JHT Project: Evaluation and Improvement of Ocean Model Parameterizations for NCEP Operations

L. K. Shay, G. Halliwell and B. JaimesMPO/RSMAS, Univ. of Miami

AOML/PhODhttp://isotherm.rsmas.miami.edu/~nick

Outline:

• Cold and Warm Core Eddies -Ivan (04) • Loop Current/Warm Core Eddy Interactions (deep OML) 1. Katrina and Rita (05) (Idealized Simulations) 2. Gustav and Ike (08)• Preliminary Comparison to Navy/NOAA Models Using DWH

T(z) Data (Assimilation Versus Non Assimilation)• 2011 HFP with NOAA IFEX/ DOI BOEMRE/NSF• Summary

Recurrent Theme: 3-D Upper ocean observations are critical in understanding

the response and interactions with atmosphere.

Ivan (2004) over the GOM SSH (cm) from HYCOM (from Halliwell et al., MWR, 2008).

NRL SEED Mooring data provided by Bill Teague (NRL-Stennis)

SST Analyses

Northern Cyclone

Southern Cyclone

Hurricane Ivan Modeling Experiment Summary

Model Attribute Baseline Experiment GOM1 Alternate Experiments

Horizontal resolution 0.04° Mercator GOM2: 0.08° Mercator

Vertical resolution 26 layers, 4-8m in OML GOM3: 21 layers, 7.5-15m in OMLGOM4: 31 layers, 3-5m in OML

Vertical mixing KPP GOM5: MYGOM6: GISS

CD Donelan GOM7: PowellGOM8: Large and PondGOM9: Large and Pond (capped)GOM10: Shay and Jacob

CEL, CES COARE3.0 algorithm GOM11: Kara et al.

Atmospheric forcing 27-km COAMPS+H*WIND GOM12: 27-km COAMPS only

Outer model NCODA GoM hindcast GOM13: Free GOM simulation

Ocean Dynamics 3-D Ocean Physics GOM14:1-D Column Models

Observed

KPP (control)

Mellor-Yamada

GISS

Vertical Mixing

Conclusion: KPP mixingproduces the mostrealistic shear response.Other two choices under-estimate entrainment,which reduces mixedlayer deepening andSST cooling.

Drag Coeff. Parameterization

Observed

Powell et al.

Donelan et al.(control)

Large and Pond

Conclusion: parameter-izations that yieldcoefficients between2 x 10-3 and 2.6 x 10-3

at high wind speedsproduce the mostrealistic shearresponse. Larger(smaller) coefficientValues increase(decrease) mixed layerdeepening and SSTcooling

Ivan Model Summary

Ivan a clear example of negative feedback (wake cooling/mixing induced by strong winds and Cold Core Ring) as opposed to positive feedback over the Loop Current and Warm Core Rings.

Relative to the NRL ocean measurements in Ivan:• Simulated shear response is a stringent test of model performance

– KPP mixing outperforms Mellor-Yamada and GISS– Drag coefficient parameterizations that yield values between 2.0 x 10-3 and

2.6 x 10-3 (e.g. Donelan et al.) outperform those that yield larger or smaller values

– No need to use vertical resolution of < 10 m with KPP mixing– Same conclusions realized at other SEED moorings (not shown)

• Model evaluation must be extended to other storms and to differing ocean regions Katrina and Rita (05); Gustav and Ike (08); Frances (2004)

The 2005 Loop Current System experimentThe 2005 Loop Current System experiment

twice the radius of maximum winds

Moorings

CCE2

LC

• Seawater temperature and conductivity data from four airborne surveys.

• Mooring data: velocity, temperature, and conductivity at ~10 m intervals.

CCE1

WCEFor the first time, the ocean response to major hurricanes was simultaneously measured inside cyclonic and anticyclonic geostrophic features.

cluster-averaged temperature profiles (Hurricane Rita)

Loop Current bulge: T ~ -1oC

26oC

26oC

CCE: T ~ -4.5oC

TC-induced cooling in geostrophic eddies

Jaimes and Shay, MWR, 2009

Modelapproach

• Isopycnic model (MICOM) to suppress spurious vertical mixing.

• Idealized, though very realistic quasi-geostrophic vortices.

• Turbulence closure for the oceanic mixed layer (OML) that considers wind erosion and vertical shear at the OML base and in the stratified ocean underneath.

• Constant wind stress fields from direct measurements during Katrina.

WCE1

WCE2

CCE1

CCE2

Jaimes et al. 2011 (JPO: in review)

• 47 isopycnic layers.

• Initial temperature and density fields from WP-3D data acquired after Katrina.

• High vertical resolution below the OML and over the thermocline (10 m, similar to vertical resolutions in MMS moorings).

CCEWCE

ObservationsModel

Jaimes et al. 2011 (JPO: in review)

Model Characteristics Model Characteristics

OML coolingOML cooling

Jaimes et al. 2011 (JPO: in review)

• Realistic OML cooling of less than 1.2oC in WCEs, and of about 4oC in CCEs.

• More intense OML cooling over more energetic WCEs and CCEs (right panels).

• More and less intense OML cooling over CCEs and WCEs, respectively, compared with quiescent background conditions.

WCE1

CCE1

WCE2

CCE2

CCE1 CCE2

Stationary vortices (f-

plane)

Freely propagating vortices (β-plane)

Quasi-geostrophic vortices transport the temperature anomaly Quasi-geostrophic vortices transport the temperature anomaly caused by tropical cyclones, distorting the cold wake.caused by tropical cyclones, distorting the cold wake.

Jaimes et al. 2011 (JPO: in review)

Black circles represent the initial position of the vortices.

• Horizontal convergence of warm water under the storm’s eye.

• Wind erosion occurs over a warm, deep, and nearly homogeneous water column.

• Kinetic energy injected by the tropical cyclones is rapidly dispersed, which reduces vertical shears and mixing in the upper ocean.

WCEsΔT < -1oC

• Upwelling of cold water under the storm’s eye.

• Wind erosion takes place over a a shallow OML and over a near-surface water column with sharp vertical thermal gradients.

• Kinetic energy injected by the storm is trapped in upper layers that increase vertical shears and mixing.

High quality ocean observations are critically important for model evaluations (e.g., NOAA IFEX, BOEMRE summer)

CCEsΔT ~ -4oC

Katrina and Rita Model SummaryKatrina and Rita Model Summary

T(z) Data Assimilation Into Models

Taylor Diagram for T(z) over All 9 Flights (30-360 m)

NOAA WP-3D profiling over BOEMRE Moorings (NOAA IFEX, HFIP)

Expendables deployed from P-3 and proposed moorings relative to the LC, WCE and CCE superposed on late Sept 05 altimetry derived 26oC isotherm Depth (After Rita).

Deliverables include:

V, T, S profiles to 1000 m @ 2-m resolution.

Surface winds (SFMR, GPS) provided by HRD.

Atmospheric profiles of V, T and RH @ 5-m resolution.