ENSO dynamics

1
ENSO dynamics predictability in the GFDL
coupled models
Andrew Wittenberg NOAA/GFDL Thanks to G.
Vecchi, A. Capotondi, Q. Song, T. Rosati, T.
Reichler
2
IPCC-AR4 Models
Flux Adjustment H Heat, M Momentum, W Water,
X None
3
Overall Performance Index courtesy Thomas
Reichler Junsu Kim, Univ. of Utah
A combined measure of how well 21 different
global climate models simulate35 different
observed climate features (time averaged, large
scale quantities). Normalized so that the
average model score 1.0 Values less than 1.0
are better.Lower Values Smaller Errors (i.e.,
greater agreement btwn the model simulation
observations)
Avg. of 21models 1.0
4 GFDL CM2.0
1 GFDL CM2.1
4
Overall Performance Index courtesy Thomas
Reichler Junsu Kim, Univ. of Utah
A combined measure of how well 21 different
global climate models simulate35 different
observed climate features (time averaged, large
scale quantities). Normalized so that the
average model score 1.0 Values less than 1.0
are better.Lower Values Smaller Errors (i.e.,
greater agreement btwn the model simulation
observations)
Error bars from bootstrapping observations
EnsembleMeann21
Avg. of 21models 1.0
Obsvs.Obs

4 GFDL CM2.0
1 GFDL CM2.1
5
Planet CM2
Wittenberg et al. (J. Climate, 2006)
6
Regressions onto NINO3 SSTAs
7
Regressions onto NINO3 SSTAs
8
CM2 lag regressions onto NINO3 SSTAs
9
ENSO period vs. y-width/longitude of zonal stress
anomalies
Capotondi, Wittenberg Masina (Ocean Modelling,
2006)
corr(T, Tp) 0.82 /- 0.15
Tp 3.05 (Ly-14)/9.6 (C-184)/30
10
CM2 sensitivity Cumulus Momentum Transport (CMT)
11
CM2 sensitivity to AGCM resolution M45-gtM90
12
CM2.3 sensitivity to parameterized convection
RAS -gt Donner
13
Blocking the Indonesian Throughflow Change in
mean SST (degC)
14
Blocking the Indonesian Throughflow Anomaly
patterns (regressed on NINO3)
15
Approach Hybrid Coupled GCM
Partition the problem
flux climatology L(T') N(T) noise
slow
fast
1. Fit to CGCM runs, verify analogue 2. Mix
match components to isolate what matters most 3.
Can atmosphere-only runs predict CGCM
behavior? 4. Intercomparison with other CGCMs
16
Partitioning the wind stress
stress F(SST) atmospheric noise
stress ltclimatologygt ltL(T')gt ltEgt
E
Many examples of this in the literature... Barnett
et al. 1993 Syu et al 1995 Eckert Latif
1997 Moore Kleeman 1999 Harrison et al.
2002 Wittenberg 2002 Zavala-Garay et al 2003,
2005
17
Partitioning the wind stress
stress F(SST) atmospheric noise
stress ltclimatologygt ltL(T')gt ltEgt
E
Many examples of this in the literature... Barnett
et al. 1993 Syu et al 1995 Eckert Latif
1997 Moore Kleeman 1999 Harrison et al.
2002 Wittenberg 2002 Zavala-Garay et al 2003,
2005
18
SST-forced AGCM(GFDL AM2)
19
Partitioning the wind stress
stress F(SST) atmospheric noise
stress ltclimatologygt ltL(T')gt ltEgt
E
Many examples of this in the literature... Barnett
et al. 1993 Syu et al 1995 Eckert Latif
1997 Moore Kleeman 1999 Harrison et al.
2002 Wittenberg 2002 Zavala-Garay et al 2003,
2005
20
Partitioning the wind stress
stress F(SST) atmospheric noise
stress ltclimatologygt ltL(T')gt ltEgt
E
Many examples of this in the literature... Barnett
et al. 1993 Syu et al 1995 Eckert Latif
1997 Moore Kleeman 1999 Harrison et al.
2002 Wittenberg 2002 Zavala-Garay et al 2003,
2005
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
E
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
Stochastic forcing A role for the Indian Ocean
Daily west-Pacific zonal stress from 10 AM2 runs
Observed SST forcing
Warm East Indian
31
Summary of greenhouse tropics in CM2.x

• 1) SST warms 3-4C loosely El Niño-like
• - Pacific trade winds ITCZs more y-symmetric
• 2) More clouds, evaporation, ocean-dynamical
  cooling
• - compensate for reduced longwave cooling at
  surface
• 3) Sharper/shallower thermocline halocline
• - shallower ocean circulation weaker surface
  winds/currents
• 4) Stronger annual cycles of SST rainfall
• - more ocean-dynamical cooling atm moisture
  convergence
• 5) ENSO amplifies slightly
• - enhanced wind stress coupling heat flux
  damping
• 6) Model biases -gt uncertainty

32
IPCC AR4 intercomparisons

• 1) GFDL CM2.0/2.1 ENSOs among top 4 (of 23
  models)
• - nice spectrum, zonal propagation, coupling
  strengths
• 2) Most models show loosely El Nino-like warming
• - dT/dy change more robust than dT/dx
• - more stratified ocean atmosphere
• - weaker atmospheric circulation trumps upwelling
  thermostat
• - eastward shift of warm pool convection
• 3) Diverse changes in ENSO spectrum/pattern
• - hard to detect in short records
• - increased damping opposes increased coupling
• - "best" models show more amplification, eastward
  propagation

Guilyardi (Climate Dyn. 2005) van Oldenborgh et
al. (Ocean Sci. 2005) Philip van Oldenborgh
(subm. GRL 2006)
Merryfield (JC in press, 2005) Liu et al. (JC
2005) Tanaka et al. (SOLA 2005)
Collins (Clim. Dyn. 2005) Jin et al. (GRL 2001)
33
Annual mean surface changes in CM2.0
34
Annual mean surface changes in CM2.0
35
Annual mean subsurface changes in CM2.0
36
Future mean state resembles El Nino, except

• 1) Evaporation increases more broadly
• - especially off-equator where winds are strong,
  SST is warm
• - reduces further off-equatorial warming of SST
• 2) More PBL moisture, reduced lapse rate
• - increases static stability of atmosphere -gt
  circulation slows
• 3) Ocean warms from above
• - by heat fluxes, not upwelling and not just at
  equator
• 4) Ocean more stratified
• - mixed layer ocean circulation shoal
• - enhanced warming near equator where dT/dz is
  strong
• 5) Surface ocean equilibrated all seasons
  affected

Knutson Manabe (JC 1994, 1995, 1998)
Collins (GRL 2000) Vecchi et al. (Nature
subm.)
37
ENSO changes in CM2.0 CM2.1
38
ENSO rainfall changes in CM2.0 CM2.1
39
Spectral changes in CM2.0 CM2.1
40
ENSO atmospheric response
CM2.0
CM2.1
41
ENSO surface heat fluxes along the equator
CM2.0
CM2.1
42
Changes in intraseasonal variability
43
CM2 sensitivity Cumulus Momentum Transport (CMT)
44
CM2.1 Natural modulation of ENSO
45
Mixed layer temperature anomaly tendency equation
Key to understanding impact of background state
on ENSO.
46
(No Transcript)
47
(No Transcript)
48
Observational Fields

• IPCC-AR4 intercomparison 35 variables
• IPCC-AR4 CMIP intercomparison15 variables
• Problem of missing fields.
• BCCM eliminated.