the summer nao in cmip3 models and related uncertainties in precipitation projections ... ·...
TRANSCRIPT
The summer NAO in CMIP3 models and related uncertainties in
precipitation projections in the Euro-Mediterranean region
Ileana Bladé (1), Didac Fortuny, Geert Jan Van Oldenborgh and Brant Liebmann
(1) Department of Astronomy and Meteorology. University of Barcelona
) • Bladé et al. 2012 (in press, JGR-Atmospheres)
Motivation: models predict substantial summer precipitation reductions in Europe and the Mediterranean
IPCC‐AR4:PREDICTEDENSEMBLE‐MEANRELATIVECHANGEINPRECIPITATIONFROM1980‐1999TO2080‐2099
The summer NAO • defined as the leading EOF of July-August
SLP in the domain [40ºN-70ºN; 90ºW-30Eº], following Folland et al. 2009 and Greenbacth and Rong (2006)
• dominant dynamical pattern of summer precipitation variability in the Euro/Mediterranean area (Bladé et al. 2011)
NOTE: results presented here are insensitive to details in the definition of SNAO !
SNAO is negatively correlated with precipitation in northern
Europe, positively correlated with precipitation in Mediterranean
PRECIP CORRELATIONS (E-OBS)
SNAO in observations
!"## !"#$ !"!# !"!$ !"%# !"%$ !"&# !"&$ !"'# !"'$ !"$# !"$$ !"(# !"($ !")# !")$ !"*# !"*$ !""# !""$ %### %##$ %#!#!&
!%
!!
#
!
%
&
����������������������������
!"## !"#$ !"!# !"!$ !"%# !"%$ !"&# !"&$ !"'# !"'$ !"$# !"$$ !"(# !"($ !")# !")$ !"*# !"*$ !""# !""$ %### %##$ %#!#!!
!#+$
#
#+$
!
���������������������������������� ��
SNAO time evolution is dominated by multidecadal fluctuations …..
but no significant long-term trend
SNAO time series
(Bladé et al. 2011)
SNAOimpactovertheMediterraneanisrelatedtoanupperleveltroughwhichispartofahemisphericpa:ernofanomalies
Upper-level trough implies mid-tropospheric cooling and thus potential instability -> favors summer convection
(Bladé et al. 2011)
OBJECTIVES
To attempt to validate climate model projections by
• documenting CMIP3 model performance with regards to the SNAO
• assessing the contribution of the SNAO in the projected summer drying in the Euro/Mediterranean region in the CMIP3 simulations
Most CMIP3 models reproduce the spatial pattern of the SNAO
a“good”model
a“bad”model
(Bladé et al. 2012)
Taylor‐likeplotcomparingsimulatedandobservedSNAO
To be fair to the models we project the observed SNAO onto the SLP field … And discard those that just don’t seem to have a SNAO (4 out of 24) We are left with 20 models and 44 simulations (1950-2100 period)
CMIP3 models also reproduce the broad-scale
dipolar precipitation signature
but the signal in the Mediterranean is too weak ….
which is related to weak troughs at 200 hPa
OBSERVATIONS MULTIMODEL ENSEMBLE
Land‐oceanprecipita=ondataset(NOAA)
(Bladé et al. 2012)
WE STRATIFY PRECIPITATION RESPONSES AND SHOW HIGH END /LOW END RESPONSES (9 out of 44)
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
NORTHWEST EUROPE SOUTHEAST EUROPE/MEDITER.
even the “best” models are too weak in the Mediterranean …
Weak response to SNAO in the Mediterranean is not due to too weak precipitation in response to an upper level trough
but to the weak SNAO-related trough in the models
-0.016
-0.014
-0.012
-0.010
-0.008
-0.006
-0.004
-0.002
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
regr
essio
n pr
ecip
itatio
n-z2
00 [(
mm
/day
)/m]
regression precipitation-SNAO [mm/day]
REGRESSION OVER THE BALKAN/ITALY REGION
(a)
r=-0.34 [-0.58,0.02]r(no outliers)=0.01
-20
-15
-10
-5
0
5
10
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
regr
essio
n z2
00-S
NAO
[m]
regression precipitation-SNAO [mm/day]
(b)
r=-0.36 [-0.66,-0.04]r(no outliers)=-0.26
bccrbcm20ccsm3
cgcm31t47cgcm31t63
cnrmcm3csiromk30csiromk35
echam4
echam5gfdlcm20gfdlcm21
gisseh
hadcm3hadgem1inmcm30miroc32h
miroc32mmricgcm212
pcmobs.
Regression of precipitation onto Z-200 trough versus regression of precipitation onto SNAO
(Bladé et al. 2012)
21stcentury:CMIP3modelsprojectanupwardSNAOtrend
The SNAO trend accounts for 50% of the ensemble-mean drying in NW Europe
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
prec
ipita
tion
trend
[(m
m/d
ay)/c
entu
ry]
SNAO trend [per century]
(a) NWE-2 TREND (2010-2099)
r=-0.80 [-0.88,-0.67]r (no outliers)=-0.69
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
prec
ipita
tion
trend
[(m
m/d
ay)/c
entu
ry]
SNAO trend [per century]
(b) BAL/ITA TREND (2010-2099)
r=-0.49 [-0.69,-0.17]r (no outliers)=-0.15
bccrbcm20ccsm3
cgcm31t47cgcm31t63
cnrmcm3csiromk30csiromk35
echam4
echam5gfdlcm20gfdlcm21
gisseh
hadcm3hadgem1inmcm30miroc32h
miroc32mmricgcm232
pcmobs.
r=0.8
and also for 64% of the spread (uncertainty) in the magnitude of this drying
HADCM3 GFDLCM2.1
SNAO trend NW
pre
cip
tren
d SNAO contribution
MME precip trend
(Bladé et al. 2012)
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0pr
ecip
itatio
n tre
nd [(
mm
/day
)/cen
tury
]
SNAO trend [per century]
(a) NWE-2 TREND (2010-2099)
r=-0.80 [-0.88,-0.67]r (no outliers)=-0.69
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
prec
ipita
tion
trend
[(m
m/d
ay)/c
entu
ry]
SNAO trend [per century]
(b) BAL/ITA TREND (2010-2099)
r=-0.49 [-0.69,-0.17]r (no outliers)=-0.15
bccrbcm20ccsm3
cgcm31t47cgcm31t63
cnrmcm3csiromk30csiromk35
echam4
echam5gfdlcm20gfdlcm21
gisseh
hadcm3hadgem1inmcm30miroc32h
miroc32mmricgcm232
pcmobs.
But in the Mediterranean, where the model’s impact on precipitation is too weak, the upward SNAO trend is
inconsistent with observations (no precipitation decreases)
SNAO trend Med
iter
rane
an p
recip
tren
d
(Bladé et al. 2012)
CONCLUSIONS • CMIP3 models reproduce the spatial SLP signature of the SNAO
and the broad-scale dipolar pattern of precipitation over Europe, but the precipitation impact over the Mediterranean is too weak. The model error is related to their inability to reproduce the observed upper-level trough that favors convection in observations.
• Many CMIP3 predict an upward SNAO trend in the XXIst century.
This trend accounts for 50% of the mean projected precipitation decrease in NW Europe. However, the large uncertainty in the SNAO trends results in a very large large uncertainty in the magnitude of this drying (64%).
• The corresponding expected increase in Mediterranean precipitation that would partially offset the drying due to other processes does not occur in CMIP3 models.
• Because the Mediterranean drying feedbacks on the circulation in central Europe via the local heat low, the model biases in the Mediterranean due to the missing SNAO teleconnection may lead to excessive drying over central Europe also.