Quantifying the Sources of

1

Quantifying the Sources of Interannual to Decadal
SST Anomalies An Overview
Clara Deser (With thanks to Michael Alexander,
NOAA/CDC)
AGU Ocean Sciences 2004
2
OUTLINE

• Overview of Key Physical Processes
• Relevant mechanisms for observed SST variability
  in the North Pacific, including the Kuroshio
  Extension System
• Contribution of the North Atlantic Thermohaline
  circulation
• Final Remarks

3
SST Variability Processes
ATMOSPHERE
MIXED LAYER Tm
H
OCEAN
4
SST Variability Atmospheric Processes
ATMOSPHERE
MIXED LAYER Tm
H
OCEAN
5
SST Variability Oceanic Processes
ATMOSPHERE
MIXED LAYER Tm
H
OCEAN
6
SST Variability Oceanic Processes
ATMOSPHERE
Qnet
Mixed Layer Tm
H
Dynamical Ocean (Main Thermocline)
1 km
?Tm
7
Entrainment Velocity (We dH/dt)

• Atmospheric contribution
• local changes in H due to wind stirring
• and convective cooling
• Oceanic contribution
• - Ekman Pumping due to wind stress curl
• - Ug H (lateral induction due to
  spatial
• gradients in H)

8
SST Variability Processes
ATMOSPHERE
Qnet
Mixed Layer Tm
H
Dynamical Ocean (Main Thermocline)
1 km
?Tm
9
Which processes are important in nature?
Qnet
(UEk Ug )
?Tm

• Extratropics
• Surface fluxes (Qnet)
• Weather forcing of a slab ocean
• Mixed Layer Processes (H and WE)
• Entrainment and the seasonal cycle of mixed layer
  depth
• Ocean Dynamics (UekUg, Tbelow)
• Wind-driven gyre circulation (Kuroshio Current
  System)
• Thermohaline circulation (North Atlantic)

10
Local Correlation (SST tendency, Qlatentsensible
)(see also Cayan, 1992 and Iwasaka and Wallace
1995)
Q drives dSST/dt
Q damps dSST/dt

• Qls drives 20-60 of winter SST tendencies

11
Weather forcing of a slab oceanthe simple
stochastic climate model (Frankignoul and
Hasselmann Tellus, 1977) The null hypothesis
for extratropical SST variability(random Qnet
due to internal atmospheric variabilityacting on
a slab ocean mixed layer)
linear damping l 10 - 20 Wm-2 K-1
No preferred time scale beyond 1-2 weeks
12
Simple Stochastic Climate Model
decorrelation time 5 mo
1 yr lag autocorr .1
decorrelation time 4 yrs
1 yr lag autocorr .8
13
Application of the Simple Stochastic Climate
Model Observed SST Lag Autocorrelations in the
North Pacific (35-45N, 155E-170W) Deser et al.
(J. Climate, 2003)
Simple stochastic climate model
Observed SST
Decorrelation time 4 months
14
Application of the Simple Stochastic Climate
Model Observed SST Lag Autocorrelations in the
North Pacific (35-45N, 155E-170W) Deser et al.
(J. Climate, 2003)
Simple stochastic climate model
Observed SST
Yr 0
Yr 2
Yr 1
2 years for winter SST
Decorrelation time 4 months
15
IDEALIZED MIXED LAYER OCEANS
Entraining
H
H(t)
Summer
Winter
Winter
Re-emergence Mechanism (Alexander, Deser and
Timlin, J.Climate, 1999)
16
Re-emergence in three North Pacific regions
Lag regression between SST anomalies in April-May
with monthly temperature anomalies as a function
of depth.
Regions
Alexander et al. (1999, J. Climate)
17
ENTRAINING STOCHASTIC CLIMATE MODEL
Entraining

if deepening 0 otherwise
Specified monthly from observed (Levitus)
climatology
T
Tb
5000 yr integration, 3 day timestep
H
Deser et al. (J. Climate, 2003)
18
Application of the Entraining Stochastic Climate
Model to Observed SST Lag Autocorrelations in the
North Pacific (35-45N, 155E-170W) Deser et al.
(J. Climate, 2003)
Heat Content to base of winter ML
Entraining stochastic climate model
Observed SST
Simple stochastic climate model with obs winter H
See DeCoetlogon andFrankignoul (2003) for North
Atlantic
19
The Entraining Stochastic Climate Model
Relevance for the Pacific Decadal Oscillation

• North Pacific ocean mixed layer integrates Qnet
  anomalies due to atmospheric circulation changes
  forced by ENSO
• Re-emergence brings back ENSO-induced SST
  anomalies in succeeding winters
• (no summer/fall PDO)
• PDO acts as a lowpass filter of ENSO
• (Newman et al., J. Climate, 2004)

20
Atmospheric Circulation Anomalies Forced by ENSO
Observations
DJF SLP Contour (1 mb)
FMA SST (shaded ºC)
Model AGCM-MLM
Specified SST
Alexander et al. J. Climate, 2002
21
Forecast PDO vs. Observed PDO (Winter)
a0.35 b0.69
Re-emergence
Atmospheric Bridge
(H125m)
ENSO Index (Deser et al., 2004)
Observed PDO
Correlation 0.71
22
Or in frequency space
Observed PDO
Model PDO
ENSO Index Deser et al. 2004
23
Beyond mixed layer processes The role of ocean
dynamics in SST variability

• Two Examples
• Kuroshio Extension System
• North Atlantic Thermohaline Circulation

24
Kuroshio Extension System

• Qiu (Jap. J. Oceanog., 2002) Mixed layer heat
  budget using Topex/Poseidon altimetry data for
  geostrophic currents.
• Kelly (J. Climate, 2004) Lateral ocean heat
  transport convergence vs. surface fluxes
• Schneider and Miller (J. Climate 2001)
  First-mode baroclinic Rossby waves

25
Winter Mixed Layer Heat Budget for Kuroshio
Extension Region
(31-37N, 140-180E)
Qiu (2002)
Temperature tendency
Correlation 0.38
Qnet Entrainment Ekman advection 1D ML
processes
Geostrophic advection
Note slow time scale! (from wind stress
curl farther east 6 yrs earlier)
Sum of forcing terms
Correlation 0.80 with T tendency
26
Similar Result from Kelly (J. Climate, 2004)
ATMOSPHERE
Kuroshio Extension Region (26-40N,
140E-180) 1970-2000
Qnet
MIXED LAYER
Flateral
OCEAN
300m
Lateral Ocean Heat Transport Convergence
contributes more to changes in Heat Storage Rate
(0-300m) than do surface fluxes on timescales
longer than a couple of years
27
Correlation with (0-300m) Heat Storage Rate
(Kelly, 2004)
Kuroshio Extension Region (140E-180)
Monthly data
gt 2 years
28
Role of Wind-generated 1st Mode Baroclinic Rossby
Waves
Atmosphere
Ocean Mixed Layer
Thermocline
West
East
29
Role of Rossby Waves for SSTVariability in the
Kuroshio ExtensionSchneider and Miller (J.
Climate, 2001)
Correlation between observed and hindcast winter
SST
0.6
Forecast equation for SST from linear Rossby
wave equation forced by observed wind stress curl
(Ekman pumping) during 1948-2000, and assuming
thermocline depth (T) anomaly is entrained into
the deep winter mixed layer.
30
North Atlantic Thermohaline Circulation

• Prominent cooling in northern North Atlantic
  during the 1960s and 1970s and warming during the
  1920s and 1930s likely due to a lagged response
  of oceanic heat transports associated with
  changes in the THC forced by NAO-related surface
  heat flux changes 10-20 years earlier. (Eden and
  Jung, J. Climate, 2001 additional references in
  Visbeck et al., AGU Monograph, 2003).

31
Final Remarks
32
Which processes are important in nature?Evidence
for All
Qnet
(UEk Ug )
?Tm

• Surface fluxes (Qnet)
• Weather forcing of a slab ocean
• Mixed Layer Processes (H and WE)
• Entrainment and the seasonal cycle of mixed layer
  depth
• Ocean Dynamics (UekUg, Tbelow)
• Wind-driven gyre circulation (Kuroshio Current
  System)
• Thermohaline circulation (North Atlantic)

33
The End
34
Summary of Processes

• Mean seasonal cycle and mixed layer physics
• Reemergence
• Atmospheric Bridge
• Cause and effect well understood
• Tropical Pacific gt Global SSTs
• Influence of air-sea feedback on extratropical
  atmosphere complex
• PDO (1st EOF of North Pacific SST)
• Ocean integration of noise Reemergence
  Atmospheric Bridge Extratropical ocean dynamics
  (?)
• Much of predictable signal likely coming from the
  tropics
• Extratropical ocean integrates (reddens) ENSO
  signal
• Decadal variability in tropics impact
  atmosphere ocean
• Unrealistic strong positive extratropical
  air-sea feedback
• Other Processes/modes of variability
• Ocean currents Rossby waves in western N.
  Pacific
• Some evidence for extratropical PDV
• Changes in the Thermohaline Circulation

35
Forecast of Annual Mean Anomalies PDO vs.
observed PDO
Re-emergence
Atmospheric Bridge
Correlation 0.74
a0.58 b0.58
Newman et al. 2003, J. Climate
36
(No Transcript)
37
Comments on theories for thePacific Decadal
Oscillation (Variability)

• Midlatitude Ocean Dynamics
• with positive-feedback atmospheric response
  (Latif and Barnett, J. Climate, 1996) unlikely
  (see Schneider et al., J. Climate, 2002)
• with white-noise atmospheric forcing (Qiu, J.
  Phys. Oceanog., 2003)
• Extratropical-Tropical Connections
• Oceanic subduction of T plus atmospheric bridge
  (Gu and Philander, Science, 1996) unlikely (see
  Schneider et al., GRL, 1999)

38
Comments on theories for thePacific Decadal
Oscillation (Variability)

• Midlatitude Ocean Gyre Circulation Dynamics
• with positive-feedback atmospheric response
  (Latif and Barnett, J. Climate, 1996) unlikely
  (see Schneider et al., J. Climate, 2002)
• with white-noise atmospheric forcing (Qiu, J.
  Phys. Oceanog., 2003)
• Extratropical-Tropical Connections
• Oceanic subduction of T plus atmospheric bridge
  (Gu and Philander, Science, 1996) unlikely (see
  Schneider et al., GRL, 1999)
  Meridional
  Subtropical Cells plus atmospheric bridge
  (Kleeman et al., 1999 McPhaden and Zhang, 2002)
• ENSO Atmospheric Bridge Re-emergence (ocean
  mixed layer processes) (Newman et al., J.
  Climate, 2004)
• ENSO Atmospheric Bridge Re-emergence Midlat
  Ocean Gyre Circulation Response

39
An Example of the Role of Qnet EOF 1 Winter SST
Anomalies, 1949-1999 (Kushnir et al., 2002)
Color SST Contours Regressed Qnet ( out of
ocean) Arrows Surface winds
ATLANTIC
PACIFIC
See also Cayan (1992 J. Climate and J. Phys.
Oceanog.) Frankignoul et al. (1998 J.
Climate), and many others
40
Kuroshio/Oyashio Extension (40N, 140-170E)
Qnet
SST
Schneider et al. (J. Climate, 2002)
Qnet damps SST
41
The Atmospheric Bridge
Relevance for the Pacific Decadal Oscillation
PDO ENSO Atmospheric Bridge
Re-emergence (Newman et al., J. Climate, 2004)
/
(Alexander 1992 Lau and Nath 1996 Alexander et
al. 2002 all J. Climate)
42
Local Correlation (SST tendency, Q
latentsensible )(see also Cayan, 1992 and
Iwasaka and Wallace 1995)
8 year
Q drives dSST/dt
Q damps dSST/dt

• Qls drives 20-60 of winter SST tendencies