U.S. Navy Global Ocean Prediction Update

Key Performers: A.J. Wallcraft, H.E. Hurlburt, E.J. Metzger, J.G. Richman, J.F. Shriver, P.G. Thoppil, O.M. Smedstad, L. Zamudio

2011 GODAE OceanView Technical Workshop onObserving System Evaluation and Inter-comparison

13-17 June 2011Santa Cruz, CA

Goals of This Project

Increase to 1/25 resolution globally (~3-4 km mid-latitude)

- Internal/external tides

- Other sub-daily effects

Provide direct boundary conditions for coastal models

- Eliminates need for regional models

Won’t run in near real time until FY12

- On next system at Navy DSRC

Test new capabilities at 1/12

- Transition some to operational use at 1/12

Modeling tides in the global model

• In the global model, the body forces due to the tidal potential, self attraction and loading have been added

• Tidal Forcing with 8 constituents:– Semidiurnal M2, S2, N2 and K2

– Diurnal O1, P1, Q1 and K1

• Topographic wave drag is applied to the tidal motions– The form of the drag is generalized from the

linear topographic wave drag, but tuned to minimize the difference with the 102 pelagic tide gauges using a barotropic version of the model

Comparison of M2 tide from Data-Assimlative Shallow Water Model (TPXO7.2) and HYCOM Simulation

TPXO7.2 M2 Tidal Model HYCOM M2 Tide

Difference with 102 pelagic tide gauges 7.8 cm rmsDifference with TPXO7.2 model 5.4 cm rms

cm

M2 internal tide amplitude: along-track altimetrydata vs 1/12 Global HYCOM

Altimetry-based analysis

HYCOM

Analysis of along-track altimetry data based upon extracting the aliased tidal signal from the repeat-cycle data as described by Ray and Mitchum (1996, 1997)

The internal tide is recovered by a spatial filter to separate the barotropic and baroclinic tides

Atlimeter analysis provided by Richard Ray

HYCOM SSH analyzed for tides, sampled along track and filtered similar to altimeter

Altimetry-based analysis

cm

Altimeter Data HYCOM M2 0.5649 0.6512 S2 0.2997 0.3897 N2 0.1497 0.1197 K2 0.0918 0.1084

HYCOM

M2 internal tide amplitude: along-track altimetrydata vs 1/12 Global HYCOM

Large Scale Prediction GOFS V3.5

1/25° non-assim. global HYCOM

1/12° non-assim. global HYCOM

Mean SSH 2004-2009

Forced with 0.5° NOGAPS winds and heat fluxes

Large Scale Prediction GOFS V3.5

1/25° non-assim. global HYCOM

1/12° non-assim. global HYCOM

SSH Variability (cm) 2004-2009Satellite observations from CLS

Hindcast NCODA Synthetics CICE

1 MVOI MODAS No

2 3DVAR MODAS No

3 3DVAR ISOP No

4 3DVAR ISOP Yes

1/12° NCODA(3DVAR) / ISOP / CICE Hindcast Matrix

Hindcasts 1-3 will cover approximately a three month period and incrementally build upon each other. It is anticipated that hindcast four will be the final configuration of GOFS V3.1 and so this will be integrated an entire year.

Plans for 1/12° Global Forecast System (GOFS 3.1)

Recent Progress

New 1/12 and 1/25 bathymetries

Based on 30 (1/120 ) GEBCO

New tidal drag based on 30 GEBCO

Testing 3D-VAR data assimilation in place of MVOI in NCODA

Initial 1/25 with tides case

Initial 1/25 with data assimilation

Status

Expect to be ready for real time 1/25 data assimilative system when new Navy DSRC computer is available in FY12

FY12 Plans for 1/25° System (GOFS 3.5)

• Near real-time 1/12° global HYCOM/NCODA prediction system output (Nov 2003-present)

• 1/12° global non-assimilative HYCOM output (2003-2006)

• 1/12° North Atlantic data assimilative HYCOM output (2003-2008)

• 1/12° Pacific non-assimilative HYCOM output (1979-2003 – monthly means)

Variables remapped to 33 standard levels: 3D – T, S, U, V, ρ; 2D – SSH, SST, SSS, MLD, …

HYCOM Data Servinghttp://www.hycom.org/dataserving

1/12º Global HYCOM

Questions?

ISOP corrects climatological or model backgrounds based on climatologically observed modes of variability

Background from climatology

or model forecast + uncertainty

in situ observations

+ uncertainty

Remote sensing (SST, SSH)

+ uncertainty

Mixed-layer depth estimate

+ uncertainty

Synthetic profile of T,S or

corrections to background

ISOP

• EOFs of T, S• Model mixed layer, deep layer, transitions

• minimize variance among background, observations

Eddy Kinetic Energy at the Surface

Eddy Kinetic Energy (cm2/s2)

Eddy Kinetic Energy at the Surface in the Gulf Stream

The 1/25 model has the highest EKE with a larger southern recirculation gyre

Data assimilation brings the EKE close to the observations from the drifters

Summary of Surface EKESummary of Surface EKE

74.2, (1/12)DA

04.2, (1/25)FR

18.2, (1/12) FR

18.5, (1/12)FR-T

2008-09 2005-07 2005-09 2005-09

Surface

Observed (Drifter) Mean EKE = 418.2; KEMM = 135.4

Mean EKE 395.4 439.1 348.3 346.3

KEMM 145.3 179.9 155.2 156.0

EKE Correlation

0.769 0.809 0.804 0.802

Increasing the resolution or assimilating data increases the EKE to levels comparable to the drifter observations

Increasing the resolution increases the KEM while assimilating data decreases the KEM

Part of the doubled resolution EKE increase may be associated with surface quasigeostrophic motions

Interannual variability about 15 cm2/s2 in EKE and 20 cm2/s2 in KEM

Why does the 1/25° Model Perform so much better?

• Much emphasis for the global models has been focused on resolving the eddy scales (Rossby radius of deformation)

• The dynamics are governed by the energy and enstrophy cascades and interactions.– In the surface enstropy cascades to small scales and energy

towards the Rossby radius, where it is transferred in the vertical (barotropization)

– The barotropic eddy kinetic energy can be transferred to the mean (a mechanism for the downstream increase in the Gulf Stream)

– The enstropy cascades are a critical part of this cycle, which are resolved better in the higher resolution model

– Increasing the eddy kinetic energy in the surface by data assimilation improves the model but does adequately resolve the enstropy cascades

1/12°

1/25° – 1/12°

1/12° Tides – 1/12° 1/12° DA – 1/12°

Differences in Surface Eddy Kinetic Energy (cm2/s2)

Surface Eddy Kinetic Energy

Deep Eddy Kinetic Energy in the Gulf Stream Region

What happens in the simulations?• We find that the resolution of the present generation of ocean general

circulation models of ~1/10° is inadequate to establish a vigorous abyssal circulation and the surface eddy kinetic energy (EKE) is only about 85% of the observed.

• Adding tidal forcing has a minimal impact on the surface circulation but increases the deep EKE by 12% and the deep KEM by 25%.

• Doubling the horizontal resolution of the model increases the surface EKE to levels comparable to the drifter observations and increases the KEM by 40%, which is greater than the drifter estimates. The deep EKE and KEM also are increased to levels consistent with the deep current meters.

• Data assimilation increases the surface EKE to levels consistent with the drifter observations and increases the deep EKE and KEM. Surprisingly, data assimilation weakens the KEM at the surface and upper thermocline to levels below the 1/12° simulation.

Mean Kinetic Energy at Surface (cm2/s2)

Kinetic Energy of the Mean Flow at the Surface

1/12°

Drifters

Large Scale Prediction GOFS V3.5

Maximenko and Niiler

1/25° non-assim. global HYCOMMean SSH2004-2009

Large Scale Prediction GOFS V3.5

1/25° non-assim. global HYCOM

1/12° non-assim. global HYCOM

Mean SSH 2004-2009

Forced with 0.5° NOGAPS winds and heat fluxes

Large Scale Prediction GOFS V3.5

1/25° non-assim. global HYCOM

1/12° non-assim. global HYCOM

SSH Variability (cm) 2004-2009Satellite observations from CLS