comparison of eastern tropical pacific simulations by ipcc ar4 coupled models simon de szoeke...

Comparison of eastern tropical Pacific simulations by IPCC AR4

coupled models

Simon de SzoekeNOAA/ESRL/PSD3, Boulder, Colorado

Shang-Ping XieInternational Paci c Research Center

Honolulu, Hawaii

Tropical eastern Pacic coupled processes

•equatorial & coastal upwelling; fronts and eddies

•stratus & trade-wind cumulus clouds

•intertropical convergence zones (ITCZ)

•variability: El Niño, TIWs, tropical depressions

•observations: satellite & TAO buoy monitoring; EPIC (2001) & VOCALS eld projects

Robert Wood

SST (° C)

EQEQ

Tropical eastern Pacic

•End of the line for the Pacic equatorial waveguide and embarkation point for the PNA atmospheric wave train

•The PNA connects ENSO to North American weather in winter.•Seasonal cycle is important.•Models fail to predict the season of ENSO variability (Saji et al., 2006).

Warm

PNAwinterpressurepattern (Horel and Wallace, 1981)

Ocean-atmosphere variability: ENSO & Pacific North American (PNA) teleconnection

convection

Outline

1. Assess current state-of-the-art coupled models: IPCC AR4 20th century simulations.•Signicant variety and biases exist among

models!•Introduce metrics.

2. Beyond biases—infer physical relationships from the ensemble of CGCMs.

3. Regional coupled modeling.

Why the eastern Pacic warm pool

is in the north

wind offshore &upwelling

Upwelling along a tilted coast (Philander 1996)Wind-evaporation-SST feedback (Xie and

Philander, 1994)

C

W

Precipitation (mm day-1)

Intergovernmental Panel on Climate Change 4th

Assessment Report (IPCC AR4) models

Canadian Centre for Climate Modelling and Analysis CGCM3.1Centre National de Recherches Météorologiques (France) CNRM CM3CSIRO Atmospheric Research (Australia) CSIRO Mk3.0Geophysical Fluid Dynamics Laboratory (USA) GFDL CM2.0Geophysical Fluid Dynamics Laboratory GFDL CM2.1Hadley Centre for Climate Prediction and Research (UK) UKMO

HadCM3Institute of Atmospheric Physics (China) FGOALS 1.0gInstitute of Numerical Mathematic (Russia)INM CM3.0Institut Pierre Simon Laplace (France) IPSL CM4Center for Climate System Research (Japan) MIROC3.2

medresCenter for Climate System Research MIROC3.2 hiresMeteorological Research Institute (Japan) MRI CGCM2.3.2National Center for Atmospheric Research (USA) NCAR CCSM3.0National Center for Atmospheric Research NCAR PCM 1International Pacific Research Center (USA/Japan) IROAM

15 models from 8 nations

Eastern Pacic average SST (Reynolds)

and rain (TRMM)

degrees west

degre

es

long

itu

de

° C

mmday

Eastern Pacific average SST (Reynolds)

and rain (TRMM)

degrees west

degre

es

long

itu

de

° C

mmday

TRMM

Reynolds

lati

tud

e

meridional seasonal cycle of SST and rain

140-90° W

model a

model b

seasonal SST and rain

lati

tude

8

e 49bf

normalized standard deviation σ/σobs

0.99

0 1 20

0.5

1

1.5

2 correlation

0.20.4

0.6 0.7

0.8

0.90.95

11

2

23

34

5

5

66

7

7

8

9

a

a

b

c

c

d

d

e

f

Eastern tropical Pacific 140-90° W,±10° N SST, ±25° N precipitation

123456789abcdef

cccma cgcm3.1 cnrm cm3 csiro mk3.0 gfdl cm2.0 gfdl cm2.1 iap fgoals1.0g inmcm3.0 ipsl cm4 miroc3.2 hires miroc3.2 medresmri cgcm2.3.2a ncar ccsm3.0 ncar pcm1 ukmo hadcm3 iroam

0.380.740.720.470.620.700.920.510.420.380.260.741.290.480.30

0.490.950.620.841.010.711.180.720.880.640.480.911.530.661.10

RMSETaylor (2001) diagram

Semiannual equatorial SST bias

lati

tud

e

observed

Boreal spring cold bias correlated to northerly wind at r=-0.6.

Westward and disconnected equatorial cold tongue bias

warmcoldwest longitude

lati

tude

SST (° C)

Zonal wind small east of 90˚ W.

meridional wind’s role inequatorial upwelling

Meridional wind drives a meridional cell in the equatorial ocean with upwelling on the windward side of the equator. cools equator

2° N

18° C

v

2° S EQ

Niño 1+2 meridional wind cools SST

2 422

24

26

28

30

1

2

3 45

6

7

89

a

b

c

d

e f

v (m s-1)

SST (

° C

)

r=–0.60

Nino 1+2:80-90° W0-10°S

meridional wind’s role in equatorial upwelling near the

coast

• Models with stronger southerly wind maintain cooler SST 80-90° W (r=-0.6).

• Bjerknes (1969) feedback amplies and propagates cooling westward (Xie, 1998).

EQ

Central American December-February gap winds and SST

Gap winds prematurely end northern warm pool and ITCZ.

0 525

26

27

28

29

1

23

45

6

7

89ab

cd

e

f

0

r=-0.55

DJF CA wind speed (m s-1)

FMA

SST (

˚ C

)

0 2 4 6 8

-4

-2

0

2

4

12345

6

7

89a

b

cd

ef

0

r=-0.87

FMA

vEQ (

m s

-1)

DJF CA wind speed (ms-1)

1 cccma cgcm3.12 cnrm cm33 csiro mk3.04 gfdl cm2.05 gfdl cm2.16 iap fgoals1.0g7 inmcm3.08 ipsl cm49 miroc3.2 hiresa miroc3.2 medresb mri cgcm2.3.2ac ncar ccsm3.0d ncar pcm1e ukmo hadcm3f iroam0 Observed

Gap winds cool northern SSTand lead equatorial northerlies

(5-20˚N, 80-110˚W)

(±2˚N)

Eastern Pacific IPCC AR4 20th centurymodel roundup

• Seasonal cycle of meridional asymmetry exhibits a wide range of behavior.

• Models form an ensemble with signicant inter-model correlations between– SST and rain meridional asymmetry (r=0.8)– meridional asymmetry and equatorial v (r=0.6)– equatorial v speed and cold tongue SST. (r=–

0.6)

• Double ITCZ bias is now mostly an alternating ITCZ bias.→semiannual cycle of wind and equatorial SST

affects interannual variability

Eastern Pacific IPCC AR4 20th centurymodel variations

1. The seasonal cycle of meridional asymmetry governs equatorial wind and SST.

2. Meridional wind is important for equatorial upwelling 0-10° offshore.

3. Central American gap northeasterlies in boreal winter cool the eastern Pacific warm pool, which sends the ITCZ south.

End of part 1

International Pacific Research Center (IPRC)

Regional Ocean Atmosphere Model (IROAM)

• Realistic boundary conditions constrain the model & isolate local coupled feedbacks.

• Better resolution of orographic and mesoscale circulations

• Verify with point observations.Sensitivity experiments• Assess parameterizations• Propose strategies to diminish GCM biases.• Regional effects of scenarios: e.g. last

glacial maximum (cool North Atlantic 2˚C at 21,000 ybp)

1990–1995 1996–2003

Ocean spin-up

Coupled

Land surface model

Ocean forced by NCEP reanalysis

IPRC Regional Ocean-Atmosphere Model (IROAM) Atmosphere forced

by observed SST

Interactive

IROAM is run on Earth Simulator in Yokohama, Japan.

Atmosphere: IRAM, Y. Wang (2003, 2004)Ocean: MOM2, Pacanowski & Griffies (1999)

Surface solar cloud radiative forcing (SSCRF)

IROAM ISCCP satellite retrieval

W m-2

Stratiform clouds reinforce meridional asymmetry

C

W

Sccool

Shallow cumulus convection scheme• Tiedtke (1989) mass flux scheme for deep,

shallow, and midlevel convection• Shallow convection is activated when

– surface evaporation > 90% of the boundary layer (PBL) moisture sourceor

– cloud layer is less than 200 hPa deep.

no shallow convection (noSC) experiment

• All convection is prohibited when either shallow convection condition is met.– Conditional instability in the lower troposphere

must be relieved by turbulence or deep convection.

Shallow cumulus convection affects stratiform clouds

• Shallow convection vents moisture from the boundary layer and entrains dry air in.

700 hPa

Shallow cumulus convection affects stratiform clouds

• Shallow convection vents moisture from the boundary layer and entrains dry air in.

• Reduced venting increases stratiform clouds,which block sunlight from reaching the surface.

noSC

Vertical structure:stratus clouds at 15° S,

September

pre

ssu

re (

hPa)

noSCIROAM control

cloud water, 10-4 kg kg-1

surface solar cloud radiative forcing(W m-2)

-50

-50

-50

-50-100

-50

IROAM control noSC

Stratiform clouds dim sunlight by 50-100 W m-2.

Seasonal cycle (80-140° W)

mm day-1° C

lati

tude

Regional Ocean-Atmosphere Modeling

• simulates realistic eastern Pacific climate.

• is useful for testing ideas:– Without SC, stratus cloud is 0.1-0.6

greater and SST 1-4° C cooler.– Thermodynamically coupled SST is

essential to many sensitivity experiments.

• IROAM will be validated against EPIC and VOCALS observations.

Eastern Pacic coupled GCM biases• are largely due to incorrect seasonal

cycle and meridional temperature asymmetry.

• affect wind on the equator and cold tongue SST.

• The mean state and seasonal cycle affect interannual variability.