Impact of interannual atmospheric forcing on the Mediterranean Outflow Water variability in the Atlantic Ocean  Hello, today I’m going to present my work on the impact of interannual atmospheric forcing on the Mediterranean Outflow Water variability in the Atlantic Ocean. Alexandra Bozec1, Eric Chassignet1, Susan Lozier2 In collaboration with George Halliwell3, and Zulema Garraffo4 1 Center for Ocean-Atmospheric Prediction Studies, Tallahassee, FL 2 Division of Earth and Ocean Sciences, Duke University, Duke, NC 3 Atlantic Oceanographic and Meteorological Laboratory, Miami, FL 4 Rosenstiel School of Marine and Atmospheric Science, Miami, FL

Background Trends in water properties of the Mediterranean Outflow Water reservoir (Potter and Lozier, 2004): Observations: Max of S of each profiles Time period : 1955 to 1993 3 year moving average Results: TEMPERATURE S-Trend (psu/decade) T-Trend (ºC/decade) 0.0283+/- 0.0067 0.101+/- 0.024 SALINITY So why studying the Mediterranean Outflow Water variability? In 2004, a study by Potter and Lozier shows that a salinity trend of 0.03psu/dec and a temperature trend of 0.1ºC/dec of the main core the Mediterranean water in this yellow box here that they call the MOW reservoir between 1955 and 1993. Especially, the temperature trend far exceeds the the average warming of the North Atlantic described by Levitus (2000). PRESSURE DENSITY 1955 1995

Source of Variability for the MOW Reservoir Mediterranean Sea Water (MSW) North Atlantic Central Water (NACW) Change in MOW properties through either MSW and/or NACW properties Change in the Atlantic circulation So what would be the sources of these large trends? Here is a schematic of the water mass involved in the production of MOW: you have the Mediterranean Sea Water product of the transformation of surface Atlantic water through its journey in the Mediterranean Sea by the air-sea interactions. Then the MSW descends along the northern edge of the Gulf of Cadiz mixing with the North Atlantic Central Water and stabilizes around 1100m in the Atlantic with a salinity of 36.2 and a temperature of 11ºC. So the source of the MOW variability in the reservoir can come from etiher the MSW and /or the NACW or it can also be induced a change of circulation in the Atlantic Ocean that would extend or not the MOW tongue. Mediterranean Outflow Water (MOW)

Mediterranean Sea Water variability Mediterranean Sea Water from E-P Lozier and Sindlinger (2009) Derived MSW salinity using different E-P products (NCEP/NCAR, ECMWF, DaSilva) NCEP/NCAR Med Sea Water: S-Trend: +0.037 ± 0.018 psu/decade Resulting MOW reservoir salinity (assuming NACW constant in Price and Yang (1998)): S-Trend:+0.0024 ± 0.0014psu/decade => 10 times lower than observations NCEP/NCAR SO first let’s take a look at the variability of the MSW: A recent study by Lozier and Sindlinger derived the MSW salinity at Gibraltar from different E-P product (here on the right) and shows that only the NCEP/NCAR salinity has a trend susceptible to explain the MOW salinity trend in the Atlantic. However, putting this salinity in the Price and Yang box model and assuming a NACW salinity constant through time, they deduced a salinity trned of the MOW in the Reservoir of 0.003psu/dec which is roughly 10 times lower than the observed trend. ECMWF DaSilva

NACW entrained variability From HydroBase 2 To test if there assumption of a constant NACW salinity was ok or not, I calculate the evolution of the NACW properties in the Gulf of Cadiz between 400 and 700m from HYDROBASE data and found no trend between 1955 and 1993. NACW trends between 1955-1993: Trend S at 600m: +0.0025±0.0090 psu/dec. Trend T at 600m: +0.0069±0.0029 ºC/dec.

The variability of the Mediterranean Sea water and of the North Atlantic Central water are too weak to be responsible for the variability of the MOW in the Atlantic. Hypothesis: The variability is due to the interannual variability of the atmosphere (wind and/or buoyancy forcing). The variability of the MSW and of the NACW are then too weak to be responsible for the variability of the MOW in the Atlantic. So our hypothesis is that the variability of the MOW is due to the interannual variability of the atmosphere through either a mechanical effect of the wind and/or the impact of the buoyancy forcing.

Experimental Set-up 3 simulations using 1/3º Atlantic HYCOM with the Marginal Sea Boundary Condition (Price and Yang, 1998). Spin-up of 30 years 60 years simulations Constant properties of the MSBC MOW: T=11ºC, S=36.2psu,Transport =4Sv. Experiments Atmospheric Forcing CLIM Climatological ECMWF (ERA15) WIND Interannual wind-stress (NCEP 1948-2006) and climatological ECMWF buoyancy forcing BUOY+WIND Interannual buoyancy and wind forcing (NCEP 1948-2006) To test this hypothesis, we set up 3 simulations using a 1/3º Atlantic HYCOM using the Marginal Sea boundary condition to simulate the outflow. All simulations starts from a 30-years spin-up and last for 60 additional years. As we saw from observations that the variability of the MOW at the exit of Cadiz was very weak, we set the outflow properties constant at 11ºC, 36.2psu and 4 Sv. The 3 simulations are CLIM with a climatological forcing, WIND that uses an interannual wind-stress from NCEP from 1948 to 2006 but a climatological forcing and finally BUOY+WIND that uses a interannual wind-stress and buoyancy forcing.

MOW Reservoir Trends in HYCOM CLIM WIND BUOY+WIND Experiments Salinity Trend (psu/decade) Temperature (C/decade) Density (kg/m3/decade) Observations 0.028±0.0067 0.101±0.0024 0.00 CLIM 0.010±0.0012 0.042±0.0043 WIND 0.019±0.0011 0.086±0.0065 BUOY+WIND 0.032±0.0024 0.127±0.0128 0.01 As the modeled outflow spreads more than in the observations, we defined a new box (here the grey box) to calculate the properties evolution of the tongue and checked that we should have the trend fro these box than in the Potter and Lozier box. So here are the results with salinity on top, temperature and density., and you can see that BUOY+WIND shows the most important variability compared with the others experiment and if we calculate the trend for 1955-1993 we find that this is the simulation the closest to the observations.

Spreading of the MOW CLIM WIND BUOY+WIND HYDROBASE We compare the MOW tongue (s2=36.52) during 1955-1965 and 1980-1990 for each simulation: If we now look at the spatial pattern of the tongue between 1955-1965 and 1980-1990 for each simulation, we see that only BUOY+WIND shows an extension of the tongue in the latter period compared with the other, extension that we can see this smooth salinity field from HYDROBASE. CLIM WIND BUOY+WIND HYDROBASE

Why is the MOW tongue expanding? So why is the MOW tongue expanding ? If we look at vertical salinity section at 36N for each simulation with the isopycnals overlaid for the two periods 1955-1965 and 1980-1990 we see that in BUOY-WIND the interface depth of the MOW density layers are not as tilted as the other simulation letting the MOW extends westward. WIND shows even more tilted isopycnals in the latter period showing the variability of the MOW is not due primarily to the mechanical effect of the wind. The cross section at 36ºN shows tilted isopycnals in BUOY+WIND compared with the other simulations.

Time Evolution of Interface Depths MOW Salinity (psu) Interface depths (m) If we look more closely at the time evolution of the interface depths in the central Atlantic, we notice that salinity increase coincides with a decrease of the thickness of the Labrador Sea water layer which is a little bit confusing since 1980-1990 is a period of High NAO and we would think lots of convection and so lots of LSW. MOW salinity increase coincides with a retraction of the Labrador Sea Water (LSW) from the Central Atlantic.

LSW thickness (s2=36.83) NAO- NAO-/+ NAO+ Salinity contour (white): However, looking at the evolution by 5-year bin of the LSW thickness in BUOY-WIND, we see a strong variability during the 60 years simulations. I also put in white 3 salinity contours of the MOW of this layer (so at the base of the salinity tongue). So we can see that at the beginning of the simulations, the LSW layer is quite thick and the MOW is restrained to the east of the basin, between 1955 and 1979, we are in a low/neutral NOA phase where no LSW is form, the thickness of the layer then decrease progressively letting the MOW water penetrates the ventral Atlantic around 1975 . Then the LSW starts to refill during 1980-1999 and starts to expands back and we start to see in the 2000’s the MOW starting to retract from the central Atlantic. Salinity contour (white): 35.15psu, 35.2psu and 35.3psu

Trend reversal in the 2000s From Leadbetter et al., 2007 And this is actually in agreement with the observations since Leadbetter et al. shows from a repeated section at 36N a reversal of the trend in 2005. Here we calculate the same profile for BUOY+WIND and see a increase of the salinity between 1959 and 1981 and a slight decrease in 2005 that is actually accentuated in 2006. I would like also to mention that we also did a interannual simulation using variable outflow properties from the Price and Yang that presents a slightly less important transport of the MOW at the exit of Cadiz, that shows even better agreement with the obs. From Leadbetter et al., 2007

Conclusions We are able to reproduce the MOW reservoir variability in the Atlantic for a constant MOW production. The observed MOW reservoir variability is due to circulation changes in the Atlantic Ocean induced by the atmospheric forcing. These circulation changes are primarily due to the variability in buoyancy forcing through the formation and flushing of Labrador Sea Water during high and low NAO periods. We can identify a 20-year cycle in phase with the NAO for the period 1950-2005. (i.e. time needed to fill and empty the “LSW reservoir”). So to conclude: