mid-depth circulation of the world ocean: a first look at the argo array josh k. willis and...

Mid-depth Circulation of the Mid-depth Circulation of the World Ocean:World Ocean:

A First Look at the Argo A First Look at the Argo ArrayArray

Josh K. Willis and Lee-Lueng Josh K. Willis and Lee-Lueng FuFu

jwillis@caltech.edujwillis@caltech.edu, , llf@pacific.jpl.nasa.govllf@pacific.jpl.nasa.govJet Propulsion Laboratory, Pasadena, CA 91109Jet Propulsion Laboratory, Pasadena, CA 911092005 AGU fall meeting2005 AGU fall meeting

San Francisco, CASan Francisco, CADec. 5-9, 2005Dec. 5-9, 2005

OverviewOverview

I.I. DataDataI.I. The Argo Array The Argo Array

II.II. Satellite sea-surface heightSatellite sea-surface height

II.II. Combining float displacements with Combining float displacements with altimeter dataaltimeter data

III.III. Maps of general circulation at Maps of general circulation at 1000m1000m

The Argo ArrayThe Argo Array

About 166,000 float displacements provide about 4300 float-years of

data

Data Hurdles: surface drift correction

Extrapolation of surface and dive positions

Technique from Davis et al., JAOT, 1992

Position error of O(5 km) implies Position error of O(5 km) implies < 1 cm/s error in subsurface < 1 cm/s error in subsurface velocityvelocity

Data Hurdles: different parking depths

For floats that park at depths near 500m, 1500m or 2000m, velocities calculated using geostrophic shear from WOA01 were added to displacements prior to processing.

Satellite Sea Surface height Satellite Sea Surface height datadata

•Mapped AVISO product•2000-2005 mean removed

Sea surface height anomaly, Jan 5, 2005

Is SSH complementary to Is SSH complementary to float displacement data?float displacement data?

If so, can we use it to If so, can we use it to reduce eddy-noise in the reduce eddy-noise in the

estimate of mean estimate of mean circulation?circulation?

Relating SSH anomalies to subsurface float displacements

From From Roemmich and Gilson, Roemmich and Gilson, JPOJPO, , 20012001

For eddies in the N. Pacific anomalous velocity decreases with depth

Because SSH anomalies reflect surface velocity, they should be scaled-down for comparison with subsurface displacements

“Pseudo-displacements” can be calculated by advecting a particle with anomalous geostrophic

surface velocity

Relating SSH anomalies to subsurface float displacements

Through the geostrophic relation:

SSH anomaly implies anomalous geostrophic velocity at the surface.

,y

gfu

Pseudo-displacement

“pseudo-displacements” from SSHA geostrophic velocity vs. actual displacements

Relating SSH anomalies to subsurface float displacements

Note that slope < 1

Displacements from a few floats in the N. Pacific

From the slope we obtain

α: the scale factor btwn. surface and subsurface velocity

Relating SSH anomalies to float displacements

The scale factor for converting surface velocities anomalies to subsurface velocity is found by minimizing displacement variance in 10 x 10 deg. squares

Using scale factor and AVISO, we compute anomalous velocity at depth and subtract them from actual float displacements to get improved estimate of mean flow

Examples for a few floats in the Southern Ocean

Blue arrows are raw float dataRed arrows are ‘corrected’

Contours are cm of 1000/2000m steric height from WOA01

To test how well the correction works, consider the RMS variability of float displacements about the 10 x 10 degree mean:

Befo

re c

orr

ecti

on

Aft

er

corr

ecti

on

diff

ere

nce

22 )()( dydydxdx

Applying the “pseudo-displacement” correction to the float data reduces the variance of float displacements by a factor of 1.5 to 2.

Note that most improvement occurs in mid- to high-latitudes.

Objective maps: Davis Objective maps: Davis (1998)(1998)

• Bin-average of u and v on ~1° x 1° gridBin-average of u and v on ~1° x 1° grid

• Exponential covariance fcn.:Exponential covariance fcn.:

C(R) = ( 1 + R + 1/6 RC(R) = ( 1 + R + 1/6 R22 – 1/6 R – 1/6 R33 ) ) ee–R–R

• R depends on f/H, to model topographic steeringR depends on f/H, to model topographic steering

• Mapped relative to WOA01 1000/2000m dynamic Mapped relative to WOA01 1000/2000m dynamic heightheight

Tiles were used to avoid inversion of large matrices

Lx = 500 kmLy = 300 kmSNR: 1

Results: 1000m dynamic height

Objectively mapped from float displacements using techniques of Davis, JGR, 1992

Results: 1000m velocity

Objectively mapped from float displacements using techniques of Davis, JGR, 1998

Maps of dynamic height with and without ‘pseudo-displacement’ correction

Although very similar, the ‘corrected’ map shows slightly less noise in the S. Pacific subtropical gyre and a shaper gradient across much of the ACC

Difference between corrected and uncorrected dynamic height maps show largest differences in eddy-rich regions.

Future work: can we detect Future work: can we detect changes in the deep circulation changes in the deep circulation

from float data?from float data?

From Davis, JPO, 2005

From Lavender et al., Nature, 2000

SummarySummary• Argo is now producing displacements Argo is now producing displacements

(and profiles) at an unprecedented (and profiles) at an unprecedented rate with near global coveragerate with near global coverage

• SSH data is complementary to float SSH data is complementary to float displacement data and can be used to displacement data and can be used to reduce error in time-averaged reduce error in time-averaged estimates of circulationestimates of circulation

• Coverage is now good enough to make Coverage is now good enough to make preliminary maps of 1000m velocity preliminary maps of 1000m velocity fieldfield

And now for something And now for something completely different…completely different…

““Recent Cooling of the Recent Cooling of the Upper Ocean”Upper Ocean”

J. Lyman, J. Willis, G. JohnsonJ. Lyman, J. Willis, G. Johnson

Globally averaged upper-ocean Globally averaged upper-ocean Heat content from profile dataHeat content from profile data

From Lyman et al., Science, submitted

The Ocean COOLED!!!

Globally averaged upper-ocean Globally averaged upper-ocean Heat content from profile dataHeat content from profile data

ReferencesReferences

Rio and Hernandez, 2004Rio and Hernandez, 2004 Rio, M.-H. and F. Hernandez, A mean Rio, M.-H. and F. Hernandez, A mean dynamic topography computed over the world ocean from dynamic topography computed over the world ocean from altimetry, in situ measurements and a geoid model. altimetry, in situ measurements and a geoid model. Journal of Journal of Geophysical ResearchGeophysical Research 109109, C12032, 2004., C12032, 2004.

Roemmich and Gilson, 2001Roemmich and Gilson, 2001 Roemmich, D., J. Gilson, Eddy Roemmich, D., J. Gilson, Eddy Transport of Heat and Thermocline Waters in the North Transport of Heat and Thermocline Waters in the North Pacific: A key to understanding interannual/decadal climate Pacific: A key to understanding interannual/decadal climate variability? variability? Journal of Physical OceanographyJournal of Physical Oceanography, , 3131, 6757-687, , 6757-687, 2001.2001.

Davis, 1998Davis, 1998 Davis, R.E., Preliminary results from directly Davis, R.E., Preliminary results from directly measuring middepth circulation in the tropical and South measuring middepth circulation in the tropical and South Pacific. Pacific. Journal of Geophysical Research – OceansJournal of Geophysical Research – Oceans, , 103103, , 24,619-24,634, 1998. 24,619-24,634, 1998.

Davis et al, 1992Davis et al, 1992 Davis, R.E., D.C. Webb, L.A. Regier, J. Dufour, Davis, R.E., D.C. Webb, L.A. Regier, J. Dufour, The autonomous lagrangian circulation explorer (ALACE), The autonomous lagrangian circulation explorer (ALACE), Journal of Atmospheric and Oceanic TechnologyJournal of Atmospheric and Oceanic Technology, , 99, 264-285, , 264-285, 1992. 1992.

Comparisons with other Comparisons with other velocity estimatesvelocity estimates

Dynamic height at the surface

Mapped 1000m dynamic height + 0/1000m dyn. ht. mapped from Argo profile data.

Combined mean dynamic topography from Rio and Hernandez, JGR, 2004

Contours are 10 cm