Advanced coupled atmosphere-wave-ocean
modeling for improving tropical cyclone
prediction models
PI: Isaac Ginis
University of Rhode Island
Co-PIs: T. Hara (URI), E. Andreas (NWR),
R. Lukas (UH), A. Soloviev (NSU)
Collaborators: J.-W. Bao, C. Fairall (NOAA/ESRL)
H. Tolman (NOAA/NCEP)
NOPP Review Meeting, 2012, Miami, FL
1
Long-Term Goals
1) To understand the physical processes that control the
air-sea interaction and their impacts on intensity changes
in tropical cyclones.
1) To develop a physically based and computationally
efficient unified air-sea interface module for use in the
next generation of research and operational coupled
atmosphere-wave-ocean-land models.
Year 2: Work Completed
• Evaluation of WAVEWATCH v 3.14 wave model in hurricane
conditions.
• Investigation of the sea state dependence of drag coefficient in
hurricanes with new URI and UM air-sea momentum.
• Investigation of the Stokes drift velocity and Coriolis-Stokes forcing
due to ocean surface waves under hurricane conditions
• Implementation of new methods for coupling the sea-spray
parameterization with the surface wave properties.
Year 2: Work Completed (cont’d)
• Implementation and testing of the air-sea interface module with
URI/ESRL air-sea coupling parameterizations into a research
versions of the GFDL and HWRF coupled hurricane-wave-ocean
model.
• Investigation of the upper and lower limits of the drag coefficient in
high wind conditions
Physics of Wind-Wave-Current Interaction
Image courtesy of Fabrice Veron
Wave model component - WAVEWATCH III
WAVEWATCH III can accurately reproduce observed
hurricane surface wave fields if:
- Wind forcing is reduced at very high wind speeds.
- Ocean current is explicitly included in the simulation.
WW3 significant wave height
field (color) at Sept. 15 2:00
UTC. The thick gray line is the
flight track.
Significant wave height comparison
between SRA measurements
(during this flight) and WW3 results
from experiments A, B (with
modified wind stress) and C (with
modified wind stress and including
ocean currents).
Comparison between modeled
and measured significant wave
heights from all flights.
.
WW3 v3.14: Extending to Finite/Shallow Water
* Previous version of WAVEWATCH III (v2.2.2) did not work well for water
depth less than 30m (grey area below)
•New version of WAVEWATCH III (v3.1.4) includes improved physics in shallower
water.
* We are validating the WAVEWATCH III (v3.1.4) results in shallower water against
observations (Scanning Radar Altimeter) in collaboration with Ed Walsh.
WW3 v3.14: Extending to Finite/Shallow Water
Hurricane Ivan (2004) significant wave height predictions
WW3 2.22
WW3 3.14
Difference
Sea state dependent drag coefficient
Isaac Ginis, Tetsu Hara, Brandon Reichl (URI), Mark
Donelan (UM)
• In the fully coupled hurricane-wave-ocean modeling
framework, the sea state dependent air-sea momentum flux
(drag coefficient) is calculated using the wave model
(WAVEWATCH III or others) output.
• In the air-sea interface module developed during the NOPP
project, different flux models (wave boundary layer models)
will be available for the flux calculations.
• We have examined two flux models (new URI and UM) as
potential candidates for the air-sea interface module.
pdf
Stokes Drift
Isaac Ginis, Tetsu Hara, Colin Hughes, Brandon Reichl,
John Bruce (URI)
•In the fully coupled hurricane-wave-ocean modeling
framework, the Stokes drift of surface waves introduces:
- Langmuir turbulence and enhanced/reduced upper
ocean mixing
- Coriolis-Stokes effect that effectively modifies the
momentum flux into subsurface current
• In addition, the Stokes drift introduces significant near
surface mass transport that may affect transport of
materials (e.g., oil).
Lagrangian floats deployed ahead of Hurricane
Gustav (2008) measured vertical kinetic energy in
the mixed layer
Courtesy of Eric D’Asaro
Langmuir turbulence
• D’Asaro et al. suggest that near surface turbulence and
upper ocean mixing may be significantly reduced when
surface waves are opposing the wind and suppress the
Langmuir turbulence.
• We have examined the Stokes drift (vertical profile)
under fetch dependent cases (uniform wind) and under
idealized hurricanes (stationary and translating).
• We are investigating the Langmuir turbulence under
hurricanes using LES in collaboration with Tobias
Kukulka (U. Del).
Angle difference between wind direction and Stokes
drift direction at z=k peak-1 under Idealized hurricane
Angle exceeds
90 degrees under
a translating
hurricane.
Left panels:
500x500 km
domain
Right panels:
100x100 km
domain
UT
Surface Mass Transport
• Empirical oil spill models assume drift speed of about
3% of the wind speed.
• Our calculations under fetch-dependent and hurricane
winds show that more than half (around 1.5-2% of the
wind speed) is due to the Stokes drift. In a moving
hurricane, this ratio varies, with the largest on the rearright side of the track.
Coriolis–Stokes Effect
The momentum flux into the ocean may be different
from the wind stress.
Ocean momentum equation with surface wave effects (Mellor 2008)
In meso-scale, deep water, hurricane ocean model:
Momentum equation is unchanged, but
- turbulence closure is modified by (unresolved) Langmuir turbulence.
- surface boundary condition is modified (momentum flux ≠ wind stress).
Air-sea momentum budget
(Fan et al. 2008) already
included in ASIM
Coriolis-Stokes
correction to
wind stress
18
Coriolis –Stokes effect
• The momentum flux into the ocean may be different
from the wind stress. The difference can be as large as
15% of the wind stress near the radius of maximum wind
to the right of the hurricane center, and it exceeds 30%
further away.
• The direction of the Coriolis-Stokes effect (vector) tends
to be to the right of the wind stress direction.
Drag coefficient parameterization
by Ed Andreas
Implications of the Straight Line
Best fit is
u*  0.0583U N10  0.243
where both u* and UN10 are in m/s.
Hence,
2
CDN10
 u* 

0.243 
 
   0.0583 

U N10 
 U N10 


4.17 
 3.40  10 1 

U N10 

3
2
2
The parameterization predicts
leveling off of the drag coefficient
We can combine linear relations in the smooth
regime and in the rough regime
Air-Sea Interface in Hurricane
Conditions
Alex Soloviev1, 2, Roger Lukas3
Silvia Matt1, Atsushi Fujimura2
1 Nova Southeastern University
2 University of Miami
3 University of Hawaii
In collaboration with
Isaac Ginis and Tetsu Hara
March 1, 2012
ONR/NOPP Review at UM RSMAS
Introduction
• In this work, we further develop the hypothesis that the
change of the air-sea interaction regime in hurricane
conditions is associated with the mechanism of direct
disruption of the air-sea interface by pressure fluctuations
working against surface tension.
• This disruption can be achieved through the KelvinHelmholtz (KH) or Tollmien-Schlichting (TS) instability and
leads to formation of two-phase transition layer.
• The transition layer has been related to the lower bound
on the air-sea drag coefficient in hurricane conditions in
an earlier work (Soloviev and Lukas 2010).
Direct Disruption of the Air-Sea
Interface
A non-dimensional number,
K  ua /  g s  w / a

2 1/ 4
which we call here the Koga number, is the criteria for the K-H
instability at an interface (Soloviev and Lukas 2010).
The instability occurs at
K > Kcr, where Kcr ~ 0.26
(corresponding to U10 ~ 30 m s-1).
In this formula, u*a is the friction velocity from the air side, g the
acceleration due to gravity, s the surface tension, w and a are
the water and air density, respectively.
Numerical Simulations
• In order to demonstrate the possibility of the direct disruption of the airsea interface under hurricane conditions, we have used an idealized 3D
model set-up.
• A series of numerical experiments has been conducted using the
computational fluid dynamics software ANSYS Fluent.
• Wind stress was applied at the upper boundary of the air layer, ranging
from no wind stress to hurricane force wind stress.
Disruption of the air-water interface
due to the K-H type instability
Wind stress 4 N m-2
Elapsed time = 2 s
10
cm
The numerical experiment with an initially flat interface illustrates
the possibility of the direct disruption of the air-water interface due
to the K-H type instability and formation of the two-phase
environment under hurricane force winds.
Soloviev, Fujimura and Matt,
submitted to JGR-Oceans
The numerical experiment with
imposed short waves
Wind stress 4 N m-2
Elapsed time = 0.5 s
The numerical experiment with imposed short waves
demonstrates the tearing of wave crests, formation
of water sheets and spume ejected into the air
Soloviev, Fujimura and Matt,
submitted to JGR-Oceans
Averaged vertical density and velocity profiles at
the air-sea interface from CFD simulation
For air-water density difference,
Boussinesq approximation is
no longer valid:
  log  
g 
N 
 g
 z
z
2
Ri  N /   u /  z 
2
2
Ricr  0.43
“Theoretical” value for
non Boussinesq: Ricr = 1/2
Cushman-Rosin (1994)
Linear profiles for log  and u follow from dimensional analysis:
 log   /  a   /  z  c gu2 , u / z  cu gu1
Schematic representation of density and velocity
profiles in the atmospheric and oceanic boundary
layers under hurricane conditions
The density profile in the atmospheric boundary layer is assumed to obey
the log layer law as well but cannot be seen in this diagram scale.
The CFD model has provided a better estimate for the lower
bound on the air-sea drag coefficient in hurricane conditions
potential local
minimum in Cd?
The two-phase layer resistance and wave resistance parameterizations in
comparison with available laboratory and field data. Transition to
hurricane force wind is associated with the drop of the drag coefficient,
which then may slowly increase with wind speed.
Thickness of the two-phase transition
layer as a function of wind speed
The two-phase environment suppresses short waves. This
effect can be included in the wave model via the condition:
kcutoff H » C
kcutoff is cut-off wavenumber
H is thickness of the transition layer
C is dimensionless constant of the order of 1.
Conclusions
• Change of the air-sea interaction regime in hurricane
conditions can be linked to the effect of direct disruption of
the air-sea interface and formation of a relatively thin twophase transition layer
• Analysis including computational fluid dynamics
experiments has provided an estimate for the lower bound
on the air-sea drag coefficient Cd in hurricane conditions
assuming regime of marginal stability in the transition layer
• An ad hoc “sweet spot” type parameterization for Cd can be
implemented
• Two-phase environment can be included into wave model
Summary
Coupled Atmosphere-Wave-Ocean Framework
Red - atmospheric parameters, Green – wave parameters, Blue - ocean parameters
• Hurricane model: air-sea fluxes depend on sea state, sea spray and include surface
current.
• Wave model: forced by sea state dependent wind forcing and includes surface current
• Ocean model: forced by sea state dependent wind stress modified by growing or decaying
wave fields and Coriolis-Stokes. Turbulent mixing is modified by the Stokes drift (Langmiur
turbulence).
Year 3 Plans for the URI Co-PIs
• Implement the new URI and UM sea state momentum
flux parameterizations into the air-sea interface module
(ASIM)
• Implement the effect of Coriolis-Stokes forcing into ASIM
• Refine the wave-driven sea spray parameterization and
implement it into ASIM (in collaboration with ESRL)
• Insure that all components of ASIM are modular, so the
same codes can be moved between the coupler and
component models as needed and different coupled
atmosphere-wave-ocean models
Year 3 Plans for the URI Co-PIs
• Implement and test ASIM into the HWRF-WAVEWATCH
III-POM/HYCOM system (in collaboration with EMC)
• Transition components of ASIM into the COAMPS TCWAVEWATCH III-NCOM system (in collaboration with
NRL)
• Run test simulations with HWRF-WW3-POM/HYCOM as
part of HFIP Stream 2 and evaluate the model results
against available observations
• Assist in transitioning the HWRF-WW3-POM/HYCOM to
the research community via DTC