AtmosphereAtmosphere--OceanOcean InteractionInteraction inin TropicalTropical CyclonesCyclones

IsaacIsaac GinisGinis UniversityUniversity ofof RhodeRhode IslandIsland

Collaborators: T. Hara, Y.Fan, I-J Moon, R. Yablonsky.

ECMWF, November 10-12, 2008 AirAir--SeaSea InteractionInteraction inin TropicalTropical CyclonesCyclones

Two U.S. operational hurricane prediction models are coupled with ocean models: GFDL (since 2001) and HWRF (since 2007)

. CriticalCritical AspectsAspects ofof TCTC--OceanOcean InteractionInteraction

• Accurate initialization of ocean mesoscale features.

• Dynamical and microphysical processes near and at the sea surface that influence air-sea momentum and heat fluxes. AccurateAccurate OceanOcean InitializationInitialization ofof MesoscaleMesoscale FeaturesFeatures

Hurricane Katrina GFDL Model Forecast: Initial time August 26, 00 UTC, 2005

SST and Surface Current Temperature and Current at 75 m

Feature-based ocean initialization assimilates satellite altimeter, and in situ data in the (Yablonsky and Ginis 2008) implemented operationally GFDLGFDL HurricaneHurricane KatrinaKatrina Forecast:Forecast: SSTSST andand SurfaceSurface CurrentCurrent

72 h Forecast 96 h Forecast HurricaneHurricane KatrinaKatrina Forecast:Forecast: ModifiedModified LCLC andand NoNo WarmWarm--corecore RingRing

SST and Surface Current Temperature and Current at 75 m HurricaneHurricane Katrina:Katrina: ModifiedModified LCLC andand nono WarmWarm--corecore RingRing

72 h Forecast 96 h Forecast HurricaneHurricane KatrinaKatrina IntensityIntensity ForecastsForecasts

Central Pressure Maximum Winds

Green– Real-time forecast Blue – Modified LC and no warm-core ring Black - Observations 3D3D vs.vs. 1D1D OceanOcean CouplingCoupling

• Some recent studies (Emanuel et al. 2004; Lin et al. 2005, 2008; Bender et al. 2007; Davis et al. 2008) suggest that coupling a 1D ocean model to a TC model may be sufficient for capturing the storm-induced SST cooling in the region providing heat energy to the TC.

• If a 1D model is sufficient, valuable computational resources can be saved as compared to coupled models that employ a fully three-dimensional (3D) ocean component. A Prescribed translation speed T M Hurricane O vortex S P H E R E

O Homogeneous initial SST C E A N Vary position of ring relative to storm track

Warm core ring evident in subsurface temperature field C0N0R2L2 DifferenceDifference inin SSTSST ResponseResponse UnderneathUnderneath thethe TCTC CoreCore UsingUsing aa 1D1D andand aa 3D3D VersionVersion ofof thethe SameSame OceanOcean ModelModel

Hurricanes have historically translated at < 5 m s-1 73% and < 2 m s-1 16% of the time in the Gulf of Mexico

at < 5 m s-1 62% and < 2 m s-1 12% of the time in the entire western tropical North Atlantic AlongAlong--tracktrack temperaturetemperature crosscross--sectionssections

3‐D

1.0 m s‐1 2.4 m s‐1 4.8 m s‐1 1‐D SST and currents 6 hrs after storm passes WCR center longitude: 3-D experiments C0 N0

WCR

Storm Storm

Speed = 2.4 m s‐1 L2 R2 WCR

Storm Storm WCR SST and currents 6 hrs after storm passes WCR center longitude: 1-D experiments C0 N0

WCR

Storm Storm

Speed = 2.4 m s‐1 L2 R2 WCR

Storm Storm WCR SST cooling within 60-km radius of storm center 3-D 1-D R2 N0

N0

R2 Mean ΔSST Mean ΔSST Speed = 2.4 m s‐1 > > 3-D WCR 1-D WCR

< <

Storm Storm

R2 SST – N0 SST at +6 hours R2 SST – N0 SST at +6 hours ConventionalConventional couplingcoupling betweenbetween TCTC modelsmodels andand oceanocean modelsmodels Model Momentum flux (τ) Wind speed (Ua ) Sensible heat flux (QH ) Temperature (Ta ) Latent heat flux (QE) Humidity (qa) (at 10 m height) Air-Sea Interface Momentum flux ( ) Surface current (U ) τ s (Kinetic energy flux) SST (Ts ) (stability correction Ocean Model omitted for simplicity) τ = ρ ( −UUC ) C Drag Coefficient saDa D Heat Transfer Coefficient C = ()− ( −TTUUCQ sasaHH ) H Humidity Transfer Coeff. C LV E (Dalton Number) QE = ()()−− qqUUC sasaE CP (CBLAST) CE CD

Black et al. (2007)

/ CC Black stars: CBLAST data DE Black dash: HEXOS data Green: COARE 3.0 (Bulk) Observations of exchange coefficients WavesWaves GeneratedGenerated byby TropicalTropical CyclonesCyclones

WAVEWATCH III wave model TSP ) Wave Field (m calculations 0m/s Hs Wind Field (m/s)0m/s TSP = 5m/s 0m/s10m/s TSP = 5m/s 10m/s )

TSP (m 5m/s Hs

TSP )

10m/s (m Hs CoupledCoupled WindWind--WaveWave ModelModel

• Near the peak : WAVEWATCH III (WW3) model. • High frequency part : Equilibrium Spectrum model of Hara and Belcher (JFM, 2002)

Near peak High frequency part Full wave spectrum

kmin

k 3 Wave Boundary Layer model k 2 of Hara and Belcher (JPO, 2004) k1

k0 kv k max Kukulka and Hara (JPO, 2008)

Δk Δk Δk 2 1 0 Explicitly calculates 25 3 2 1 0 -1 -max wave-induced stress

Wind profile and drag coefficient over any given seas SeaSea statestate dependencedependence5

4 Constant zch (0.0185)

Extrapolation of Large & Calculations are based on Pond (1981)’s formula observed hurricane winds in the 3 Atlantic basin. Upper and lower CD * 1000 bounds of Cd from At high wind speeds, Cd levels all earlier exp. 2 off and even decrease with wind speed

1

Powell et al. (2003) Large & Pond (1981) GPS sonde observation under various WW3 (upper and lower bound) Bulk, 0.0185 hurricanes (Powell et al., 2003). Our results (upper and lower bound) 0 20 30 40 50 U10 (m/s) CharnockCharnock CoefficientCoefficient vs.vs. WindWind SpeedSpeed andand InputInput WaveWave AgeAge

Red indicates the range of realizable input wave age for given wind speed

Input wave age is one of the output parameters of WW3 and is a measure of the development stage of locally wind forced waves, excluding the effects of long swell and waves that are misaligned with the local wind SeaSea StateState DependenceDependence inin CoupledCoupled TCTC--WaveWave ModelModel

Zo Cd m

m/s m/s

Input Wave age

m/s WindWind--WaveWave--CurrentCurrent InteractionInteraction

Wind

Sea state τr air r Atmosphere Current τ air Surface waves Ocean τr Current r r c τ ≠ τ cair Ocean currents EnergyEnergy andand MomentumMomentum FluxFlux BudgetBudget AcrossAcross AirAir--seasea InterfaceInterface

EFair = EFc +c (EF growth + EFdivergence)

r r r r τ =τ cair +c ( τ growth +τ divergence )

Horizontal wave Wave growth ( EF , τ r ) propagationr growth growth (EFdivergence ,τ divergence )

Ocean Currents c c WindWind--wavewave--currentcurrent interactioninteraction

™ Ocean response

W at 90 m τair Currents WindWind--wavewave--currentcurrent interactioninteraction ™ Differences in Momentum Flux τc in A / τair in Control τair in B / τair in Control

Exp A Exp B

Budget No budget No current Current →wind →wind No current Current →wave →wave

τair in C / τair in Control τc in D / τair in Control Exp C Exp D

No budget Budget Current Current →wind →wind No current Current →wave →wave EffectEffect ofof WindWind--WaveWave--CurrentCurrent InteractionInteraction onon DragDrag CoefficientCoefficient

Wind-wave coupling only Wind-wave-current coupling

Cd Cd EffectEffect ofof AtmosphereAtmosphere--WaveWave--OceanOcean InteractionInteraction onon TCTC forecastsforecasts (Idealized experiments)

Central Pressure Maximum Winds

Track AirAir--SeaSea CouplingCoupling StrategiesStrategies forfor TropicalTropical ModelsModels

•• InIn thethe TCTC modelmodel,, parameterizationsparameterizations ofof thethe airair-- seasea heatheat andand momentummomentum fluxesfluxes andand seasea sprayspray sourcesource functionsfunctions explicitlyexplicitly includeinclude SST,SST, seasea statestate dependencedependence,, andand effects..

•• TheThe wavewave modelmodel isis forcedforced byby a)a) seasea--statestate dependentdependent momentummomentum fluxflux andand includesincludes oceanocean currentcurrent effectseffects.

• The ocean model is forced by sea-state dependent momentum and kinetic energy fluxes calculated from the air-sea flux budget. SummarySummary

• Accurate initialization of ocean mesoscale features is critical for skillful coupled TC-Ocean forecasts.

• By neglecting upwelling 1D mixed layer models are inadequate for TCs translating at 5 m/s or less and misrepresent TC-induced SST cooling in the vicinity of oceanic fronts and eddies.

• Improved predictions of TC intensity, structure, and motion will require fully coupled ocean-wave- atmospheric models that explicitly resolve the effects of sea state on air-sea fluxes and spray generation.