlation, well yes…

Enough to constitute two indices: one showing the location of the Gulf Stream and the other its intensity.  Sylvain Watelet then compared the results with satellite data, only available since 1982.  “Correspondence is available covering these past thirty years and this validated the construction of my indices” This helped him to reach the last phase, that is to say, comparing with the NAO, for the same year but also with 1 and 2 year time lags.    

“The correlations between the location of the Gulf Stream and the North Atlantic Oscillation have never been significant, except between 1940 and 2012 with a one year time lag. 
 Since 1982 the correlation between the location of the Gulf Stream and the North Atlantic Oscillation has not (exactly) been significant (it is too short a period), but on doing the calculations over a longer period (1940-2012), it becomes significant.  That indeed shows that there is a link between the NAO and the Gulf Stream.  Between 1900 and 1940, the reduction in the correlation is logical: the amount of data has decreased dreadfully (much weaker than between 1940 and 2012) and our analyses overall are not more reliable.  If we weight the correlation depending on the amount of data (the less there is, the less we’ll take these years into consideration in the calculation), it then becomes significant.” Regarding the intensity, we had to concentrate on the years 1960-2012 and not on 1940-2012 (as we did for the location), which is probably linked to the fact that the right evaluation of the intensity of the Gulf Stream requires a greater mass of data that the evaluation of its location. In this context, a significant correlation appeared between 1960 and 2012, in the same year or with a two year time lag. “Why this time lag of 0 or 2 years whereas in the case of the location, the time lag is one year? For the moment we do not have an answer to this question”, regrets Sylvain Watelet.