Part I. Mike McPhaden--What We Know and What

is Unresolved: An Observational Perspective

Part II. David Battisti--What We Know and What is Unresolved: A Theoretical and Modeling Perspective

Sleeping Lady Mountain Retreat

Leavenworth, WA

19 September 2005

ENSO Dynamics and Global Impacts(“A Benign Problem”)

Sleeping Lady Mountain Retreat

Leavenworth, WA

19 September 2005

Part I. What We Know and What is Unresolved:

An Observational Perspective

1997-98 El Niño: Global Impacts

Fatalities: 23,000Economic Losses: $36 Billion

El Niño is often followed by or preceded by La Niña: an unusual cooling of the tropical Pacific

Upwelling zones

Western Pacific “warm pool”

El Niño happens roughly every 2-7 years, lasts 12-18 months, and peaks at the end of the calendar year

Anomalous

Geostrophic

Divergen

ce

Every few years, the trade winds weaken…

Anomalous

Geostrophic

Convergen

ce

Ocean-Atmosphere Feedback Loops During El Niño

Winds SST

Fast Positive Feedback Warms

Ocean-Atmosphere Feedback Loops During El Niño

Winds SST

Fast Positive Feedback Warms

Thermocline Depth

Slow Negative Feedback Cools

Ocean-Atmosphere Feedback Loops During La Niña

Winds SST

Fast Positive Feedback Cools

Thermocline Depth

Slow Negative Feedback Warms

NINO3.4 and SOI, 1980-2005

NINO-3.4

Darwin Tahiti

NINO3.4 and SOI, 1980-2005

Darwin Tahiti

NINO-3.4

SOI and NINO3.4 Correlation=-0.9 (maximum @zero lag)

NINO3.4 and SOI, 1980-2005

Darwin Tahiti

NINO-3.4

El Niño/Southern Oscillation (ENSO):Warm phase (El Niño) // Cold Phase (La Niña)

Thermocline Depth (20°C)

Thermocline Depth (20°C)

Thermocline Depth (20°C)

Build up of excess heat content along equator is a necessary precondition for El Niño to occur.

The time between El Niños is determined by the time to recharge.

El Niño purges excess heat to higher latitudes, which terminates the event.

Upper Ocean Heat Content and El Niño(Recharge Oscillator Theory*)

*Wyrtki, 1985; Cane et al, 1986; Jin, 1997

Niño3.4 SST

Janowiak et al (2003) rainfall & ERS wind

velocity

Reynolds et al (2003) SST & ERS wind

stress

Peak Phase, 1997

Peak Phase, 2004

Janowiak et al (2003) rainfall & Quikscat

wind velocity relative to ERS climatology

Reynolds et al (2003) SST & Quikscat wind stress relative to ERS

climatology

DRY

Processes Affecting Equatorial SST

Enhanced Surface

Heat Fluxes

Zonal Advection

Suppressed Upwelling

Atmospheric Circulation Changes During El Niño

Changes in tropical rainfall patterns affect the global atmospheric circulation via “teleconnections”

Heavy rain

Pacific-North American (PNA) Pressure Pattern

Subtropical Jet Stream in NH Splits, Southern Branch Shifts

South and Intensifies

Impacts on Global Weather PatternsEl Niño shifts the probability of droughts, floods, heat waves, and

extreme weather events in large regions of the globe.

El Niño tends to suppress formation of Atlantic hurricanes.

El Niño tends to increase intensity and geographic range of Pacific hurricanes.

Opposite tendencies occur during La Niña.

Impacts on Tropical Storms

El Niño shifts the probability of droughts, floods, heat waves, and extreme weather events in large regions of the globe.

Magnitude of impacts scales with magnitude of Pacific SST anomalies

La Niña impacts roughly opposite to those of El Niño

Impacts on Global Weather Patterns

El Niño shifts the probability of droughts, floods, heat waves, and extreme weather events in large regions of the globe.

Washington State:

For 9 El Niños between 1941-1995 Expected by Chance

6 warm winters (~2°F increase) 3

2 neutral winters 3

1 cold winter 3

Impacts on Global Weather Patterns

Social and Economic Consequences

El Niño can affect life, property, and economic vitality due to weather related hazards.

1997-98 El Niño: Global Impacts

Fatalities: 23,000Economic Losses: $36 Billion

1997-98 El Niño: U.S. Impacts

• Negative– 189 Fatalities– $4-5 billion in economic losses

• Positive– 850 lives saved– $20 billion in economic gains

Weather Noise and ENSO Stability If ENSO is a freely oscillating instability of the ocean-atmosphere

system governed by basin scale dynamics (Schopf & Suarez, 1988: Battisti & Hirst, 1989), weather “noise” is not essential but introduces irregularity.

If ENSO is a stable or weakly damped oscillator, external forcing in the form of weather noise is essential for initiation and development of warm events (Penland & Sardeshmukh, 1995; Moore & Kleeman, 1999; Kessler, 2002).

Mean thermocline depth

Stability characteristics determined by strength of ocean-atmosphere coupling and may vary decadally with changing background conditions (Kirtman and Schopf, 1998; Fedorov and Philander, 2000)

Fedorov & Philander, 2000

A=1980-90s; B=1960-70s

ATLAS Mooring

TAO/TRITON: A U.S./Japan collaboration

Current ConditionsNear normal conditions prevail

Current Conditions

Thermocline slopes down to west because of trade wind forcing.

Cold subsurface (100-200 m) temperature anomalies may indicate trend towards La Niña cooling.

Weather Noise and Stochastic ForcingEpisodic westerly wind

forcing and downwelling Kelvin

wave responses

Weather Noise and Stochastic Forcing

June-July 2004 Westerly Wind Burst

BEFORE:

“It is…likely that ENSO-neutral conditions will continue for the next 3 months (through August 2004).”

NOAA/NCEP10 June 2004

AFTER:

“El Niño conditions are expected to develop during the next 3 months.”

NOAA/NCEP5 August 2004

Westerly Wind Bursts Amplitude and Phase of ENSO

Westerly Wind Bursts (2°N-2°S)

0

2

4

6

8

10

12

1 2 3 4 5 6 7 8

Number of Westerly Events (Anomalies > 2 m/s)Maximum Zonal Wind (m/s)Maximum Zonal Fetch (x1000 km)

1997 200019991998 2001 20032002 2004

Westerly Wind Bursts Amplitude and Phase of ENSO

Westerly Wind Bursts (2°N-2°S)

0

2

4

6

8

10

12

1 2 3 4 5 6 7 8

Number of Westerly Events (Anomalies > 2 m/s)Maximum Zonal Wind (m/s)Maximum Zonal Fetch (x1000 km)

1997 200019991998 2001 20032002 2004

29°C

Stochastic forcing not entirely random

Effects of Westerly Wind Bursts on Equatorial SST

Westerly wind bursts cool the western Pacific, and warm the central and eastern Pacific Ocean via processes similar to those that operate on longer time scales (Shinoda & Hendon, 1998; Zhang, 2001; McPhaden, 2002).

Nonlinear processes can rectify short time scale variations into lower frequency changes (Lukas and Lindstrom, 1991; Kessler et al, 1995; Kessler and Kleeman, 2000; Waliser et al, 2003)

Spatial structure resembles “optimal perturbations” in some coupled models of ENSO (Moore and Kleeman, 1999).

Enhanced Surface

Heat FluxesZonal

Advection

Suppressed Upwelling

Sleeping Lady Mountain Retreat

Leavenworth, WA

19 September 2005

Part II. What We Know and What is Unresolved: A Theoretical and

Modeling Perspective

ENSO and Models/Theory: Resolved Issues• ENSO is the result of coupled atmosphere-ocean physics

in the tropical Pacific.– Ocean models must be forced by Southern Oscillation to produce ‘El Ninos’;– Atmosphere models must be forced by ‘El Nino’ SST to produce the Southern

Oscillation.

• ENSO is a true mode of the coupled system. • ENSO is a strong function of the climatological mean

state.– The annual cycle acts to coordinate the ENSO mode to peak at the end of the

calendar year.

• ENSO is predictable 12 months in advance.• ENSO affects are teleconnected from the tropical Pacific

by well-understood atmosphere and ocean dynamics:– Atmospheric impact is nearly global in extent.

www.atmos.washington.edu/~david/enso_pcc.pdf

Examples from an intermediate atmosphere/ocean model

Some models of the tropical Pacific Atmosphere and Ocean system have realistic

ENSO cycles

The processes that affect SST during an ENSO cycle are time and space dependant

Horizontal advection, vertical mixing, entrainment and surface heat fluxes are all important for ENSO

Nino3

Nino1

SST

Ten

denc

y

Time ( model years)

Basic elements of ENSO in observations and models

The “delayed oscillator physics” and “recharge oscillator” are complimentary toy-model descriptions of the ENSO mode.

Oce

an A

djus

tmen

t

&

Bje

rkne

s

ENSO is a true eigenmode of the coupled atmosphere/ocean system in the Pacific

The Bjerknes Mechanism and the Ocean Dynamics are features of the ENSO mode

Few Global Climate models have realistic ENSOs

First EOF of tropical Pacific SST (all IPCC ‘04 models)

Models w/ realistic ENSO space/time variability

•Global Climate models can produce realistic ENSO variability:

- Eg, the high resolution GFDL tropical Pacific-global atmosphere model (Philander et al 1992); the paleo-CSM.

•Unfortunately, none of the current generation global climate models have realistic ENSO variability. Why?

- lousy mean states in the tropical Pacific (why?); too low resolution in equatorial ocean; etc.

Only one model used in the last IPCC assessment simulated ENSO variability that did not violate the robust observational constraints.

The ENSO mode (cont).

ENSO exists because of the structure in the annual average state of the atmosphere/ocean system in the tropical Pacific, and …

ENSO tends to peak at the end of the calendar year because of theannual cycle in the in the basic state.

Spectrum of the pure ENSO mode

Thompson and Battisti 2001

ENSO as a linear, stochastic system

Results from a linear coupled atmosphere-ocean model forced by white noise

ObservedModel

ENSO peaks at the end of the calendar year because the of the annual cycle in the mean state

Observation

Model

ENSO is Predictable

• Skill depends on knowing where you are on the ENSO mode

• The long-lead time for skillful forecast is due to long period of the ENSO mode

• Presently, empirically based forecast models are at least as skillful as coupled GCMS

ENSO is predictable

Empirical models skillfully predict the state of ENSO 12 months in advance.

Skill 1965-1993Persistence

Empirical Models

•Skill depends on the start month.

Empirical Model

Persistence

These plots are from an empirical model using tropical Pacific SST (1981-present). For this month’s forecast, see www.atmos.washington.edu/~wroberts/ENSO/LIM.html

ENSO is predictable

March Starts September Starts

Why is ENSO irregular?The Seasonal Footprinting Mechanism

•ENSO is variable, in part, because the mode interacts with the annual cycle.

•The Seasonal Footprinting Mechanism accounts for about 1/4 to 1/3 of ENSO variance.

•Other factors responsible for irregularity? - Westerly wind bursts?

Vimont et al. 2001, 02, 03

ENSO: Unresolved Issues

• Is the ENSO mode stable or unstable in the present climate? (consensus - stable)

– If stable, what are the sources of energy for ENSO?– Does stability matter for predictability?

• What is the limit of predictability?– 24 months? 12 months?

• How does ENSO affect the mean state?• Will ENSO change due to Global Warming?• What is the cause of the decadal ENSO-like

variability in the present (1900-2005) climate?

ENSO: Unresolved Issues• The leading pattern of variability in the tropical Pacific has an ENSO

like pattern in SST and atmosphere circulation.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Something new, or the debris that results from averaging over past ENSO events? (see Vimont 2005 if you aren’t already convinced)

ENSO and Models: Resolved Issues• ENSO is the result of coupled atmosphere-ocean physics

in the tropical Pacific.– Ocean models must be forced by Southern Oscillation to produce ‘El Ninos’;– Atmosphere models must be force by ‘El Nino’ SST to produce the Southern

Oscillation.

• ENSO is a true mode of the coupled system. • ENSO is a strong function of the climatological mean

state.– The annual cycle acts to coordinate the ENSO mode to peak at the end of the

calendar year.

• ENSO is predictable 12 months in advance.• ENSO affects are teleconnected from the tropical Pacific

by well-understood atmosphere and ocean dynamics:– Atmospheric impact is nearly global in extent.

www.atmos.washington.edu/~david/enso_pcc.pdf

ENSO: Unresolved Issues

• Is the ENSO mode stable or unstable in the present climate? (consensus - stable)

– If stable, what are the sources of energy for ENSO?– Does stability matter for predictability?

• What is the limit of predictability?– 24 months? 12 months?

• How does ENSO affect the mean state?• Will ENSO change due to Global Warming?• What is the cause of the decadal ENSO-like

variability in the present (1900-2005) climate?

Extratropical forcing of ENSO

•Skill depends on the start month.

Empirical Model

PersistenceMar

ch S

tart

sS

epte

mbe

r S

tart

s

These plots are from an empirical model using tropical Pacific SST (1981-present).

For this month’s forecast, see www.atmos.washington.edu/~wroberts/ENSO/LIM.html

ENSO is predictable