CBLAST Hurricane Modeling Update - Summary of Recent Results
June 13, 2003
PI's: Michael L. Banner and Lance M. Leslie
The University of New South Wales, Sydney, Australia
Background
A major part of our effort to date has been on evaluating and refining the performance of a new spectral
dissipation rate source term, together with the modeling of spectral and statistical properties of breaking
waves. We are also working on the incorporation of a refined wind input source function for very
strongly forced conditions from our allied ONR Lake George project (Donelan, Babanin, Banner and
Young). These developments are all central to our modeling of breaking wave enhancements of the
basic air-sea fluxes, such as the wind stress, in our development of a more physically-based coupled
atmospheric-spectral wind wave modeling for hurricane conditions. Our effort also includes the
formulation of a physically-based spray/spume source function that depends on the computed sea state,
particularly the wave breaking activity, rather than the wind speed alone.
A. Refined spectral dissipation rate and wind input terms
Recently, we have concentrated on optimising our new spectral-saturation-based threshold form of the
dissipation rate source term Sds, in the context of spectral evolution calculations using the exact form of
nonlinear transfer Snl. This new Sds term extends the groundwork of Alves and Banner (2003) by
incorporating recent advances in predicting wave breaking probabilities at different wave scales (Banner
et al., 2002). That study reported a high correlation of breaking probability with the spectral saturation
B = k4 (k) = (2)4f 5F(f)/2g2, for wave scales from the spectral peak frequency fp out to 2.5fp, and
revealed a very strong threshold behavior. After normalization to offset the increasing directional
spreading of the waves with increasing f/fp, the saturation breaking threshold was found to be nearly
constant from fp out to 2.5fp. Our refined Sds now incorporates the observed breaking saturation
threshold.
Extensive testing for 10m/s winds (where most of the data observations are available) indicated that this
representation provides greater flexibility and improved performance in modeling fetch-limited wind
wave evolution. In recent months we have been pushing the model towards hurricane wind speeds, but
encountered stability problems at these higher wind speeds. After considerable effort, the problem was
diagnosed recently and is no longer an obstacle to progress.
Some recent results are shown in the Appendix. Computations of fetch-limited evolution have been
done with full spectral bandwidth out to 1.5 Hz waves for 10 m wind speeds of 10 and 15 m/s. Both runs
compare very favourably with standard energy and peak frequency evolution curves. The 10 m/s runs
were used to extract (c), the spectral breaking crest length in (c, c+dc). We have been using the
spectral wave breaking data from aircraft video imagery (Melville and Matusov, 2002) to test our model
predictions and found our calculations are in remarkably close agreement with the observations to date.
In the context of the forthcoming CBLAST hurricane measurement program, the most useful validation
data for the our modelling of spectral breaking wave properties are: E. Walsh's SCR directional wave
spectra together with K. Melville's spectral wave breaking spectral data from aircraft video imagery.
2
The planned concurrent surface flux measurements will be essential validation data for the fluxes
predicted by our coupled model, described below.
B. Coupled wind-wave model with explicit wave breaking impact.
It is known from laboratory and recent field observations that wave breaking can enhance both the
momentum flux from the wind to the wave field and the overall wind stress. We have been developing a
code to allow the iterative computation of the wind stress and wave spectrum in response to changes due
to breaking-induced wave drag. Our approach is based on the calculation of wave breaking properties
from the wave spectrum, the consequent enhanced wave drag and adjustment of the aerodynamic
roughness length z0 in the logarithmic mean velocity profile for the surface layer wind field. In turn, the
wave spectrum is modified in response to the updated wind profile. This is currently being integrated
into our coupled test-bed model based on fetch-limited growth. With the availability of the latest results
from the Lake George, and soon to be finalised Showex experiments, we are looking closely at
optimising the choice of wind input source function, Sin.
C. Sea spray source function
Joint research on refining sea spray/spume parameterization was initiated at the inaugural CBLAST PI
meeting in Warrenton, Virginia (January 2001) with C. Fairall (NOAA, Boulder, Co) and W. Asher
(APL, U. Washington). This resulted in a laboratory experiment conducted in January/February 03 at
the University of NSW Water Research Laboratory aimed at developing a refined model for spume
droplet production based on sea state (particularly wave breaking properties) rather than on wind speed
alone. An initial wave breaking/turbulent flow model of spume droplet production has been developed
by C. Fairall and one of the aims of the experiment is to validate/refine this formulation. The data
analysis is partially completed and these results should be available within the next few months. Our
aim is to be able to incorporate the new sea spray parameterization into our coupled model as soon as
possible.
D. Operational model implementation
A background effort has been the acquisition of the official release of COAMPS(TM)3 and
WaveWatchIII, for atmospheric and sea state prediction. Due to differences in the implementation of the
Message Passing Interface (MPI) parallel protocol, a substantial effort has gone into coupling of these
models. WaveWatchIII is run routinely at UNSW and the new implementation of COAMPS3 is being
implemented.
Information needed from other CBLAST modelers:
Much of the modeling effort is on generating robust model strategies for realistic hurricane evolution,
particularly intensity, structure and track. We would like to be kept informed on the related key issues
such as model sensitivities to assumed air-sea flux parameterisations, as well as any other coupling
strategies being developed or implemented.
Data needed from the field program, coordination with the observationalists.
At this stage, we particularly need:
 directional wave spectra, as far into the spectral tail region as possible, to monitor spectral level,
spreading and saturation.
 breaking wave data, especially breaking probabilities at all scales and spectral breaking crest
length per unit area
3
These needs have been foreshadowed to Ed Walsh and Ken Melville.
Allied data sets essential to our model validation are:
 wind stress and heat fluxes with concurrent sea state data. We will be coordinating initially with
C. Fairall on this aspect.
 sea spray data from the hurricane flights from C. Fairall and W. Asher, hopefully synchronized to
the above wave data sets.
References:
Alves, J.H and M.L. Banner, 2003: Performance of a saturation-based dissipation source term for wind
wave spectral modeling. J. Phys. Oceanogr. 33, 1274-1298.
Banner, M.L., J.R. Gemmrich and D.M. Farmer, 2002: Multiscale measurements of ocean wave
breaking probability. J. Phys. Oceanogr. 32, 3364-3375.
Hwang, P.A., D.W. Wang, E.J. Walsh, W.B. Krabill & R.W. Swift, 2000 Airborne measurements of the
wavenumber spectra of ocean surface waves. Part II: Directional distribution. J. Phys. Oceanogr., 30,
2768-2787.
Melville, W.K. and P. Matusov, 2002: Distribution of breaking waves at the ocean surface. Nature 417,
58-63.
Appendix
Typical computational results to date are shown below.
Fig. 1(a)
Fig. 1(b)
Fig. 1(a) and (b) show the fetch evolution of dimensionless energy and peak frequency for 10 m/s
winds, indicating the close correspondence with observations for both the growth evolution and
asymptotic level (bands indicate uncertainty range of Pierson-Moskowitz limit). This demonstrates the
flexibility of our new saturation-based Sds source function to produce the key integral wave energy
properties. Very similar results are obtained for 15 m/s winds.
4
Fig. 2(a)
Fig. 2(b)
Fig. 2(a) shows the azimuthally-integrated spectral saturation B (defined above) for U10=10 m/s and old
wind sea conditions, as a function of f/fp , in relation to the saturation data reported in BGF02. The
observations were made for growing wind seas, so the comparison is not ideal. Nevertheless, the
computation captures the levels and overall shape of the observations, aside from the reduced level at
the spectral peak due to the difference in conditions. Fig. 2(b) shows the saturation spectrum,
normalized by the local angular spreading width (Fig. 2(d), against k/kp, in relation to the common
breaking threshold saturation level of 0.005 observed in BGF02. This indicates the spectral regions
where breaking is active and how these change with wave age. The wind speed is U10 =10 m/s and the
inverse wave age is U10/Cp = 1. Computations made for 15 m/s winds show very similar properties.
Fig. 2(c)
Fig. 2(d)
Fig. 2(c) shows the directional spreading at different k/kp and shows a strong bimodal signature at the
higher k/kp values, in accordance with the recent observations of Hwang et al. (2000). This is attributable
to the strong influence of Snl in the tail region. Fig. 2(d) shows the close reproduction of the mean
directional spreading width with k/kp, from the model, and the Hwang et al. (2000) data for comparable
conditions (U10~10 m/s and U10/Cp~1). Again, computations made for 15 m/s winds show very similar
properties.
5
Fig. 3(a)
Fig. 3(b)
Fig. 3(a) compares the computed one-dimensional transect spectrum, (k1), with the observations of
Melville and Matusov (2002 and Vandemark (private communication). All spectra conform to the
expected k1-3 behavior, with the modeled spectral level is corresponding very closely to the data. Fig.
3(b) shows the first model computations of (c), the spectral density of breaking crest length per unit
area of sea surface, expressed as a function of the phase speed c for comparison with the observations of
MM02. The computed results are seen to follow very closely the level and spectral trend as measured by
MM02. The wind speed is 10 m/s, with an inverse wave age of 0.83 that matches the observational
conditions.