Transcript Slide 1 - International Pacific Research Center

Changing
South Pacific
rainfall bands in
a warming
climate?
Image from 8 February 2012
MTSAT-2 visible channel, Digital Typhoon, National Institute of Informatics
Spotlight on the South Pacific
Convergence Zone:
How will Pacific rainfall bands respond to a
warming climate?
Collaborators:
Shayne McGregor1,
Matthew H. England1,
Matthieu Lengaigne2,
and Wenju Cai3
University of Hawaii colleagues:
Axel Timmermann,
Niklas Schneider, and
Karl Stein
1Climate
Change Research Centre,
University of New South Wales
2LOCEAN, France
3CSIRO Marine and Atmospheric
Research, Australia
Matthew J. Widlansky
International Pacific Research Center
Southern Hemisphere Convergence Zones
Austral summer (DJF) climatology of satellite observed rainfall
(GPCPv2.1)
South Indian
Convergence Zone
(SICZ)
South Pacific
Convergence Zone
(SPCZ)
South Atlantic
Convergence Zone
(SACZ)
5 mm day-1 contour indicated by blue line
SPCZ is the largest rainband in the Southern
Hemisphere and provides most of the rainfall for
South Pacific island nations
Historical perspective: Very early ship observation
“I was struck by the precise similarity of the
clouds, sky, peculiarities of wind, and weather, to
what we had been accustomed to meet with off
the coast of Patagonia: and I may here remark
that, throughout the southern hemisphere, the
weather, and the turn or succession of winds, as
well as their nature and prognostications, are
remarkably uniform.”
Captain Fitz-Roy,
Narrative of the surveying voyages of His Majesty’s Ships Adventure and
Beagle between the years 1826 and 1836
Historical perspective: Early satellite observations
Cloud Cover Percentage (DJF)
S. Pacific
~30%
Quasi-stationary Southern Hemisphere
cloud band locations are
“related closely to that of the longwave hemispheric pattern.”
Satellite cloud brightness
1968-1971 composite of 5 day averages
S. Indian
~20%
S. Atlantic
~30%
(Streten 1973, Mon. Wea. Rev.)
Historical perspective: Literature review
SPCZ
SPCZ
SICZ
SACZ
• SPCZ is a region of widespread cloud cover and
rainfall extending southeastward from New
Guinea into Southern Hemisphere mid-latitudes.
(Streten 1973; Trenberth 1976)
• Tropical convection is oriented zonally and
collocated with warmest SST. (Vincent 1994)
• Baroclinic-type disturbances influence the
diagonal region. (Kiladis et al. 1989)
• Orientation changes during different phases of
the El Niño-Southern Oscillation (ENSO).
(Trenberth 1976; Streten and Zillman 1984; Karoly and Vincent 1999;
Folland et al. 2002)
Why does the SPCZ extend diagonally away
from the equator in the Southern Hemisphere?
Outline

1) Historical perspective
2) Snapshot of SPCZ science, circa 2010
3) Recent advancements in understanding:
•
Why is there a diagonal rainband?
•
How will the rainband respond to climate change?
•
Will frequency of future extreme SPCZ events change?
Answers to these questions are based on the
underlying sea surface temperature (SST)
distribution and its projected change
August 2010: State of the science
Workshop on the SPCZ
Apia, Samoa (Aug. 2010)
Secretariat of the Pacific
Regional Environment
Programme
The Pacific climate change
Science Program
1) Hypothesis for dynamics of the SPCZ
2) SPCZ related extreme events on interannual
timescales such as droughts, floods, and tropical cyclones
3) Projections of the SPCZ response to climate change
1) Dynamics of the SPCZ
Observed rainfall and SST climatology during DJF
28°C
(mm day-1)
26°C
~ GPCP rainfall
~ NOAA SST
• Tropical SPCZ adjacent the meridional SST gradient (equatorial)
(e.g., Lindzen and Nigam 1987)
• Subtropical SPCZ transects the meridional SST gradient (mid-latitudes) and is west
of maximum zonal SST gradient
(2011, Clim. Dynam.)
2) Interannual variability of the SPCZ
Observed rainfall and SST climatology during DJF
(mm day-1)
Extreme
El Niño
28°C
26°C
El Niño
La Niña
~ GPCP rainfall
~ NOAA SST
Neutral (mean)
Extreme El Niño (anomaly)
Tropical Cyclone
Genesis
Adapted from Figure 12 (Vincent et al. 2011, Clim. Dynam.)
3) Uncertain rainfall projection in IPCC AR4
DJF (2080 to 2099)
Rainfall projection (%)
%
IPCC Fourth Assessment Report
“Regional Climate ProjectionsSmall Islands”:
1) Rainfall is likely to increase along
equator and decrease in the Southeast
Pacific (where it is already dry)
Number of models > 0
CMIP3 (A1B, 21 models)
Adapted from Figure 11.25 (IPCC AR4, Chapter 11)
2) Multi-model mean trend is small in the
SPCZ and inter-model uncertainty is large
3) Impact of coupled model biases on
future rainfall projections not addressed
Fundamental questions unanswered in Samoa
Why is there a diagonal rainband in the Southern Hemisphere, but
not in the Northern Hemisphere?
Why is the tropical Pacific rainfall response to greenhouse warming
so uncertain?
How will extreme events, such as strong El Niño occurrences and
zonally oriented SPCZ events, respond to climate change?
Today, I will present three papers (2012)
addressing each question individually
Question #1
Why is there a diagonal rainband in the Southern Hemisphere, but
not in the Northern Hemisphere?
Why is the tropical Pacific rainfall response to greenhouse warming
so uncertain?
How will extreme events, such as strong El Niño occurrences and
zonally oriented SPCZ events, respond to climate change?
(Q. J. Roy. Meteor. Soc., 2012)
Influence of SST forcing on basic-state
SST Climatology
240 W m-2 OLR (rainfall proxy) climatology indicated by blue line
SPCZ (A) is west of the maximum zonal gradient (B-C)
The background quasi-stationary 200 hPa flow is partially dictated
by the SST distribution (e.g., Gill 1980)
Upper-troposphere zonal flow
200 hPa Zonal Winds200
& Negative
Zonal
Stretching Deformation
hPa Zonal
Winds
A decelerating jet stream creates a band of upper-tropospheric negative
zonal stretching deformation (s-1, 200 hPa) near the subtropical SPCZ:
U
0
x
Distribution of mean zonal winds acts to refract Rossby waves
(e.g., Hoskins and Ambrizzi 1993, J. Atmos. Sci.)
SPCZ acts as a synoptic ‘graveyard’
From Matthews (2012, Q. J. Roy. Meteor. Soc.):
“The propagation of Rossby waves in a spatially varying mean
flow can also be interpreted in terms of accumulation of wave
energy (Webster and Holton, 1982). In particular, in jet-exit regions where
the mean westerly wind u decreases eastward (∂u/∂x < 0), the
zonal wavenumber will increase along a ray path. This leads to a
decrease in the wave group speed and an increase in the wave
energy density (Webster and Chang, 1998). When applied in the region of
the SPCZ (Widlansky et al., 2011), this can explain the observation that the
SPCZ acts as a synoptic ‘graveyard’ (Trenberth, 1976).”
TRANSIENT
WAVES
∂U/∂x < 0
Modes of SPCZ variability
Observed rainfall and 200 hPa zonal wind (DJF)
‘Shifted SPCZ’ mode 1 (12%)
~ TRMM rainfall
~ NCEP Reanalysis U
SPCZ position and intensity varies on
multiple timescales:
• Synoptic, Rossby waves
• Intraseasonal, MJO
• Interannual, ENSO
(e.g., Widlansky et al. 2011, Clim. Dynam.)
Adapted from Figures 1 and 3 (Matthews 2012, Q. J. Roy. Meteor. Soc.)
Later, we will look at mode 2
Synoptic disturbances from higher latitudes
‘Shifted SPCZ’ (mode 1) composite:
OLR (rainfall proxy) and 200 hPa vorticity anomalies
Path of wave
propagation
Mean diagonal SPCZ is the sum of equatorward
propagating synoptic waves from the subtropical jet
towards the equatorial westerly wind duct
Adapted from Figure 5 (Matthews 2012, Q. J. Roy. Meteor. Soc.)
Change in basic-state during ENSO
La Niña minus El Niño:
SST anomaly (shading)
ULa Niña = 0
UEl Niño = 0
Path of wave
propagation
Westerly wind duct constricts during El Niño,
hence synoptic waves refract equatorward further east,
shifting the diagonal SPCZ northeastward
Adapted from Figure 11 (Matthews 2012, Q. J. Roy. Meteor. Soc.)
Why no diagonal rainband in North Pacific?
(Q. J. Roy. Meteor. Soc., 2012)
A diagonal rainband is the default, triggered by equatorward
refraction of synoptic waves, but in the North Pacific:
1) Subtropical jet is strong and narrow (topography)
2) Equatorial westerly wind duct is absent during
Northern Hemisphere summer (weaker Walker circulation)
3) NH warm pool is confined near equator during winter
SPCZ orientation determined by warm pool
configuration and its projected change
Question #2
Why is there a diagonal rainband in the Southern Hemisphere, but
not in the Northern Hemisphere?
Why is the tropical Pacific rainfall response to greenhouse warming
so uncertain?
How will extreme events, such as strong El Niño occurrences and
zonally oriented SPCZ events, respond to climate change?
(2012, in press)
Uncertainty remains in CMIP5
21st century projection (shading)
20th century control (black lines)
Regional rainfall trend
Rainfall change (% Control)
Rainfall trend (RCP 4.5, 21 models)
Inter-model standard deviation
Equatorial islands
SPCZ islands
Inter-model standard deviation
Inter-model uncertainty is larger than
ensemble mean projected rainfall trend
Blue lines enclose simulated 20th century rainfall > 5 mm day-1
Model bias and projected rainfall change
(% 20th century control)
scaled by warming at equator, K-1
Rainfall projection
Shifted South Pacific rainfall bands in a
warming climate?
r2 = 0.27
(n = 74)
Tropical SPCZ
(10°S-20°S,
150°E-150°W)
Rainfall bias
RainfallOBS >5 mm day-1
(% observed climatology)
Procedure: Uncertain rainfall projection?


SST gradients influence the observed location and
strength of the SPCZ
Coupled GCMs yield uncertain 21st century rainfall
projections, especially in Southwest Pacific
1) Removing SST bias improves simulated diagonal
rainband
2) Bias-corrected climate experiments suggest future
drying as regional SST gradients weaken
3) Net rainfall change depends on balance of two
mechanisms (of opposite sign)
Goal is to explain inter-model uncertainty
Removing SST bias improves climatology
SST bias (CMIP5, 20 models)
Equatorial Pacific is too cold and Southeast
Pacific is too warm
• Double-ITCZ bias partly related to SST biases
Warm Pool
(e.g., Wittenberg et al. 2006, J. Clim.)
• Atmosphere GCMs (observed SST) simulate a more
diagonal SPCZ
AMIP rainfall is too heavy
Rainfall climatology (CMIP5, 21 models)
Rainfall bias (CMIP5, 21 models)
Rainfall climatology (AMIP, 5 models)
Rainfall bias (AMIP, 5 models)
Robust SST warming pattern
SST trend
(tropical mean removed)
21st century projection (RCP 4.5 W m-2, 20 models)
Inter-model standard deviation
Maximum equatorial
warming is a
robust response to
greenhouse warming
(e.g., Xie et al. 2010, J. Clim.)
Green lines enclose simulated 20th century Warm Pool (27.5 °C)
Biased SST climatology does not affect SST projection
Warm Pool (27.5°C), climatology
Shading, warming trend
Coupled GCM (CCSM3) response to 2xCO2
No flux correction
Radiative flux correction
Removing SST bias does not change the warming
pattern and improves rainfall climatology
Each experiment projects more rain along equator and drying in the South
Pacific, but drying in SST bias-corrected experiment occurs in Southwest Pacific
collocated with observed SPCZ
CO2 increased 10% per year to 710 ppm
Projections from last 20 years of 90 year simulations
Bias-corrected island rainfall projections
Coupled GCM (CCSM3) response to 2xCO2
Equatorial islands
SPCZ islands
CCSM3 experiment with no flux correction
shows no consistent rainfall projection for
the SPCZ islands
Equatorial islands
SPCZ islands
SST bias-correction experiment projects
drying for SPCZ islands (typically 5-10%)
and more rain along some parts of the
equator
Rainfall response to changing SST gradients
21st century trend (tropical mean removed)
CMIP3 A1B scenario
2 and ½ Layer Atmospheric Model
Green lines enclose observed Warm Pool (27.5°C) and the
changing threshold for deep convection (dashed)
Idealized Atmospheric GCM (ICTP)
(Graham and Barnett 1987, Science;
Johnson and Xie 2010, Nature Geosci.)
• SST bias-corrected experiments have a
more realistic SPCZ climatology than
coupled GCMs.
• In response to
century SST gradient
pattern, rainfall increases where SST
warms the most and decreases elsewhere.
21st
• SPCZ drying is a robust response
regardless of model resolution or
convection parameters.
Full Atmospheric GCM (CAM3)
21st century projection (shading)
20th century control (blue & black lines)
Increasing model complexity
Tropical Channel Run
Rainfall response to tropical mean SST increase
Rainfall response
(Total SST trend)
=
Rainfall response
(SST gradient pattern)
Warmest regions tend to get wetter
(Ma et al. 2012, J. Clim.)
+
Rainfall response
(Uniform SST warming, 2.2°C)
?
Wet regions tend to get wetter
(Held and Soden 2006, J. Clim.)
“Wet gets wetter”
Thermodynamic mechanism
Rainfall response to tropical mean
SST increase (2.2°C):
Mean specific humidity increases over entire
tropical Pacific supporting an enhanced
hydrological cycle.
(Held and Soden 2006, J. Clim. and Seager et al. 2010, J. Clim.)
Contours depict projected moisture increase (lower troposphere)
as simulated by AGCM forced with 21st century SST trend (A1B)
“Warmest gets wetter”
Rainfall and wind response to
prescribed SST gradient:
Dynamic mechanism
• Anomalous divergence of moisture away
from minor warming regions, such as SPCZ.
• Moisture convergence towards warmest
waters accounts for increased rainfall at
equator.
(Ma et al. 2012, J. Clim.)
Red contours depict warming more than tropical mean
21st century multi-model trend (CMIP3 A1B emissions)
Delicate balance of opposing rainfall mechanisms
Warmest
gets wetter
Warmest gets
wetter
Wet gets
wetter
Wet gets
wetter
~2
CO2scenario
scenario
4 xxCO
2
How does this balance change for more extreme
greenhouse-warming?
Rainfall response to tropical mean
SST increase for 4 x CO2 (4.4°C):
AMIP-future ensemble (4 x CO2 SST)
projects rainfall increase for parts of SPCZ
For 4 x CO2 conditions, wet gets wetter mechanism
almost completely offsets SPCZ drying associated with
diminished SST gradient between SPCZ and Equator
Moisture convergence in the SPCZ
SPCZ rainfall response to greenhouse warming influenced by two opposing mechanisms:
1) Increasing moisture convergence in lower troposphere (Thermodynamic mechanism)
76 experiments
% 20th century observations
Moisture convergence (g kg-1 s-1)
in the SPCZ
2) Divergence of moisture away from the rainband towards equatorial
regions of greater warming (Dynamic mechanism)
Robust response
Large inter-model
spread
Net drying
Net moisture increase
Projected SST trend (°C) in the SPCZ
Answers: Uncertain rainfall projection?


SST gradients influence the observed location and
strength of the SPCZ
Coupled GCMs yield uncertain 21st century rainfall
projections, especially in Southwest Pacific
1) Removing SST bias improves simulated diagonal
rainband, but rainfall intensity is prone to errors
2) According to bias-corrected experiments, summer
rainfall may decrease by 10-20% for some South
Pacific islands, assuming moderate warming
3) Net rainfall change depends on delicate balance of
opposing thermodynamic and dynamic mechanisms
Multi-model scatter of net moisture convergence helps
explain inter-model variance in CMIP5 rainfall projections
Question #3
Why is there a diagonal rainband in the Southern Hemisphere, but
not in the Northern Hemisphere?
Why is the tropical Pacific rainfall response to greenhouse warming
so uncertain?
How will extreme events, such as strong El Niño occurrences and
zonally oriented SPCZ events, respond to climate change?
(Nature, 2012)
Defining a “zonal SPCZ event” : PC1 > 1 and PC2 > 0
GPCP rainfall
La Niña
Zonal SPCZ
Neutral
Moderate El Niño
Nonlinear behavior of 2nd principal component
PC2
PC1
1997/98
El Niño
1997/98
El Niño
How will “zonal SPCZ events” respond to
climate change?
CMIP5 experiments
Considering only models able to simulate the nonlinear
behavior of the SPCZ (12 out of 20 models)
20th century
95 zonal SPCZ events
21st century
201 zonal SPCZ events
RCP 8.5 W m-2
Correcting SST errors
Flux adjusted perturbed physics experiments with HadCM3 CGCM
(12 out of 17 experiments considered)
20th century
57 zonal SPCZ events
21st century
179 zonal SPCZ events
CO2 increased 1% per year
Meridional SST gradient & zonal SPCZ events
Observed rainfall and SST climatology during DJF
(mm day-1)
Box 1
28°C
26°C
Box 2
~ GPCP rainfall
~ NOAA SST
1997/98
El Niño
= [Box 1 SST – Box 2 SST]
Smaller future meridional SST gradient
SST trend
(departure from tropical mean)
Box 1
Box 2
21st century projection (RCP 4.5 W m-2, 20 models)
Maximum equatorial warming
is a robust response to
greenhouse warming
(e.g., Xie et al. 2010, J. Clim.)
Increased number of zonal SPCZ events
Flux adjusted perturbed physics experiments with HadCM3 model
(12 out of 17 experiments considered)
1
2
Greenhouse warming is likely to cause:
1) More summers with small meridional SST gradients
2) Increased frequency of zonal SPCZ events
Pacific island communities experience extreme weather
–droughts, floods, & tropical cyclones–
during zonal SPCZ events
Extreme zonally oriented SPCZ event:
4 January 1998
GMS-5
IR water vapor
6.70-7.16 μm
Susan
(125 kts)
Katrina
(28 days)
Ron
(Tonga: 67%
damaged)
Increased
frequency of
extreme events
is consistent
with projected
SST warming
pattern
Thank you