recent developments in nemo within the albatross project

HAL Id: hal-02418218https://hal.inria.fr/hal-02418218

Submitted on 18 Dec 2019

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Recent developments in NEMO within the Albatrossproject

Florian Lemarié, Guillaume Samson, Xavier Couvelard, Gurvan Madec,Romain Bourdallé-Badie

To cite this version:Florian Lemarié, Guillaume Samson, Xavier Couvelard, Gurvan Madec, Romain Bourdallé-Badie.Recent developments in NEMO within the Albatross project. DRAKKAR 2019 - Drakkar annualworkshop, Jan 2019, Grenoble, France. �hal-02418218�

https://hal.inria.fr/hal-02418218

https://hal.archives-ouvertes.fr

Recent developments in NEMO within the Albatrossproject

F. Lemarie1, G. Samson2, X. Couvelard3, G. Madec1,4, R. Bourdalle-Badie2

1 Inria (Airsea Team), Univ. Grenoble-Alpes, LJK, France2 Mercator Ocean International, Toulouse, France3 Laboratoire d’Oceanographie Physique et Spatiale (LOPS), Brest, France4 Sorbonne Universites-CNRS-IRD-MNHN, LOCEAN Laboratory, Paris, France

Context: Albatross project funded by CMEMS

→ Representation of ocean/waves/atmosphereinteractions in global eddying operational systems

- Focus on the characteristic scales of the oceanic mesoscale (a few kilometersand from a few days to a few weeks)

. Demonstrate the contribution of ocean/waves coupling

. Account for eddy-scale wind-SST and wind-currents effects

1. Implementation of a 2-way ocean/waves coupling (more conventional)

2. Explore the possibility to dynamically downscale atmospheric data to theoceanic resolution via a simplified model (more exploratory)

F. Lemarie – Recent developments in NEMO within Albatross 2


1Inclusion of wave-induced terms in NEMO & couplinginfrastructure

Inclusion of wave-induced terms in the PE modelsMathematical derivation• McWilliams et al. (2004)

- Eulerian frame- Asymptotic expansion of the wave effects to some order in wave steepness

• Ardhuin et al. (2008)- Hybrid Lagrangian-Eulerian frame (Generalized Lagrangian Mean)- Effects of vertical shear are ignored

→ nearly equivalent wave-current equations at leading orders (Ardhuin et al.,

2017)

→ the wave effects on the current momentum expressed in the form of thevortex force

. Examples of practical implementations in PE models:Uchiyama et al. (2010); Bennis et al. (2011)


Curl form of NEMO equations with generalized vert. coordinate (e3 = ∂kz)

∂tu = +(f + ζ)(v + vs)−∂x‖uh‖2

2−

(ω +ωs)e3

∂ku−∂x(ps + pJ )

ρ0−

(∂x(ph + p)

)z

ρ0+Fu

∂tv = −(f + ζ)(u + us)−∂y‖uh‖2

2−

(ω +ωs)e3

∂kv −∂y(ps + pJ )

ρ0−

(∂y(ph + p)

)z

ρ0+F v

∂kph = −ρge3 − ∂kp + ρ0

(us∂ku + vs∂kv

)“wavy hydrostatic” balance

∂t(e3θ) = −∂x(e3θ(u + us)− ∂y(e3θ(v + vs))− ∂k(θ(ω +ωs)) + Fθ

∂te3 = −∂x(e3(u + us))− ∂y(e3(v + vs))− ∂k(ω +ωs)ρ = ρeos

• (us, vs, ωs) : Stokes drift velocity (assumed to be non-divergent)

• pJ : depth-uniform wave-induced kinematic pressure term• p : shear-induced 3D pressure term associated with vertical component of VF

WSt−Cor =

fvs

−fus0

, WVF =

ζvs − ωs

e3∂ku

−ζus − ωs

e3∂kv

us

e3∂ku + vs

e3∂kv

, WPrs =

pJ + p

pJ + pp


Curl form of NEMO equations with generalized vert. coordinate (e3 = ∂kz)

∂tu = +(f + ζ)(v + vs)−∂x‖uh‖2

2−

(ω +ωs)e3

∂ku−∂x(ps + pJ )

ρ0−

(∂x(ph + p)

)z

ρ0+Fu

∂tv = −(f + ζ)(u + us)−∂y‖uh‖2

2−

(ω +ωs)e3

∂kv −∂y(ps + pJ )

ρ0−

(∂y(ph + p)

)z

ρ0+F v

∂kph = −ρge3 − ∂kp+ ρ0 (us∂ku+ vs∂kv) “wavy hydrostatic” balance

∂t(e3θ) = −∂x(e3θ(u + us)− ∂y(e3θ(v + vs))− ∂k(θ(ω +ωs)) + Fθ

∂te3 = −∂x(e3(u + us))− ∂y(e3(v + vs))− ∂k(ω +ωs)ρ = ρeos

. Small vertical shear limit (Bennis et al., 2011)→ hydrostatic relation is unchanged

. Neglect wave-induced non-conservative forces (wave dissipation, rollers, etc)

. Adjustments in the barotropic mode due to the convergence of the Stokes drift

. No need to explicitly compute ωs


Reconstruction of Stokes drift velocity profile

Inputs from wave model: ush(η), ‖Ts‖, → ush(z, ke) = ush(η)S(z, ke)

• ke : degree of freedom to impose that ‖∫ush(z, ke)dz‖ = ‖T

s‖• S(z, ke) from Breivik et al., 2016

• Stokes drift interpreted in a FV sense ush(zk) =

ush(η)

(e3)k

∫ zk+1/2

zk−1/2

S(z, ke)dz

Computation of surface momentum flux

• In NEMO : τatm from IFS bulk formulation (Aerobulk) using αch from wave model.

• In wave model : τatmww3 assumes neutral stratification (wave models hyper

sensitivite to stress computation).

A way to account for the momentum flux consumed by the wave field

τ oce = τ atm − (τ atmww3 − τ oce

ww3)

. Pragmatic choice which breaks momentum conservation.


Wave-enhanced mixing (1-equation TKE scheme)

Additional terms in wavy hydrostatic balance affects TKE equation

ush∂z⟨u′hw

′⟩ =⟨u′hw

′⟩ ∂zush + ∂z(ush⟨u′hw

′⟩). Modification of shear production term + adjust Ave value

∂te =Avm

e23

[(∂ku)

2+ (∂kv)

2+ (∂ku)(∂ku

s) + (∂kv

s)(∂kv

s)]−AvtN2

+1

e3∂k

[Ave

e3∂ke

]−cε

e3/2

l2ε

Surface B.C. for e :(Ave

e3∂ke)z=z1

= −ρ0g∫ 2π

0

∫∞0Sdsdωdθ

Surface B.C. for mixing length : lz=η = κ (Cmcε)1/4

Cm(η + z0), z0 = 1.6Hs


Wave-enhanced mixing (1-equation TKE scheme)

Additional terms in wavy hydrostatic balance affects TKE equation

ush∂z⟨u′hw

′⟩ =⟨u′hw

′⟩ ∂zush + ∂z(ush⟨u′hw

′⟩). Modification of shear production term + adjust Ave value

∂te =Avm

e23

[(∂ku)

2+ (∂kv)

2+ (∂ku)(∂ku

s) + (∂kv

s)(∂kv

s)]−AvtN2

+1

e3∂k

[Ave

e3∂ke

]−cε

e3/2

l2ε

Langmuir cells parameterization : Axell (2002)

→ Additional source term in TKE equation ∆tPLC with PLC = w3LC/HLC

wLC =

{cLC‖us‖ sin

(− πzHLC

), if − z ≤ HLC

0, otherwise, −

∫ η

−HLC

N2(z)zdz =

‖us‖22

‖us‖ = max {us(η) · eτ , 0} intensity of LC scales with θusτ ( Van Roekel et al., 2012)


Practical implementation

• OASIS−MCT Interface shared with Croco and Mars3d

• Interface in NEMO inline with the generic interface with atmosphere

Variable descriptionuh(z = η) Oceanic surface currents O→W

uatm10 10 m-winds from external dataset O→W

ush(z = η) Sea-surface Stokes drift W→O

‖Ts‖ norm of the Stokes drift volume transport W→O

Φoc TKE surface flux multiplied by ρ0 W→O

αch Charnock parameter W→O

Hs Significant wave height W→O

τww3w Wave-supported stress W→O

pJ Bernoulli head W→O

Table : Variables exchanged between NEMO and WW3 via OASIS−MCT.

• Examples of ORCA025 numerical results in next talkF. Lemarie – Recent developments in NEMO within Albatross 8


2Downscaling of atmospheric data at the oceanic resolution

Objectives

SURROGATE MODEL�atm

LS (x, z, t) = MAGCM(. . . ,�oceLS (x, z = zsfc, t))

�oceHR(x, z = zsfc, t)

e�atm

HR (x, z = 10 m, t)

How to define such surrogate model ?• Estimate ∂MAGCM/∂φ

ocesfc via sensitivity analysis and subsequent model

reduction• Build a surrogate model via learning strategies (huge computational and

data requirements)• Select feedback loops of interest and mimic the physical mechanisms

→ The newly developed computational framework allows all thosepossibilities to be investigated


Air-sea interactions at the oceanic mesoscales

Wind-SST coupling (thermal coupling)

1. Modulation of PBL turbulence (e.g.Chelton, 2013; Frenger et al., 2013){

∇× τ = F1 (∇SST)∇ · τ = F2 (∇SST)

2. Pressure gradient adjustment (e.g.Minobe 2008; Lambaerts et al. 2013)

∇ · τ ∝ −‖∇2SST‖

Current feedback (dynamical coupling)

τ = ρaCD‖ua − uo‖(ua − uo)

• Strongly reduced mesoscale activity(”eddy damping”) (e.g. Dewar & Flierl,1987; Renault et al., 2016)

• Strongly increases vertical velocityanomalies associated to eddies (e.g.Oerder et al., 2017)


WEk [m day�1]

Absolute winds

Relative windsOerder et al., 2017

Air-sea interactions at the oceanic mesoscales

. A good representation of those interactions requires an interactive MABL- Under-estimation of wind-SST coupling in bulk mode- Over-estimation of wind-current coupling in bulk mode

. The atmospheric resolution must be ”eddy-resolving” (i.e. ∆xoce ≈ ∆xatm)

. Interactive MABL required for a consistent integration of surface waves

Similar initiatives :• Advective atmospheric mixed layer model of Seager et al., 1995• CheapAML (Deremble et al., 2013)→ dynamically passive ABL model


Current status of the project (1/2)

1. Definition of a single-column model (ABL1d)

Integrate winds u, potential temperature θ and specific humidity q∂tu = fk× u + ∂z (Km∂zu)−

(1ρ∇p)LS

∂tθ = ∂z (Ks∂zθ) + λs(S(θ)− θLS)∂tq = ∂z (Ks∂zq) + λs(S(q)− qLS)

Blue terms are specified via large-scale dataRed terms are given by turbulent closure

. Radiative forcing is kept as it is

. Surface boundary conditions for Km∂zu|z=0,Ks∂zθ|z=0, Ks∂zq|z=0 via IFS (aerobulk) bulkformulation

. Relaxation term scales with PBL height


Current status of the project (2/2)

2. Turbulent closure scheme : TKE-based scheme of Cuxart et al. (2000). used operationally at Meteo-France (e.g. in Arome and Meso-NH models)

. recoded from scratch to allow more flexibility and better performances

3. Development of preprocessing tools forlarge-scale forcing

4. Implementation in NEMO surface module- Option to split NEMO and SAS on separate nodes- Standalone mode- Compatible with sea-ice

Computational cost (50 vertical levels, no sea-ice)• + 12% in memory• + 7 - 12 % en elapsed time (depending on options)

GLS ext. mode ABL1d I/OBulk mode 19.44% 11.3% - 0.34%ABL mode 18.06% 10.5% 6.3% 0.64%

waves

BULK formulation

Simplified Atmospheric Boundary layer model

3D Large-scale data

Surface module

LIMsea-ice

OCEAN


Current status of the validation strategy (1/2)

1. Standardized test-cases from ABL community (see GABLS initiative)

→ Neutral turbulent Ekman layer at 450N (Cuxart et al., 2000)

-1 0 1 2 3V [m/s]

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Z f /

u*

nn_amxl = 1

nn_amxl = 3

Cuxart et al., 2000

MESO-NH 1D

LES

MESO-NH 1D

LES

SIMBAD1D

2 4 6 8 10 12U [m/s]

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Z f /

u*

nn_amxl = 1nn_amxl = 3

→ Stably stratified boundary layer (typical situation over sea-ice)

Rodier et al., 2017a) b)

0 2 4 6 8 10Wind speed [m/s]

0

100

200

300

400

altit

ude

[m]

nn_amxl = 0nn_amxl = 1nn_amxl = 2nn_amxl = 3

263 264 265 266 267 268Potential temperature [K]

0

100

200

300

400

altit

ude

[m]

nn_amxl = 0

nn_amxl = 1

nn_amxl = 2

nn_amxl = 3

SIMBAD1D


Current status of the validation strategy (2/2)

2. Winds across a Midlatitude SST Front (Kilpatrick et al., 2014)

ABL_1D MESO-NH LES

2D x-z (solution at equilibrium)

3. NEMO1D / ABL1D coupling at PAPA station (50.1oN, 144.9oW)→ Theo Brivoal MSc


Examples of numerical results

Large-scale winds over an idealized eddy

100

200

300

Relative winds (ABL)

50 100 150 200 250 300

100

200

300

Relative winds (Bulk)

8

6

4

2

0

2

4

6

81e 7

⇒ Reduction by 30% of wind stress curl

Bay of Biscay 1/12◦ realistic simulations

Courtesy of Mercator Ocean


Conclusions

• Implementation of a 2-way ocean-wave model including the wave-inducedterms thought to be relevant at global eddying scales

• Representation of some mesoscale air-sea feedbacks with a simplifiedmodel

- ABL response qualitatively and quantitatively ok in idealized experiments→ extension to realistic cases

- Validation of implementation over sea-ice

- More advanced formulation including horizontal advection and fine-scalepressure gradient will be tested (based on Konor, 2013; Durran, 2008)

Publications to come• Couvelard X., F. Lemarie, G. Samson, J.L. Redelsperger, F. Ardhuin, R. Benshila, G.

Madec, Development of a 2-way coupled ocean-wave model: assessment on a globaloceanic configuration, Geosc. Mod. Dev.

• Lemarie F., G. Samson, J.-L. Redelsperger, H. Giordani, G. Madec, R. BourdalleBadie, T. Brivoal, A simplified atmospheric boundary layer model for oceanicmodeling purposes, Geosc. Mod. Dev.


recent developments in nemo within the albatross project

Documents