Use of long-term MSU/AMSU data to examine
the weakening of Walker Circulation in CMIP5
climate simulations
1B.J.
Sohn, 2Eui-Seok Chung, and 1Hwanjin Song
1Seoul
National University, Seoul, Korea
2 University of Miami, US
Current problems: Climate model simulations and theoretical assessment
suggest that Walker circulation should be weakened under the global
warming conditions. In contrast observations at least over the past three
decades indicate the strengthening.
Mail goal: to diagnose the relative importance of diabatic heating vs dry
static stability for the Walker circulation variability & trend.
Method: The first-order thermodynamic energy balance of large-scale
tropical circulation (at any time and location) is between diabatic heating &
adiabatic cooling,
𝑸 ≈ −𝝎 × (−
𝑻 𝝏𝜽
) ≈ −𝝎 × (𝑺𝒑)
𝜽 𝝏𝝏
where Q is the diabatic heating; Sp = -(T/θ)(∂θ/ ∂p), the static stability
parameter, and ω = dp/dt.
MSU data: To obtain the stability parameter, satellite-derived temperatures
are needed at least two layers. Of course we need a long-term data set to
have an idea of climate change. MSU data may satisfy such conditions.
20
TTT Ch2+4(T24)
TLT Ch2
(nadir - limb)
Height (km)
15
TLT: MSU channel 2 (nadir − limb)
T24: (= a TMT – b TLS; a=1.1, b=0.1)
10
TLS: MSU Channel 4
TMT: MSU Channel 2
5
0
0.00
0.04
0.08
0.12
Weighting Function
 ℑ ( p )  ∂ℑ ( p )
0
ps
∂ℑ ( p )
TB ν = εν Ts ℑν ( ps ) + (1 − εν ) ℑν ( ps ) ∫ −T ( p )  ν s 2  ⋅ ν
dp + ∫ T ( p ) ν
dp
0
ps
∂
∂
p
p
 ℑν ( p ) 
()
0
∂ℑν ( p )
ps
∂p
=ενTs ℑν ( ps ) + ∫ Jν ( p )
MSU channel TB - TS term =
∫
0
ps
dp
Jν ( p )
∂ℑν ( p )
∂p
Data: ERA-Interim, MERRA reanalysis data
20 CMIP5 model simulations
Analysis period: 1979 - 2005
dp
Mean MSU T24 (K)
MSU
ERA
MERRA
CMIP5
Mean MSU TLT
0.3
0.3
OBS
ERA-Interim
MERRA
CMIP5
T24 - TLT (K)
0.2
0.2
0.1
0.1
0.0
0.0
-0.1
-0.1
-0.2
-0.2
-0.3
-0.3
1980
1985
1990
1995
2000
2005
T24 – TLT Anomaly Timeseries over the Tropics (20N-20S) for the
Period 1979-2005
Error bars in grey: ±1 intermodel standard deviation
OBS
OBS
ERA-Interim
ERA-Interim
MERRA
MERRA
Ensemble Mean
Ensemble Mean
ACCESS1-0
ACCESS1-0
BNU-ESM
BNU-ESM
CCSM4
CCSM4
CNRM-CM5
CNRM-CM5
GFDL-CM3
GFDL-CM3
GFDL-ESM2G
GFDL-ESM2G
GFDL-ESM2M
GFDL-ESM2M
GISS-E2-R
GISS-E2-R
HadGEM2-ES
HadGEM2-ES
INMCM4
INMCM4
IPSL-CM5A-LR
IPSL-CM5A-LR
IPSL-CM5A-MR
IPSL-CM5A-MR
MIROC-ESM-CHEM
MIROC-ESM-CHEM
MIROC-ESM
MIROC-ESM
MIROC5
MIROC5
MPI-ESM-LR
MPI-ESM-LR
MPI-ESM-P
MPI-ESM-P
MRI-CGCM3
MRI-CGCM3
NorESM1-M
NorESM1-M
NorESM1-ME
-0.02
NorESM1-ME
0.00
0.02
0.04
0.06
0.08
0.10
0.12
T24 - TLT Trend (K/decade)
T24 – TLT Trend over the Tropics (20N-20S) for the Period 1979-2005
Horizontal error bars: ±2 standard error of the linear trend except for multi-mo
del ensemble mean (±2 intermodel standard deviation)
∂θ
Mean ∂p (×100) estimated from T24 and TLT for 1979-2005
MSU
ERA-Interim
MERRA
CMIP5
More unstable!
Decadal trend of
MSU
∂θ
(x 103 K/hPa/decade)
∂p
Decadal trend of SST(K/decade)
(1979–2005)
HadISST
ERA
CMIP5
MERRA
CMIP
ER M
A ER
-In R
te A
r
SS im
M
M
/I
PI
-E
S
C MM G GF CS P
IR F D M
O DL L- 4
C
C
IP -ES -ES M3
SL M M
- C - C 2M
M HE
5
B A-M M
N
U R
I
H NM -ES
ad
- M
M GE CM
R M 4
M I-CG 2-E
PI C S
-E M
IP CN SM 3
SL R -C M- LR
C
N M5 M
or A 5
E G S M LR
N ISS 1
or - -M
E
M E S 2G IRO M1- R
FD C M
L -E E
A -E S S M
C M
C
E S 2G
M S1
IR -0
O
C
5
-1
WCI trend [m s /10 years ]
Walker Circulation trends (1979-2012)
0.12
0.10
0.08
+
_
Observations
0.06
+ WCI models
0.04
0.02
- WCI models
0.00
-0.02
-0.04
 The WCI trends of CMIP5 models were highlighted with red/blue boxes if the t-test satisfies the 95 % confidence level.
𝐓 𝛛𝛉
−
𝛉 𝛛𝛛
ERAInterim
Mean Sp over 20N-20S (1979-2012)
•••• • • •••• • • •••• • •
WP
EP
•
MERRA
CMIP5
+WCI
moist adiabat
CMIP5
-WCI
10
𝑸 ≈ −𝝎 × (𝑺𝒑)
� × Δ𝑺𝒑
Δ𝑸 ≈ −Δ𝝎 × 𝑺𝒑 − 𝝎
ERA (1999-2012) minus (1979-1998)
ΔQ
WP
EP
Color
(ΔSp)
Contour
(Δ ω)
−𝚫𝛚 × 𝐒�𝐩
20%
� × 𝚫𝐒𝐩
−𝛚
Q [K/day]
ω [Pa/h]
Sp [K/MPa]
11
CMIP5 +WCI (1999-2012) − (1979-1998)
ΔQ
Color
(ΔSp)
Contour
(Δ ω)
−𝚫𝛚 × 𝐒�𝐩
50%
� × 𝚫𝐒𝐩
−𝛚
Q [K/day]
ω [Pa/h]
Sp [K/MPa]
12
CMIP5 –WCI (1999-2012) − (1979-1998)
ΔQ
Color
(ΔSp)
Contour
(Δ ω)
−𝚫𝛚 × 𝐒�𝐩
50%
� × 𝚫𝐒𝐩
−𝛚
Q [K/day]
ω [Pa/h]
Sp [K/MPa]
13
Conclusions
In contrast to MSU and reanalysis data CMIP5 models suggest:
(1) overall more unstable atmosphere over the tropics
(2) Lacks the 800-hPa local maximum, and appears to closely resemble the Sp
profile of the moist adiabatic lapse rate
(3) Lacks the stable layer below 800 hPa in the eastern Pacific.
(4) Smaller vertical variations in Sp than in the reanalysis
(5) The models appear to closely follow the moisture adiabatic profile
(6) Stability trend closely resembles SST trend.
Changes in diabatic heating should be the main element of inducing
weakening/strengthening of Walker circulation.
Stabilities in CMIP models looks simpler, but the contributions to the
circulation intensity looks far higher than shown in analysis data.
CMIP models seems to follow moist adiabatic lapse rate (which is supposed to
be more stable -- explaining weakening of the circulation), but in contrast
CMIP models are overall more unstable (inducing El Nino-like climate).