14
A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle localisation of filaments stirring intensity mesoscale + Time variability First application to satellitte observations Chlorophyl filaments (res. 1 km) and lyaponov eponnts

A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

Embed Size (px)

Citation preview

Page 1: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

A NEW APROACH TO STIRRING

Lehahn, d'Ovidio, et al., JGR, 2007

Méthod: compute Lyapono exonent

d'Ovidio et al., GRL, 2004

sub-mesoéchelle

► localisation of filaments► stirring intensity

mesoscale+

Time variability

First application to satellitte observations

Chlorophyl filaments (res. 1 km) and lyaponov eponnts

Page 2: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

Altimetry-derived unstable manifolds(spatial+temporal variability) vs. chlorophyll pattern

Submesoscale tracer filaments are predicted from altimetry!

Altimetric velocities vs. chlorophyll pattern

The agreement seems to hold at scales larger than altimetric resolution (>1/3 deg.) and not below.

Page 3: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

= initial separation

= amplification factor

= time needed for the perturbation to grow

Aurell et al., Phys. Rev. Lett. 77, 1262 (1996)Boffetta et al., J. of Phys. A, 30, 1 (1997) chao-dyn/9904049

The intensity of stirring can be measured with local Lyapunov Exponents

Page 4: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

A field of Lyapunov exponents from surface velocities

Comparisonwith SST pattern

Page 5: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

Method: 2km-reolution model : SST 3000 km x 2000 km

How is stirring affected by degrading in space and time the velocity field?

Page 6: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

PRELIMINARY RESULTS

2 days

4 days

6 days

8 days

Current altimetry products

Pixel-by-pixel comparison:space variability dominates >16 kmtime variability dominates below

Mean comparison (400x400 km):Smoother behaviour (temporal variability less important)

the reduced resolution underestimates stirring (up to 50%)critical resolution associated to filament position (<-visible in pixel by pixel comparison, not on regional means)

Page 7: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

SUMMARY

Stirring is an important (main?) effect of the horizontal velocities at the submesoscale. It depends on both the spatial and temporal variability of the velocity field. Its quantification through the Lyapunov Exponent is a natural and objective way of quantifying the error when reducing the spatial and/or temporal resolution of a velocity field.

An analysis performed with the high resolution model GYRE suggests that, starting from current altimetry resolution, the highest gain is obtained increasing the spatial resolution up to 16 km. After that, temporal resolution should be increased.

Time resolutiion appears to be critical for filament position not stirring intensityIn general, a reduced resolution underestimates stirring intensity (up to 50%).

The analysis should be repeated resampling the model SSH over simulated (multi-)satellite tracks.

Page 8: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

The impact of mesoscale temporal variability on dispersion processes

Page 9: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

We are re-processing the entire altimetric dataset15 years, global coverage; near-real time analysis possible

Page 10: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

2003-2006 average: patterns correlated to Eulerian diagnostics (EKE, etc.)

0

0.5

day-1

stirring (Lyapunov exponents)

Page 11: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

2003-2006 average (no time variability): we compute stirring rates also at frozen velocity fields.

0

0.5

day-1

“frozen” stirring (Lyap. exp. from frozen velocity field)

Page 12: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

The difference between stirring and stirring with frozen velocities unveil the role of time variability for mesoscale dispersion processes.

2003-2006 average (fraction of stirring due to time variability)

0

1

(stirring-(frozen stirring))/stirring

Page 13: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

2003-2006 average (fraction of stirring due to time variability)

0

1

(stirring-(frozen stirring))/stirring

In the case of strong currents, like the Gulf stream, stirring processes are dominated by persistent jets and weakly affected by mesoscale time variability.

On the contrary, the weaker stirring occurring in the subtropical gyres is easily affected by mesoscale time variability.

In the subpolar regions mesoscale time variability has also a strong effect, by breaking the “trapping” effect of stationary eddies. This is shown by very localized mesoscale signal in correspondance of the cores of the most stationary eddies.

Note the colorscale: in same cases, the time variability is responsable of almost all (~1) the stirring!

Page 14: A NEW APROACH TO STIRRING Lehahn, d'Ovidio, et al., JGR, 2007 Méthod: compute Lyapono exonent d'Ovidio et al., GRL, 2004 sub- mesoéchelle ► localisation

CONCLUSION (Time variability of the ocean currents: the role of chaos)

The time variability of surface currents (in fact, chaos) has a strong effect outside permanent jets, especially in supolar regions and inside the subtropical gyres.

These regions are likely to be the ones that will most benefit from next-generation altimetry data at improved temporal resolution.

For high-resolution circulation models, these are also the regions where small-scale transport crucially depends on a correct representation of eddy temporal dynamics.In these regions, a wrong temporal variability of eddies corresponds to wrong mesoscale dispersion even if the EKE distribution is perfect.

A dynamical interpretationA two-dimensional system with no time dependence has no chaos: trajectories (and fronts) follow altimetric isolines. In this case, eddies are perfectly isolated from outside. In this case, Lyapunov exponents measure basically the strain. The temporal variability has the main effect of breaking the closed orbits, creating spirals and filaments that connect the eddy interior with the environment. This chaotic dynamics adds to the strain contributing to the stirring.