Henk Dijkstra (IMAU)

1
Physics of multidecadal variability in the North
Atlantic
in a hierarchy of models
Lianke te Raa Institute for Marine and
Atmospheric Research Utrecht (IMAU), Netherlands

• Henk Dijkstra (IMAU)
• Jeroen Gerrits (OSU)
• Maurice Schmeits (KNMI)

with
2
Atlantic Multidecadal Oscillation
Annual mean SST averaged over 0oN-60oNx75oW-7.5
oW
AMO index (Sutton en Hodson, 2005)
SST WARM (1950-1964) SST COLD (1970-1984)
What is the physical cause of the AMO?
0.8
-0.6
Kushnir (1994), from Delworth et al (1993)
3
Approch model hierarchy
1) Idealized model
-North Atlantic only -constant salinity -idealized
geometry -steady heat flux forcing
2) More realistic configurations
-continental geometry -bottom topography -salinity
-wind stress
3) Coupled climate model
-global -coupled climate model
4
Idealized model (MOM 3.1)
Configuration
Meridional overturning (Sv)
-North Atlantic only -constant salinity -idealized
geometry -no wind
P 50 yr
5
1) Idealized model (MOM)
t5887.5 yr
t5880 yr
Surface temperature anomaly (oC)
t5895 yr
t5902.5 yr
6
Linear stability analysis (THCM)
Computation underlying steady state
Steady state meridional overturning (Sv)
7
Spatial pattern multidecadal eigenmode
(THCM)
Surface temperature
Multidecadal oscillation results from linear
eigenmode
8
Physical mechanism
Thermal wind balance
t0
Te Raa and Dijkstra (2002)
9
2) Add continental geometry
(MOM 3.1)
P45 yr
Surface temperature anomalies
t3810 yr
t3800 yr
latitude
latitude
longitude
longitude
Mechanistic indicator
?Z
?M
Zonal overturning
averaged overturning (Sv)
Meridional overturning
Te Raa et al (2004)
time (years)
10
3) Global climate model GFDL R30
Data courtesy T. Delworth
Oceanic resolution 1.875o x 2.25o x 18
Run 1000 yr Atmospheric resolution 3.75o x
2.25o x 14 Data Annual mean values of last
500 yr
THC index (Sv)
Delworth et al, Clim Dyn. (2002)
11
MSSA mode, P44 yr
t600 yr
t606 yr
Anomalies potential density (kg/m3) and velocity
at 20 m depth
t618 yr
t612 yr
12
MSSA mode, P44 yr
t606 yr
t600 yr
Anomalies potential density (kg/m3) and velocity
at 683 m depth
t618 yr
t612 yr
Dijkstra et al (2006)
13
Comparison with observations
Ocean GCM (MOM)
GFDL R 30
Observations
P44 yr
P45 yr
SST (1950-1964) - SST (1970-1984)
SST (warm) - SST (cold)
SST (warm) - SST (cold)
Kushnir (1994)
longitude
Dijkstra et al (2006)
14
Conclusions

• Multidecadal variability in an idealized model
  of the North Atlantic is caused by an internal
  oceanic mode of variability

• The physical mechanism of this multidecadal
  variability is a lagged response between zonal
  and meridional flow perturbations, due to
  westward propagating temperature (density)
  anomalies

• The signatures of the variability can be
  followed towards more complex model configurations

• A dominant multidecadal statistical mode of
  variability exists in the GFDL R30 climate model

15
Our results seem to support the hypothesis that
The timescale of the AMO is caused by the basin
crossing time, and its pattern is that of a
deformed multidecadal mode