SWARP: WP2, wave modelling
Fabrice Ardhuin, Mickael Accensi,
Pierre Queffeulou, Fanny Ardhuin (Ifremer)
Fabrice Collard (Oceandatalab)
Aron Roland (TU Darmstadt)
http://wwz.ifremer.fr/iowaga
http://www.previmer.org/previsions/vagues
1
Outline
1. Building on other projects and developments : IOWAGA
(ERC), Waves-NOPP (U.S. ONR) ...
2. Specific developments for SWARP:
• Numerics (implementation of new model version)
• Development of parameterizations:
- friction below the ice
- shorelines & ice-edge reflections
- scattering
Estimation of energy & momentum fluxes
3. Hindcasts & forecasts
wwz.ifremer.fr/iowaga
2
SWARP Kick-off, February 3, 2014
1
Building on other projects
and developments
wwz.ifremer.fr/iowaga
3
1. a multi-parameter
global wave hindcast
The ERC-funded IOWAGA project integrates
- Coastal hydrodynamics
- Air-sea fluxes
- Remote sensing
- Microseisms
into a single consistent wave modelling system.
→ new version 4.18 of WAVEWATCH III (distributed by NOAA/NCEP)
→ hindcast database, >40 different parameters (http://tinyurl.com/iowagaftp)
- Results from global multigrid (1993-2013) :
GLOBAL : (30' resolution)... up to 80°N only
PACE (10'): East Pacific (Alaska to Peru), with Hawaii & ...
ATNW (10'): U.S. East coast + Gulf of Mexico
ATNE (10'): Iceland to Morocco (6').
NC (3'): New Caledonia & Vanuatu (3')
CRB (3'): Lesser Antilles (Puerto Rico to Venezuela, 3')
1 or 2 runs (with CFSR and / or ECMWF op. Winds)
Recent runs : > 9000 full spectra output (along all shorelines + others)
+ finite element zooms in coastal areas
wwz.ifremer.fr/iowaga
4
1. a multi-parameter
global wave hindcast
IOWAGA hindcast particularly focused on
- better swells (swell dissipation based on Ardhuin et al. GRL 2009)
- better high frequency energy ( → Stokes drift, mss, see JPO 2009, 2010)
- better directional parameters (Ardhuin & Roland JGR 2012, Ardhuin et al. JASA 2013)
important for acoustic and seismic noise
General validation : Rascle & Ardhuin (Ocean Modelling 2013)
– Here are a few plots with data from http://tinyurl.com/iowagaftp
wwz.ifremer.fr/iowaga
… but GLOBAL : (30' resolution)... up to 80°N only
5
SWARP Kick-off, February 3, 2014
2
Specific developments for
SWARP
wwz.ifremer.fr/iowaga
6
2. developments for
SWARP
Extension to Arctic : just one more grid ?
- 12 km polar grid, same as NSIDC or Ifremer-CERSAT sea ice products
(concentration and drift from SSM/I + ERS / QuikSCAT / ASCAT)
Bathymetry for new grid (IBCAO)
-first tests : forced by global grid
- next steps :
– Two-way nested in multigrid
Possible higher resolution if
needed (12 km Arctic is less
expensive that global hindcast)
wwz.ifremer.fr/iowaga
7
2. developments for
SWARP
Parameterizations:
Implementation of shoreline reflection on curvilinear grid
- could be added for ice edge …
Implementation of viscous & turbulent friction under the ice :
Generalization of Liu & al. (JGR 1991) following work on swell dissipation
(Ardhuin & al. 2009, Rascle & Ardhuin 2013, Perignon et al., submitted)
- non linear dissipation rate for turbulent cases, no “eddy viscosity”
- transition threshold (laminar to turbulent), random waves → smoothing
Scattering term: to be defined
Solution method taken from bottom topography scattering (Ardhuin & Magne 2007)
wwz.ifremer.fr/iowaga
8
2. developments for
SWARP
It runs! …
Example of wave data retrieval from SAR
imagery (2011/03/04 9h14, Envisat)
sea ice water
WW3
results :
Wave heights from SAR
(not adapted yet to sea ice)
wwz.ifremer.fr/iowaga
9
SWARP Kick-off, February 3, 2014
3
Hindcasts and forecasts
wwz.ifremer.fr/iowaga
10
2. Tests with different
forcing products
Sept 16th :
Sept 17th :
Sept 18th :
Yearly sea ice
extent minimum
Freeze observed by scatt., not
by radiometer
Freeze observed by both
sensors
Concentration
from radiometer
wwz.ifremer.fr/iowaga
Standard deviation of
backscatter data
Concentration
from radiometer
Concentration
from radiometer
Interest
of the use
of both
and
radiometer
Scatterometer
sensor
canscatterometer
detect the very
first
stage ofdata
sea ice
for MIZ monitoring
11
3.
fo
2008-2015: Arctic hindcasts
- multi-grid system including 12 km grid
- ECMWF winds + ?? ice
- higher resolution mesh focusing on ice edge ?
2016: forecasts
Deliverables:
Jan 2015 : First version of wave and sea ice models. Data exchanged between the partners,
deliverable 2.1 “First wave model results covering the MIZ”
Jan 2016 : Validation completed.
“Upgraded wave model”
Data flows integrated to NavPlanner, deliverable 2.2
July 2016 : deliverable 2.3 “Forecast and hindcast simulations for demonstration”
wwz.ifremer.fr/iowaga
12
3. Validation of the
wave model
a) wave parameters
Generic validation:
• Altimeters (Hs & mss ): in ice-free water
• Permanent buoys: general context (Iceland + Barents Sea )
rms error for Hs (%)
Ardhuin
Ardhuin
et al. 2010
et al. 2010
(Prévimer)
WIFAR 2012 + other field data validation:
• Detailed estimation of spectra, specific validation of wave attenuation
rates
SAR-derived attenuation rates
wwz.ifremer.fr/iowaga
13
1. a multi-parameter
global wave hindcast
The ERC-funded IOWAGA project integrates
- Coastal hydrodynamics
- Air-sea fluxes
- Remote sensing
- Microseisms
into a single consistent wave modelling system.
→ new version 4.18 of WAVEWATCH III (distributed by NOAA/NCEP)
→ hindcast database, >40 different parameters (http://tinyurl.com/iowagaftp)
- Results from global multigrid (1993-2013) :
GLOBAL : (30' resolution)... up to 80°N only
PACE (10'): East Pacific (Alaska to Peru), with Hawaii & ...
ATNW (10'): U.S. East coast + Gulf of Mexico
ATNE (10'): Iceland to Morocco (6').
NC (3'): New Caledonia & Vanuatu (3')
CRB (3'): Lesser Antilles (Puerto Rico to Venezuela, 3')
1 or 2 runs (with CFSR and / or ECMWF op. Winds)
Recent runs : > 9000 full spectra output (along all shorelines + others)
+ finite element zooms in coastal areas
wwz.ifremer.fr/iowaga
14
AGU meeting, San Francisco, December 2013
3
Scattering in the MIZ inferred
from seismic/acoustic
noise ?
wwz.ifremer.fr/iowaga
15
3. Scattering and noise : theory
Hasselmann (1963) : nearly opposing waves generate seismic noise
k2
K
Movie of sea surface elevation
Z = Z1 + Z2
k1
The interaction of k1 and k2 gives
noise at K = k1 + k2 and f = f1 + f2
Any 2nd order quantity like Z2 will
thus contain
K = k1 ± k2 and f = f1 ± f2
Resonant interaction if 2 pi f / K = Cs ,
the phase speed of one seismic mode.
For any f, this selects K.
Higher order interactions : K = k1 ± k2 ±
k3 and f = f1 ± f2 ± f3 … and so on …
wwz.ifremer.fr/iowaga
16
5.1 Validation of the
wave model
a) wave parameters
Generic validation:
• Altimeters (Hs & mss ): in ice-free water
• Permanent buoys: general context (Iceland + Barents Sea )
rms error for Hs (%)
Ardhuin
Ardhuin
et al. 2010
et al. 2010
(Prévimer)
WIFAR 2012 + other field data validation:
• Detailed estimation of spectra, specific validation of wave attenuation
rates
SAR-derived attenuation rates
wwz.ifremer.fr/iowaga
17
3. Scattering and noise :
When do we get noise ?
From wave spectrum to acoustic or seismic noise :
deep water waves : noise source proportional to
Noise is linked to waves in opposite directions. Loud sources in MIZ ??
Example of seismic records around the Pacific
due to « swell collision » :
FF
wwz.ifremer.fr/iowaga
Obrebski et al. (GRL 2012)
18
3. Scattering and noise :
Different modes
Different modes :
- seismic waves (Rayleigh or body waves) → can be
recorded on land !
- acoustic gravity waves (evanescent : local sea state)
100 to 500 m
depth
Evanescent
modes
dominate
(see Cox &
Jacobs GRL
1989)
wwz.ifremer.fr/iowaga
19
3. Scattering and noise :
Example of a noise event ...
KBS
Greenland
DBG
SCO
wwz.ifremer.fr/iowaga
20
3. Scattering and noise :
Example of a noise event ...
Modeled seismic sources
Hs
Seismic spectra at SCO
wwz.ifremer.fr/iowaga
21
1. Bottom friction
Cliffs north of Jan Mayen : strong wave reflection (another day)...
wwz.ifremer.fr/iowaga
22
AGU meeting, San Francisco, December 2013
4
Perspectives
wwz.ifremer.fr/iowaga
23
- Introducing wave dissipation and scattering in the MIZ :
Les MNT dans la modélisation côtière -
– Verification of dissipation with SAR-derived Hs and Tm02
Journée
Valor'IG09
– Scenarii for scattering : how
large can
it be given seismic records ?
- Deploying pressure sensors 100 to 500 m from the surface?
– Direct measurements of evanescent modes → measurement of I(f) integral,
a very strong constraint on scattering strength.
Anybody has a mooring out there?
wwz.ifremer.fr/iowaga
It works ! Data from SBE26 in 100 m depth (Ardhuin & al. JASA 2013)
24