Verifying the water vapor feedback using ENSO
A. E. Dessler and S. WongDepartment of Atmospheric Sciences
Texas A&M University
2
forcing
e.g., CO2
increase
Ts
increase
Atmospheric
humidity
increases
feedback
How can we measure this?
3
Volcano
ENSO+interannual
seasonal cycle
decade-scale warming
Ts
increase
Constant Relative
Humidity
Partial Radiative
Perturbation
Atmospheric
humidity
increases
4
VolcanoSoden et al., 2002; Forster and Collins,
2004
ENSO+interannualSoden, 1997; Dessler et al., 2008;
Minschwaner and Dessler, 2004;
Gettelman and Fu, 2008;
seasonal cycleInamdar and Ramanathan, 1998;
Wu et al., 2008
decade-scale warmingHall and Manabe, 1999; Soden et al.,
2005
Observational tests of the
water vapor feedback
Partial Radiative Perturbation
5
Ts-a
qa
Tatm-a
icea
Ts-b
qb
Tatm-b
iceb
Wetherald and Manabe, 1988
Colman, 2003; Forster and Collins, 2004;
Soden and Held, 2006
6
Ts-a
Tatm-a
icea
Ts-b
Tatm-b
iceb
TOA net flux Rb
qa qb
new
Wetherald and Manabe, 1988?
Colman, 2003; Forster and Collins, 2004;
Soden and Held, 2006
TOA net flux Ra
7
Ts-a
Tatm-a
icea
Ts-b
Tatm-b
iceb
TOA net flux Rb
qa qb
new
Wetherald and Manabe, 1988?
Colman, 2003; Forster and Collins, 2004;
Soden and Held, 2006
!R = change in global average TOA
flux due to !q
!Ts = change in global avg. surface
temperature change (Ts-b-Ts-a)
associated with !q
!R = change in global average TOA flux
due to !q between two climate states
!Ts = change in global avg. surface
temperature change (Ts-b-Ts-a) associated
with !q
8
Wetherald and Manabe, 1988?
Colman, 2003; Forster and Collins, 2004;
Soden and Held, 2006
!
" =#R
#TS
the water vapor feedback
!
Ts=
"G
#o
+ #1+ #
2+!
Method
• AMIP models from the PCMDI archive
• Reanalysis: ERA40 & MERRA
• For each strong DJF ENSO month (e.g., Jan. 1983),
calculate !R and !Ts between that month and the long-
term average for all Januaries in the model run
• Obtain one estimate of !R and !Ts for every strong ENSO
month
9
10
Each point is a strong ENSO DJF
!
" =#R
#TS
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
!R (
W/m
^2
)
0.80.40.0-0.4
!Ts (K)
NCAR CCSM
11
12
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
q f
eedback (
W/m
^2
/K)
A B C D E F G H I J K L
Model
(a)
ER
A40
ME
RR
A
AMIP models
Des
sler
2008
!
" =#R
#TS
Soden and Held estimate of long-term feedback
Conclusions 1
• Models, reanalysis, and pure obs. agree that
the feedback is positive and strong
• Reanalysis is larger than the models
– if reanalysis is correct, then models are
underestimating the feedback
– uncertainties generally overlap
• ENSO WV feedback larger than in response
to long-term warming (e.g., Colman, Soden and Held)
13
Questions for the rest of the talk
• What is the source of disagreements among
the models and between the models and
reanalyses?
• Speculation: Why is the ENSO feedback
larger than the feedback in response to long-
term warming?
14
15Fig. 2 of Soden et al., 2008
Water vapor feedback is primarily a “tropical” phenomenon
Change in R per unit change in q(x,y,z): !R/!q(x,y,z)
* !R determined by tropical UT !q
* tropical q controlled by tropical
surface temperaturese.g., Minschwaner and Dessler, 2004
* !R (and the WV feedback) is
controlled by tropical surface T
!R
(!q)
!(global avg. T)
Regress !R vs. global surface
temperature
!R
(!q)
!(tropical avg. T)
Regress !R vs. tropical surface
temperature
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
q f
eedback (
W/m
^2
/K)
A B C D E F G H I J K L
Model
(a)
!
"R
"TS
ER
A40
ME
RR
A
AMIP models
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
!R
q v
s. !
TT (
W/m
^2
/K)
A B C D E F G H I J K L
Model
(b)
!
"R
"TT
What does this mean?
• A consistent relationship exists between
tropical surface !TT and the radiative
response to water vapor !R
– GCMs
– reanalyses
– pure obs.
19
+1
+1
+1
!
" =#R
#TS
+1
-1
-1
!R for these two worlds is the same
!Ts is different
smaller feedback larger feedback
21
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
!T
G/!
TT (
K/K)
4.54.03.53.02.52.0
Feedback strength (W/m^2/K)
A
B
C D
EF
G
H
I
J K
L
Extratropics
warms more
Extratropics
warms less
!
" =#R #q #TT( )( )
#TG
Summary• In response to ENSO climate change, models, reanalysis,
and pure obs. show a strong and positive water vapor feedback
• Models, reanalysis, and pure obs. show that the radiative response to WV between two climate states is determined by change in the tropical surface T
• WV feedback determined by tropical UT water
• Tropical UT water is controlled by surface T described by Minschwaner and Dessler [2004] (see me for reprints)
• Differences among models and between models and reanalyses are due to differing estimates of extratropical surface T changes --- they are NOT due to the radiative response of WV
• This work is in press at J. Climate. See me for preprints.