High resolution modelling of dense water formation in the Northwestern Mediterranean: benefits from an improved initial

stratification in summerC. Estournel, P. Testor, P. Damien, L. Mortier, P. Marsaleix, J.M. Lellouche, C. Ulses, F. Kessouri,

P. Raimbault, L. Coppola

D’Ortenzio & Ribera d’Alcalà, 2009Bergamasco & Malanotte-Rizzoli, 2010

Importance of convection and dense water formation in the western Med basin- thermohaline circulation- ecosystem productivity- ventilation of deep layers

Dense water formation is suspected to be highly sensitive to climate change

Monitoring DWF at interannual scale (characteristics of newly-formed waters) is a priority

Time series from the « Lion » mooring line

(de Madron et al., 2013; Houpert)

does not provide the volume of dense water

Objective : to derive a method based on observations and modelling to characterize the interannual variability of dense water formation

Numerical modelling should be able to provide detailed information on newly-formed waters if accurate initial conditions and meteorological forcing are used

The method is tested on winter 2012-2013 as many data are available

D’Ortenzio, pers. comm.

The numerical domain

3D ModelS26 (Marsaleix et al., 2008, 2009, 2011, 2012)1 km horizontal resolution

ForcingOGCM : MERCATOR operational modelMeteo: 3 hrs ECMWF forecasts. Turbulent fluxes calculated by Bulk F.

Data available

Each summer : cruise of the MOOSE monitoring programme

numerical domain

2012 : 88 CTD + 52 ARGO profiles

used to improve the initial state of the model

During convection in 2013, DEWEX cruise of the MERMEX programme

used to check the simulation of convection

75 CTD + 139 ARGO profiles

Initial conditions

1 August 2012

Anomalies at 1500 m

Interpolation of anomalies at 1500 m

The initial state (and boundary conditions) is corrected from these anomalies

z1

z2

0

Assessment of the simulation based on the CTD

available during convectionDiagnostic : stratification index profile

dzzzzFz

))((0

111

simulation without initial state correction

simulation with initial state correction

Anomaly of the stratification index(simulation-observation)

averaged over all the CTD

6 months later

Anomaly of the stratification index(simulation-observation)

Sensitivity study to meteorological fluxes

without initial state correction

Model

Obse

rvati

on

without initial state correction with initial state correction without meteo correction

Latent heat flux X 1.25 Wind X 1.13

correlation: 0.32 correlation: 0.5

correlation: 0.59 correlation: 0.64

Density at 2000 m

Map of the stratification index at 1000 m

model V x1.13Observation

(logarithmic scale)

2

1

0

-1

15000 km3

46000 km3

53000 km3

Difference of volume by density class15 March 2013 – 1 September 2012

Conclusion

Observations in summer are unvaluable to improve the model initial statethe simulation keeps the benefits of the improved initial state until winter

The method combining observations and modelling seems promising to monitor dense water formation in the northwest basin

After correction of the initial state, a stratification bias remains. Increasing wind (13%) or directly heat fluxes (LHF x 1.25) allowed to reduce the bias and to be very close to observations

The priority is now to apply the method on other years to estimate the robustness of these results

15000 km3

46000 km3

53000 km3

Difference of volume by density class15 March 2013 – 1 September 2012

Mean buoyancy flux over autumn winter kg/m2/day

-0.727

-0.783

-0.863

X 3 X 1.08