CoRP Symposium, Fort Collins, CO
Coupled Interactions Between Surface
Winds and Sea Surface Temperature
Larry W. O’Neill
Dudley B. Chelton, Antonio Fetter,
Steve Esbensen, Ricardo Matano, Nicolai Thum
College of Oceanic and Atmospheric Sciences
Oregon State University, Corvallis, Oregon
Mean QuikSCAT Wind Speed and AMSR-E SST
Colors are 1-yr average QuikSCAT wind speed and
contours are 1-yr average AMSR-E SST for the year 2003
spatially high-pass filtered to remove large-scale wind
and SST variations
Goals of this study:
-Characterize the
interaction between SST
and surface winds from
global, near allweather satellite
observations
-SST influences on wind
speed, curl and
divergence fields
-Investigate feedbacks
onto the ocean using
an ocean GCM over
the Southern Ocean
Cold water intrusions decelerate surface winds while
warm water intrusions accelerate surface winds
Background
-Over basin scales (1000’s of
kms), surface winds and SST
are negatively correlated as
atmospheric circulation
anomalies cause ocean
warming/cooling through
evaporation and ocean
mixing (e.g. Mantua et al.
1997; Okumura et al. 2001)
-The processes described
here, however, occur on
smaller scales (about 50-1000
km)
- At these scales, small-scale
SST perturbations along ocean
frontal zones alter the
turbulent stress divergence
(through changing vertical
turbulent mixing and
boundary layer height) and
surface hydrostatic pressure
gradients
Ultimately, small-scale wind and SST
perturbations are positively correlated
Satellite observations indicate that coupling
is prevalent everywhere in the World’s oceans
where significant SST fronts exist
Small-Scale Coupled Ocean-Atmosphere Interactions
Small-scale spatial
SST anomalies
Ocean Dynamics
?
Observable by
satellite
Small-scale
surface wind
stress perturbations
~
Changes in surface
heat fluxes
Modification of
atmospheric boundary
layer stability and
hydrostatic pressure
Satellite Data Description
• Vector winds measured by the SeaWinds
scatterometer on the QuikSCAT satellite
• Active microwave imager
• 25km resolution, gridded at 1/4 degree
• Accuracy of individual swath measurements of wind
speed: 1.7 m/s (Chelton and Freilich, 2005)
• SST from the Advanced Microwave Scanning
Radiometer on the EOS-Aqua satellite (AMSR-E)
• Passive microwave
• 56km resolution, gridded at 1/4 degree
• Accuracy of about 0.4°C (Chelton and Wentz, 2005)
• Microwave radiometers allow measurement of
ocean surface through non-precipitating clouds
• important over mid-latitudes, where clouds mask ~70% of
mid-latitude ocean surface annually
Statistical Wind Speed Response to SST
Spatially High-Pass Filtered QuikSCAT wind
speed bin-averaged as a function of the
AMSR-E SST
-Rule of thumb: about
0.3 to 0.4 m/s change
in wind speed per °C
change in SST
-Regional differences
in slopes due to
detailed regional
differences in
boundary layer
structure (possibly
including large-scale
wind speed and BL
depth)
-Errorbars are a
measure of the
variability within each
bin
Wind Stress and SST over the Southern Ocean
Agulhas Retroflection
Agulhas Return Current
Brazil Current
New Zealand
Malvinas Current
Drake Passage
Ice
1-yr average for 2003
Wind stress and SST over the Southern Ocean
Over weekly and longer time-scales, correlation between smallscale wind stress and SST perturbations is >0.8
Curl and Divergence Formation near SST Fronts
Small-scale curl and divergence perturbations
form as a result of cross-frontal changes in
wind speed
Cross-frontal changes in wind direction are also
important, but ignored here
Small-scale SST influence on curl and divergence
Surface wind vectors
A similar model
holds for Cold SST
“blobs”
Warm SST “blob”
Curl perturbations should depend on crosswind SST gradient
perturbations
Divergence perturbations should depend on downwind SST gradient
perturbations
Statistical assessment of curl and divergence
Curl vs. crosswind SST gradient (black points)
Divergence vs. downwind SST gradient (grey points)
-Fields spatially high-pass filtered to
isolate effects of small-scale SST
perturbations
-Slopes are different! Curl response
to crosswind SST gradients about 1/3
to 1/2 smaller than divergence
response to downwind SST gradients
-Difference in response due to
SST-induced changes in wind
direction (not discussed here)
-Slopes exhibit seasonal variations;
larger in wintertime
Comparison between models and observations
-Curl and divergence dependence between crosswind and downwind
SST gradients provides a rigorous means to validate atmospheric
circulation model performance
-Atmospheric mesoscale models are needed to diagnose upper-air
boundary layer dynamics (i.e. kinematic and thermodynamic structure at
sufficient vertical resolution)
QuikSCAT/AMSR-E
WRF model
observations
-Model run over a large
portion of the Agulhas
Return Current using a 1month AMSR-E SST
composite by Nicolai
Thum and Steve
Esbensen at OSU
-Model agrees very well
with observations,
engendering confidence
that BL vertical structure
is well-simulated
Coupled ocean response to small-scale wind
• Use Modular Ocean Model (MOM_2) to investigate the ocean
response to small-scale SST perturbations (Antonio Fetter and
Ricardo Matano at OSU)
• Two runs are analyzed using 5-yr average (1999-2004)
QuikSCAT wind stress field:
• One forced with spatially-smoothed winds (denoted as
SMOOTHED case)
• One forced with unsmoothed winds (denoted as UNSMOOTHED
case)
• Analyzed 5 yrs of simulation after a 10 yr spin-up
• Goal is to identify and describe upper ocean circulation
differences between the two models using otherwise realistic
boundary conditions and topography over the Southern
Ocean
• The only difference between the SMOOTHED and
UNSMOOTHED model runs is the wind stress forcing
• Analysis is ongoing
Wind Forcing Comparison
-Notice small-scale wind variability – mostly caused by small-scale
SST perturbations associated with meanders in ocean currents
- Smoothing chosen to approximate smoothness of NCEP and ECMWF
reanalyses fields
Zonal Wind Stress Component
Wind Stress Curl
Spatially
smoothed
winds
Unsmoothed
winds
Black line is along the zero wind
stress curl line
Transport Comparison for Agulhas Return Current
-Meanders almost completely
straightened out in UNSMOOTHED
case
-Zonal transport into the western
edge is reduced in the
UNSMOOTHED case
-Horizontal current structure east of
60ºE (over the Kerguelen Plateau)
is very different
-Circulation is surprisingly very
different!
Vectors represent barotropic transport;
pink numbers represent transport through
that side of box (Sv = Sverdrup = 106
m3/s); grey shading is water depth
Meridional-Depth Transect at 55ºE
-Frontal zone further south in
SMOOTHED case
-Subtropical gyre fronts very
different, as frontal zone
extends deeper and further
south in UNSMOOTHED case
Meridional-Depth Transect at 55ºE
-Agulhas Return Current is
stronger and narrower in
UNSMOOTHED case and
is shifted northward by
~1° lat
-More variability north of
current suggesting more
eddy variability in
UNSMOOTHED case
-Current may be more
baroclinically unstable in
SMOOTHED case and
more barotropically
unstable in UNSMOOTHED
case
Mean Zonal Velocity
Zonal Velocity Std Dev
Horizontal Momentum Fluxes
Velocity variance ellipses are
much larger in the
UNSMOOTHED case, indicating
more eddy momentum flux out
of the current
Surface velocity variance
ellipses, surface current velocity
(black curves), water depth
(grey shading)
Meridional Transient Heat Flux
-More meanders in the SMOOTHED
case, and the northward shift in the
current, cause a huge difference in
transient heat flux
-Factor of 1.3 difference in transient
heat flux between SMOOTHED and
UNSMOOTHED model runs
-Heat transport differences over this
part of the current are partly
balanced by heat transport by the
mean flow and by transport outside
of this transect
Summary
• Small-scale SST perturbations can cause 2-4 m/s
change in surface wind speed between cold and
warm SST perturbations
• Equivalent to an ~30% change in surface wind stress
• Resulting SST-induced horizontal gradients in wind
speed create large perturbations in wind stress curl
and divergence
• Resulting feedback onto the ocean is significant
• Ocean circulation strongly influenced by smallscale SST-induced wind stress perturbations
• Influences transient heat and momentum transport
• Results in different horizontal and vertical structure and
strength of Agulhas Return Current
Conclusions and Implications
• Small-scale wind structure, caused by smallscale SST perturbations, significantly
influences ocean circulation (surprising!)
• Ocean models typically forced by coarse spatial
resolution NWP analyses wind stress fields
• Coupled feedbacks are likely important
• Implications for ocean modelling and
coupled climate simulations
• Small-scale surface wind and SST structures are
important components
• Small-scale SST fields needed for weather
forecast models
Curl and crosswind SST gradients
Spatially High-Pass Filtered Wind and SST
Wind Speed
Perturbations
(colors) and SST
perturbations
(contours)
Spatial high-pass
filter removes
large-scale wind
variability
unrelated to SSTinduced wind
perturbations
Surface winds accelerate over warm SST perturbations and
decelerate over cool SST perturbations
Small-scale perturbations in the time mean wind field
mostly determined by small-scale SST perturbations