White cap measurements and parameterizations based on thedissipation source term
Ø Breivik, B Scanlon, J-R Bidlot, P A E M Janssen, AH Callaghanand B Ward
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 1 / 16
Outline of presentation
Whitecaps and their relation to the dissipation of ocean wavesParameterizing wave breaking with a spectral wave modelMeasuring whitecapsParameterizing whitecap coverageConclusions and further work
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 2 / 16
Motivation
Estimates of the energy dissipated from the oceanic wave field due to wavebreaking is needed for among other things
upper-ocean turbulence models
gas transfer estimatescalibrating remotely sensed estimates of sea surface temperature
Parameterizing white cap coverage and comparing it to observationsis an indirect estimate of our ability to model the dissipation frombreaking waves
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 3 / 16
Motivation
Estimates of the energy dissipated from the oceanic wave field due to wavebreaking is needed for among other things
upper-ocean turbulence modelsgas transfer estimates
calibrating remotely sensed estimates of sea surface temperature
Parameterizing white cap coverage and comparing it to observationsis an indirect estimate of our ability to model the dissipation frombreaking waves
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 3 / 16
Motivation
Estimates of the energy dissipated from the oceanic wave field due to wavebreaking is needed for among other things
upper-ocean turbulence modelsgas transfer estimatescalibrating remotely sensed estimates of sea surface temperature
Parameterizing white cap coverage and comparing it to observationsis an indirect estimate of our ability to model the dissipation frombreaking waves
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 3 / 16
Motivation
Estimates of the energy dissipated from the oceanic wave field due to wavebreaking is needed for among other things
upper-ocean turbulence modelsgas transfer estimatescalibrating remotely sensed estimates of sea surface temperature
Parameterizing white cap coverage and comparing it to observationsis an indirect estimate of our ability to model the dissipation frombreaking waves
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 3 / 16
Observing whitecaps
Cruise track R/VKnorr June-July2011. Day-of-yearis marked alongthe cruise trackand stations aremarked as redsquares.
80oW 72
oW 64
oW 56
oW 48
oW 40
oW
36oN
45oN
54oN
63oN
72oN
176 177 178
180181
182
183
184185
189190191
192
194
Newfoundland
Greenland
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 4 / 16
Observing whitecaps
Time series of neutral wind speed (U10N), modeled significant wave height(Hs), mean period of the modeled wind-wave spectra (T̄ ) during theNorth Atlantic campaign of 2011. The grey columns mark the periodsduring which whitecap coverage fraction was measured.
Day of year180 185 190 195
0
5
10
15
20
U10N
(
m s−1)
HS (m)T̄m1 (s)
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 5 / 16
Observing whitecaps
Whitecaps observed with 5 mega pixel 16 mm camera mounted onR/V Knorr.> 114,000 images were processed first with Automated WhitecapExtraction (AWE) algorithm (Callaghan et al, 2009)Secondly, manual inspection using the Spatial Separation of WhitecapPixels (SSWP) method (Scanlon and Ward, 2013) distinguishesactive breaking (stage A, blue) from decaying (stage B, green)
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 6 / 16
Wave model integration
The ECWAM wave model wasrerun for the cruise period with11-km spatial resolution. Wavefields are output with 1-hourlytemporal resolution.The wind was taken fromoperational analyses whichcompare well with the observedwind speed
U10N (m s−1)0 5 10 15 20
Umod
10N
(ms−
1)
0
5
10
15
20M=0.95, C=0.24, R
2=0.90
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 7 / 16
Parameterizing whitecaps from a spectral wave model
Craig and Banner (1994) assumed that the energy flux from breakingwaves
Φoc = ρwg∫ 2π
0
∫ ∞0
Sds dωdθ [Wm−2] (1)
is proportional to the cube of the friction velocity, ie,
Φoc ≈ ρwαCBw3∗ = ρamu3
∗ , (2)
where w∗ is the water friction velocity and typically 50< αCB < 150. Craigand Banner (1994) assumed αCB = 100 (or, equivalently, an air-sidecoefficient m ≈ 3.5)
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 8 / 16
Parameterizing whitecaps from a spectral wave modelThe ECWAM model integration for the cruise period shows that thisproportionality is a good first order approximation
u⋆ (m s−1)0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
(
Φoc
ρa
)
1/3(m
s−1)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
WFA=
0
WFA>
0
WFT = 0
WFT > 0
Φoc = 4.64ρa (u⋆)3, R2 = 0.99
Φoc = 0.013 W m−2 (HS08)u⋆T = 0.065 m s−1 (SW15)
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 9 / 16
Parameterizing whitecaps from a spectral wave modelBut the proportionality factor αCB is not constant, neither geographicallynor in time as it depends on the maturity of the sea state. A one-monthaverage shows that there are large geographical differences
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 10 / 16
Parameterizing whitecaps from a spectral wave model
Kraan et al (1996) parameterized the whitecap fraction (area covered bywhitecaps) from the flux Φoc as
W modF = Φoc
γρwg ω̄E , (3)
where ω̄ = 2π/T̄ is the circular mean frequency, γ a tuning factor andE = (Hs/4)2.
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 11 / 16
Parameterizing whitecaps from a spectral wave model
The relationship between observedand modelled whitecap coverage isquite impressive, but there is stillsome room for tuning. The factor γis here used to separately tune WFAand WF.
A: γ = 0.01 for both active andtotal whitecap coverageB: γ set separately for the activeand the total whitecap coverageimproves the fit somewhat. Thecorrelation remains unchangedat R = 0.88 for the totalwhitecap.
(
Wmod
F
)
1/3
0
0.1
0.2
0.3
0.4(a)
WFA
(γ=0.01)
WF
(γ=0.01)
M=1.5, C=0.033, R2=0.74
M=0.9, C=0.041, R2=0.78
WF1/3
0 0.1 0.2 0.3 0.4
(
Wmod
F
)
1/3
0
0.1
0.2
0.3
0.4(b)
WFA
(γ=0.034)
WF
(γ=0.007)
M=1, C=0.021, R2=0.74
M=1, C=0.045, R2=0.78
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 12 / 16
Conclusions
1 The manual Spatial Separation of Whitecap Pixels (SSWP) method isfound to accurately distinguish between stage A and stage Bwhitecaps
2 The Kraan et al (1996) parameterization is found to yield a goodparameterization of the total whitecap coverage
3 Rerunning the wave model improved the results compared with usingERA-Interim estimates with six-hourly resolution
4 The parameterization is straightforward to implement as it relies onintegrated parameters only
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 13 / 16
Further workMeasuring wave breaking continuously from a fixed platform in the CentralNorth Sea is planned (Norwegian WAVEMIX proposal)
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 14 / 16
Further workThe footbridge is located 22 m above still water level, facing N-NW, with alaser array already in place and a WaveRider buoy nearby.
References:Breivik, Ø, K Mogensen, J-R Bidlot, MA Balmaseda, and PAEM Janssen (2015). Surface Wave Effects in the NEMO OceanModel: Forced and Coupled Experiments, J Geophys Res Oceans, 120, doi:10.1002/2014JC010565, arXiv:1503.07677Callaghan, A. H., and M. White (2009). Automated processing of sea surface images for the determination of whitecapcoverage. J. Atmos. Ocean. Tech., 26(2), 383–394, doi:10.1175/2008JTECHO634.1Kraan, G., W. A. Oost, and P. A. E. M. Janssen (1996). Wave energy dissipation by whitecaps. J. Atmos. Ocean. Tech.,13(1), 262–267, doi:10/dnp59mScanlon, B., and B. Ward (2013). Oceanic wave breaking coverage separation techniques for active and maturing whitecaps.Methods in Oc., 8, 1–12, doi:10.1016/j.mio.2014.03.001Scanlon, B, Ø Breivik, J-R Bidlot, P Janssen, A H Callaghan, B Ward (2015). Modelling whitecap fraction with a wave model,J Phys Oceanogr, doi:10.1175/JPO-D-15-0158.1
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 15 / 16
20 years is a long timeKraan et al (1996)
Øyvind Breivik (MET Norway) Modelling whitecap coverage 7 September 2016 16 / 16