1、AMOC MULTI-DECADAL VARIABILITY: MECHANISMS, THEIR ROBUSTNESS, AND IMPACTS OF MODEL CONFIGURATIONS,Gokhan Danabasoglu and Steve Yeager National Center for Atmospheric Research,OUTLINE,Many Coupled General Circulation Models (CGCMs) exhibit (multi-)decadal variability in their AMOCs.,Brief review of s
2、ome proposed mechanisms and their robustness- primarily CGCM studies - few idealized studies- not comprehensive Some results from the Community Climate System Model (CCSM4) to show impacts of model configuration and parameterizations on AMOC variability,CONSENSUS: Density anomalies in the sinking re
3、gions of AMOC drive these AMOC fluctuations.,T85x1,T42x1,T31x3,AMOCmax in CCSM3,Proposed AMOC Variability Mechanisms,Delworth et al. (1993, DMS93): GFDL R15 coupled model, 40-80 yr period,weak AMOC,reduced heat transport,cold, dense pool in middle North Atlantic,T anomalies generate cyclonic gyre ci
4、rculation,anomalous circulation transports S into the sinking region,S, density, and AMOC all increase,AMOCmax vs density regressions,DMS93 and Griffies & Tziperman (1995): Damped ocean-only mode excited by atmospheric noise,Weaver & Valcke (1998): coupled mode,Delworth & Greatbatch (2000): Damped o
5、cean-only mode, continuously excited by low-frequency atmospheric forcing, implications for NAO,Dai et al. (2005): Same DMS93 mechanism in PCM (25 yr),Dong & Sutton (2005): Same DMS93 mechanism in HadCM3 (25 yr),Both Dai et al. (2005) and Dong & Sutton (2005) suggest stronger ties to the NAO.,Freshw
6、ater Transport to / from Arctic Ocean and Nordic Seas,Delworth et al. (1997), GFDL R15, 40-80 yr: Enhanced transport of relatively fresh water and sea ice from the Arctic via the East Greenland Current and Denmark Strait. These anomalies propagate around the subpolar gyre into the Labrador Sea, capp
7、ing the convection. . Greenland Sea oscillations are implicated, but how they are generated is unknown.,Jungclaus et al. (2005), ECHAM5/MPI-OM, 70-80 yr: Storage and release of freshwater from the central Arctic to the Labrador Sea convection site along with circulation changes in the Nordic Seas du
8、e to Atlantic heat and salt transports. Damped ocean mode excited by atmosphere as in DMS93.,Freshwater Transport to / from Arctic Ocean and Nordic Seas,Oka et al. (2006), MIROC, 30-50 yr: Freshwater transport through the Denmark Strait results in deep convection see-saw between the Labrador Sea and
9、 Greenland Sea. Wind Stress / NAO changes are implicated and the variability is interpreted as a coupled mode.,Winter-time mixed layer depth (m),March-mean boundary layer depth (BLD) EOF1,AMOCmax vs density regressions,AMOC lagging,AMOC leading,-10,-5,0,5,10,Danabasoglu (2008), CCSM3 T85x1 present-d
10、ay control, 20 yr:,m,DOrgeville & Peltier (2009), CCSM3 T31x3 pre-industrial control, 60 yr: Similar in-phase T and S contributions to density, less role for NAO, but suggest gyre bathymetry interaction,Role for NAO,Same Model and Integration, but 2 Different Mechanisms (HadCM3, 90 yr),Vellinga & Wu
11、 (2004): Involves large scale air-sea interaction,AMOC +,increased NHT,generates cross Equatorial SST gradient,northward ITCZ shift,increased rainfall and FW flux into the ocean,surface salinity decreases,low salinities advected north into sinking regions,AMOC -,Same Model and Integration, but 2 Dif
12、ferent Mechanisms (HadCM3, 90 yr),Hawkins & Sutton (2007): Internal ocean modeChanges in the Nordic Seas convection lead to AMOC changes. Variations in salinity transports from the Arctic and from the North Atlantic are the main controlling feedbacks. Similar to Delworth et al. (1997) and Jungclaus
13、et al. (2005) but convection region, and hence the mechanism, are different.,Role for Southern Ocean,Saenko et al. (2003): AMOC / NADW is affected by the Southern Ocean freshwater perturbations.,Delworth & Zeng (2008), GFDL CM2.1: Strength and position of the Southern Hemisphere mid-latitude westerl
14、y winds impact AMOC strength.,Biastoch et el. (2008): Dynamic signals originating in the Agulhas leakage region have influence on decadal AMOC variability.,Park & Latif (2008), Kiel Climate Model: Multi-centennial and multi-decadal variabilities are both associated with sea-ice extent and the former
15、 is driven in the Southern Ocean. Coupled ocean atmosphere sea-ice mode.,Some Other Mechanism,Marshall et al. (2001): Intergyre gyre concept with links to NAO,Msadek & Frankignoul (2009), IPSL-CM4, 100 yr: Convection is primarily influenced by the anomalous advection of salinity due to changes in th
16、e East Atlantic Pattern coupled mode.,Zhu & Jungclaus (2008), coarse resolution ECHAM5/MPI-OM, 30 and 60 yr: They are interpreted as ocean-only and coupled modes, respectively. The 30-year variability is related to T anomalies moving along the cyclonic subpolar gyre and leading to fluctuations in ho
17、rizontal density gradients and subsequent weakening and strengthening of AMOC. . Consistent with TeRaa & Dijkstra (2002, 2003).,Timmermann et al. (1998), Farneti & Vallis (2009), Cheng et al. (2004), ,AMOC in CCSM4,CCSM4 includes many physical and computational improvements in all its components com
18、pared to CCSM3. The ocean model uses spatially and temporally varying eddy mixing coefficients, submesoscale mixing parameterization, lower horizontal viscosities, An overflow parameterization is used to represent the Nordic Sea overflows. The ocean model resolution is nominal 1o in the horizontal w
19、ith 60 vertical levels.,No Overflows (CCSM3),With Overflows (CCSM4),Sv,Sv,Strong salinity restoring reduces model error in the subpolar seas, but itweakens AMOCsignificantly damps AMOC variability north of 30oNreduces max Atlantic northward heat transport to below 1 PW,Increased North Atlantic bias,
20、AMOC Maximum Transports at 26.5oN in Ocean Ice Hindcast Simulations with CORE Forcing,1 month,1 year,4 years,no restoring,RAPID,Impacts of surface salinity restoring,AMOC maximum north of 28oN,AMOC maximum at 26.5oN,AMOC Properties in Ocean Ice Hindcast Simulations,AMOC Maximum Transports in CCSM4 P
21、re-Industrial Control Simulations,99%,95%,120,67,CCSM4_1: 1o FV atmosphere,CCSM4_2: 2o FV atmosphere,Impacts of Parameterized Nordic Sea Overflows on AMOC Variability,Preliminary CCSM4 present-day simulations with 2o atmosphere and 1o ocean resolution,AMOC maximum transport,99%,95%,Period (years),Pe
22、riod (years),90,50,200+,70,Density and section-normal velocity at 45oW,color: density (kg m-3)line: velocity (cm s-1),SUMMARY WHAT HAVE WE LEARNED IN THE LAST 5 YEARS? (CGCM view),AMOC variability and predictability are (perhaps) more complicated than originally thought.Proposed variability mechanis
23、ms are not (really) robust across different models.Unresolved processes, e.g., mesoscale eddies, Nordic Sea overflows, oceanic mixing, appear to influence AMOC significantly.AMOC variability in CCSM4 is muted compared to that of CCSM3 and preliminary results indicate influence of overflows and a dif
24、ferent mechanism than in CCSM3.,Key observational priorities listed in the AMOC 2009 report will be certainly helpful in discriminating against some of the proposed mechanisms.,0,-15,-10,-5,+5,max AMOC,“max” SST,max density BLD,min AMOC,positive NAO,negative NAO,strong subpolar gyre,reduced sea-ice
25、cover,increased surface heat loss,increased upwelling of salt,increased sea-ice cover,reduced surface heat loss,reduced upwelling of salt,diffusive fluxes,Simplified Diagram of Phase Relationships in CCSM3,DOrgeville & Peltier (2009), CCSM3 T31x3 pre-industrial control, 60 yr: Similar in-phase T and S contributions to density, less role for NAO, but suggest gyre bathymetry interaction,
