Transcript Document

Winds of Change:
Evolution of the El Niño Southern Oscillation
(ENSO) from the Last Ice Age to Today
Andy Bush
Dept. of Earth & Atmospheric Sciences
University of Alberta
Winds of Change:
Evolution of the El Niño Southern Oscillation
(ENSO) from the Last Ice Age to Today
Andy Bush
Dept. of Earth & Atmospheric Sciences
University of Alberta
A.B.G. Bush, 2006, Journal of Climate, in press.
Winds of Change:
Evolution of the El Niño Southern Oscillation
(ENSO) from the Last Ice Age to Today
Andy Bush
Dept. of Earth & Atmospheric Sciences
University of Alberta
A.B.George Bush, 2006, Journal of Climate, in press.
Winds of Change:
Evolution of the El Niño Southern Oscillation
(ENSO) from the Last Ice Age to Today
Andy Bush
Dept. of Earth & Atmospheric Sciences
University of Alberta
A.B.George Bush, 2006, Journal of Climate, in press.
[email protected]
Motivation: To understand the climatological factors
that determine the period and intensity of interannual
variability (ENSO).
Motivation: To understand the climatological factors
that determine the period and intensity of interannual
variability (ENSO).
Past climates provide altered mean states within which
interannual variability exists.
Animation of TOPEX/Poseidon sea surface height data
QuickTime™ and a
Cinepak decompressor
are needed to see this picture.
Some human impacts of ENSO:
1) Impact on disease spread
(malaria and dengue)
2) Food production
El Niño winter snow anomalies
Observed Composite
Temperature Anomalies
El Niño
La Niña
SST anomalies, Aug.-Sept. 2006: El Niño’s coming…
Existing numerical models for ENSO prediction are
anomaly models in which a background climate state is
assumed. Predicted variables are perturbations on that
background state. The two climate variables that must be
assumed are:
1) Mean depth of the thermocline
2) Strength of the climatological easterly trade winds
These quantities are known for today’s climate, so anomaly
models work quite well for ENSO prediction.
Existing numerical models for ENSO prediction are
anomaly models in which a background climate state is
assumed. Predicted variables are perturbations on that
background state. The two climate variables that must be
assumed are:
1) Mean depth of the thermocline
2) Strength of the climatological easterly trade winds
These quantities are known for today’s climate, so anomaly
models work quite well for ENSO prediction.
However, one or both of these quantities appear to have
been different in the past (aeolian deposits, upwelling indices,
planktonic foraminifera, etc.). Changes in the strength of the
general circulation can cause changes in these quantities.
Our atmosphere exhibits dynamic variability associated
with midlatitude baroclinic waves, or eddies.
Our atmosphere exhibits dynamic variability associated
with midlatitude baroclinic waves, or eddies.
Eddies may be either TRANSIENT (not fixed to a specific
geographic location) or STATIONARY (fixed geographically;
caused by mountain ranges, continent-ocean contrasts, etc.)
Eddies play a very important role in governing the strength
of the general circulation.
QuickTime™ and a
FLIC Animation decompressor
are needed to see this picture.
Atmospheric eddies are the primary mechanism by which low
latitude HEAT is transported poleward (v’T’>0). This occurs in
the growth phase of baroclinic waves. (Idealized life cycle)
Atmospheric eddies are the primary mechanism by which low
latitude HEAT is transported poleward (v’T’>0). This occurs in
the growth phase of baroclinic waves.
They are also the primary mechanism by which the zonal mean
(and, by angular momentum conservation, the meridional mean)
flow is forced (u’v’>0). This occurs in the Rossby wave decay
phase of the baroclinic wave, in which easterly momentum is
transported equatorward.
Atmospheric eddies are the primary mechanism by which low
latitude HEAT is transported poleward (v’T’>0). This occurs in
the growth phase of baroclinic waves.
They are also the primary mechanism by which the zonal mean
(and, by angular momentum conservation, the meridional mean)
flow is forced (u’v’>0). This occurs in the Rossby wave decay
phase of the baroclinic wave, in which easterly momentum is
transported equatorward.
Global Implications?
Atmospheric eddies are the primary mechanism by which low
latitude HEAT is transported poleward (v’T’>0). This occurs in
the growth phase of baroclinic waves.
They are also the primary mechanism by which the zonal mean
(and, by angular momentum conservation, the meridional mean)
flow is forced (u’v’>0). This occurs in the Rossby wave decay
phase of the baroclinic wave, in which easterly momentum is
transported equatorward.
Global Implications?
More Eddy ActivityStronger Circulation
Eddy activity depends on the meridional temperature
gradients of the climatological background state.
Stronger temperature gradients increase the rate of
eddy formation (can be shown from linear theory).
Eddy activity depends on the meridional temperature
gradients of the climatological background state.
Stronger temperature gradients increase the rate of
eddy formation (can be shown from linear theory).
Meridional temperature gradients were quite different
in the past for a variety of reasons (ice sheets, orbital
parameters, greenhouse gases, etc.).
Eddy activity depends on the meridional temperature
gradients of the climatological background state.
Stronger temperature gradients increase the rate of
eddy formation (can be shown from linear theory).
Meridional temperature gradients were quite different
in the past for a variety of reasons (ice sheets, orbital
parameters, greenhouse gases, etc.).
Also, during an Ice Age, topographic forcing of stationary
waves was very different because of the massive ice sheets.
The numerical experiments
A global coupled atmosphere-ocean general circulation model is
used to simulate 80 years of climate for:
The numerical experiments
A global coupled atmosphere-ocean general circulation model is
used to simulate 80 years of climate for:
1) Last Glacial Maximum (LGM, 21,000 years ago)
-massive continental ice sheets
-decreased atmospheric carbon dioxide
-sea level lowering of 120 m
-surface vegetation different
Schematics of ice sheet extent at
the Last Glacial Maximum
The numerical experiments
A global coupled atmosphere-ocean general circulation model is
used to simulate 80 years of climate for:
1) Last Glacial Maximum (LGM, 21,000 years ago)
-massive continental ice sheets
-decreased atmospheric carbon dioxide
-sea level lowering of 120 m
-surface vegetation different
2) 9,000 years ago
-orbital parameters
-remnants of Laurentide ice sheet
Obliquity was high in the early-mid Holocene (9,000-6,000 years ago).
This accentuates the seasonal cycle; warmer summers and colder winters.
The numerical experiments
A global coupled atmosphere-ocean general circulation model is
used to simulate 80 years of climate for:
1) Last Glacial Maximum (LGM, 21,000 years ago)
-massive continental ice sheets
-decreased atmospheric carbon dioxide
-sea level lowering of 120 m
-surface vegetation different
2) 9,000 years ago
-orbital parameters
-remnants of Laurentide ice sheet
3) 6,000 years ago
-orbital parameters
The numerical experiments
A global coupled atmosphere-ocean general circulation model is
used to simulate 80 years of climate for:
1) Last Glacial Maximum (LGM, 21,000 years ago)
-massive continental ice sheets
-decreased atmospheric carbon dioxide
-sea level lowering of 120 m
-surface vegetation different
2) 9,000 years ago
-orbital parameters
-remnants of Laurentide ice sheet
3) 6,000 years ago
-orbital parameters
4) Today (control)
The numerical experiments
A global coupled atmosphere-ocean general circulation model is
used to simulate 80 years of climate for:
1) Last Glacial Maximum (LGM, 21,000 years ago)
-massive continental ice sheets
-decreased atmospheric carbon dioxide
-sea level lowering of 120 m
-surface vegetation different
2) 9,000 years ago
-orbital parameters
-remnants of Laurentide ice sheet
3) 6,000 years ago
-orbital parameters
4) Today (control)
5) Doubling of atmospheric carbon dioxide (2xCO2)
El Niño and La Niña events are defined by sea surface
temperature anomalies in the Nino 3.4 region. Values
are typically normalized by the standard deviation.
Control simulation produces good statistics for ENSO.
QuickTime™ and a
GIF decompressor
are needed to see this picture.
Control
S.D.=0.83
Observations
S.D.=0.87
Wavelet
analysis
Power
Spectrum:
Averaged in time:
Observations
Control
S.D.=0.83
Observations
S.D.=0.87
Values
Normalized by
Their standard
Deviation
LGM
9,000 B.P.
6,000 B.P.
Control
Observations
2xCO2
Values
Normalized by
Their standard
Deviation
LGM
9,000 B.P.
6,000 B.P.
Control
Observations
2xCO2
Values
Normalized by
Their standard
Deviation
LGM
9,000 B.P.
6,000 B.P.
Control
Observations
2xCO2
Values
Normalized by
Their standard
Deviation
LGM
9,000 B.P.
6,000 B.P.
Control
Observations
2xCO2
Values NOT
Normalized by
The standard
Deviation
S.D.=0.58
S.D.=0.55
S.D.=0.81
S.D.=0.83
Increasing
Period of
ENSO
(decrease in
Frequency)
S.D.=0.87
S.D.=1.05
Changes in the climatological mean states
Not much change
in mean thermocline
depth (except for
CO2 case)
Change in east-west tilt
of thermocline consistent
with change in strength
of mean atmospheric
trade winds
Changes in the climatological mean states
Not much change
in mean thermocline
depth (except for
CO2 case)
Change in east-west tilt
of thermocline consistent
with change in strength
of mean atmospheric
trade winds
~20% reduction in
Easterly trade winds from LGM to today
Linear Stability Analysis of the coupled atmosphere-ocean
system (Fedorov and Philander, Science, 2001)
PERIOD
Control
(increases with
decreasing wind
speed)
Growth
Rate
Linear Stability Analysis of the coupled atmosphere-ocean
system (Fedorov and Philander, Science, 2001)
PERIOD
(increases with
decreasing wind
speed)
Control
Growth
Rate
* LGM
* 9,000
* 6,000
*2xCO2
Is there a reason why the strength of the atmospheric winds
should be different in these simulations?
Is there a reason why the strength of the atmospheric winds
should be different in these simulations?
Yes. The eddy fields are very different.
Is there a reason why the strength of the atmospheric winds
should be different in these simulations?
Yes. The eddy fields are very different.
LGM: enhanced meridional temperature gradient (transient eddies)
presence of massive continental ice sheets (stationary eddies)
Is there a reason why the strength of the atmospheric winds
should be different in these simulations?
Yes. The eddy fields are very different.
LGM: enhanced meridional temperature gradient (transient eddies)
presence of massive continental ice sheets (stationary eddies)
9,000 and 6,000 B.P.: more seasonal climate because of enhanced
obliquity of the planet and summertime perihelion. Effect is
stronger at 9,000 B.P. than 6,000 B.P. Hotter summers and
colder winters should produce more wintertime transient
eddy activity.
Is there a reason why the strength of the atmospheric winds
should be different in these simulations?
Yes. The eddy fields are very different.
LGM: enhanced meridional temperature gradient (transient eddies)
presence of massive continental ice sheets (stationary eddies)
9,000 and 6,000 B.P.: more seasonal climate because of enhanced
obliquity of the planet and summertime perihelion. Effect is
stronger at 9,000 B.P. than 6,000 B.P. Hotter summers and
colder winters should produce more wintertime transient
eddy activity.
CO2: reduced meridional temperature gradient (transient eddies)
The Eliassen-Palm (E-P) flux vector:
Equatorward momentum flux
(u’v’>0)
Z
Poleward heat flux
(v’T’>0)
E-P flux
Y
Increased midlatitude
eddies drive stronger
subtropical subsidence,
a stronger Hadley
cell and, through
angular momentum
conservation,
stronger equatorial
easterlies.
Peak easterly (i.e. negative)
winds over the Pacific
occur near the solstices of
end-December and end-June
when baroclinic eddy activity
is greatest
Zonally symmetric
atmosphere-only
integration for a
land-covered planet
--wind changes are not
related to the Asian
monsoon.
Coupled model
TAO winds
Changing
Teleconnection
Patterns:
Temperature
LGM
9K BP
6K BP
Today
2xCO2
Changing
Teleconnection
Patterns:
Precipitation
LGM
9K BP
6K BP
Today
2xCO2
Implications of changing teleconnection patterns:
Implications of changing teleconnection patterns:
Caution must be used when interpreting paleoclimate proxy records
for ENSO. For example, Rodbell et al (1999) interpreted the
absence of a distinct ENSO signal from early Holocene sediments of
coastal Peru to mean that ENSO was absent between 15,000 and
~6,000 years ago. This assumed “stationarity” of the teleconnection
pattern is incorrect.
Implications of changing teleconnection patterns:
Caution must be used when interpreting paleoclimate proxy records
for ENSO. For example, Rodbell et al (1999) interpreted the
absence of a distinct ENSO signal from early Holocene sediments of
coastal Peru to mean that ENSO was absent between 15,000 and
~6,000 years ago. This assumed “stationarity” of the teleconnection
pattern is incorrect.
Also, Koutavas et al (2002) interpreted a reduced zonal SST gradient
in the tropical Pacific at the LGM to mean the glacial climate was in
an El Niño state. This assumption does not take into account the
changed mean state of the LGM climate.
Implications of changing teleconnection patterns:
Caution must be used when interpreting paleoclimate proxy records
for ENSO. For example, Rodbell et al (1999) interpreted the
absence of a distinct ENSO signal from early Holocene sediments of
coastal Peru to mean that ENSO was absent between 15,000 and
~6,000 years ago. This assumed “stationarity” of the teleconnection
pattern is incorrect.
Also, Koutavas et al (2002) interpreted a reduced zonal SST gradient
in the tropical Pacific at the LGM to mean the glacial climate was in
an El Niño state. This assumption does not take into account the
changed mean state of the LGM climate.
Teleconnections produce high latitude fingerprints of ENSO changes.
Conclusions
The period of ENSO increases from the LGM to today
Conclusions
The period of ENSO increases from the LGM to today
The amplitude of ENSO increases from the LGM to today
Conclusions
The period of ENSO increases from the LGM to today
The amplitude of ENSO increases from the LGM to today
These changes are consistent with the decrease in
strength of the climatological easterly trade winds over
the Pacific
Conclusions
The period of ENSO increases from the LGM to today
The amplitude of ENSO increases from the LGM to today
These changes are consistent with the decrease in
strength of the climatological easterly trade winds over
the Pacific
Decrease in trade wind strength is consistent with the
decrease in midlatitude eddy activity
Conclusions
The period of ENSO increases from the LGM to today
The amplitude of ENSO increases from the LGM to today
These changes are consistent with the decrease in
strength of the climatological easterly trade winds over
the Pacific
Decrease in trade wind strength is consistent with the
decrease in midlatitude eddy activity
Decrease in eddy activity related to topographic forcing
at the LGM, to orbital forcing at 9,000 and 6,000 B.P., and
to radiative forcing in the 2xCO2 environment
What’s been observed?
Increased tropical Pacific wind speeds during the LGM
(aeolian deposits, upwelling indices from ocean cores)
What’s been observed?
Increased tropical Pacific wind speeds during the LGM
(aeolian deposits, upwelling indices from ocean cores)
Increased tilt of the tropical Pacific thermocline during
the LGM (planktonic foraminifera)
What’s been observed?
Increased tropical Pacific wind speeds during the LGM
(aeolian deposits, upwelling indices from ocean cores)
Increased tilt of the tropical Pacific thermocline during
the LGM (planktonic foraminifera)
Increase in ENSO amplitude and period from LGM to today
from corals (Tudhope et al, 2001)
What’s been observed?
Increased tropical Pacific wind speeds during the LGM
(aeolian deposits, upwelling indices from ocean cores)
Increased tilt of the tropical Pacific thermocline during
the LGM (planktonic foraminifera)
Increase in ENSO amplitude and period from LGM to today
from corals (Tudhope et al, 2001)
Decrease in easterly trade wind strength (Nature, May 2006)
With many thanks to…
Canadian Foundation for Climate and
Atmospheric Research (CFCAS; Polar
Climate Stability Network)
With many thanks to…
Canadian Foundation for Climate and
Atmospheric Research (CFCAS; Polar
Climate Stability Network)
…and all of you!
Wavelet analysis
Power Spectrum:
Averaged in time:
Seasonal Cycle
Seasonal Cycle
QBO?
Seasonal Cycle
QBO?
Decadal Variability
Seasonal Cycle
QBO?
ENSO
Decadal Variability
Changing teleconnection patterns: composite temp. anomalies
Control
6,000 B.P.
9,000 B.P.
LGM