Overview of GOFS:
The U.S. Navy Global Ocean
Forecasting System
GOVST-V
Beijing, October 2014
GODAE OceanViewwww.godae-oceanview.org
Symposium, Hilton Baltimore, 4-6 November 2013
GOFS Descriptions
GOFS 3.0: 1/12° 32 layer HYCOM/NCODA-MVOI /MODAS synthetics/e-loan ice
• Pre-operational system that ran on Navy DSRC Cray XT5
• OPTEST report accepted by AMOP in Apr 2012
GOFS 3.01: 1/12° 32 layer HYCOM/NCODA-3DVAR/MODAS synthetics/e-loan ice
• Operational system running on Navy DSRC IBM iDataPlex computers
• Switch from MVOI to 3DVAR
• Switch from NOGAPS to NAVGEM 1.1 forcing in August 2013
• Switch from NAVGEM 1.1 to NAVGEM 1.2 forcing in March 2014
GOFS 3.1: 1/12° 41 layer HYCOM/NCODA-3DVAR/ISOP synthetics/CICE
• Add nine near surface layers
• Two-way coupled HYCOM with Los Alamos CICE model
• Replace MODAS synthetics with Improved Synthetic Ocean Profiles
• Validation test report completed
GOFS 3.5: 1/25° 41 layer HYCOM/NCODA-3DVAR/ISOP synthetics/CICE/tides
• Increase equatorial horizontal resolution to ~3.5 km
• Tidal forcing
• Scheduled to be operational in 2016-Q4
REANALYSIS 1993-2012: 1/12° 32 layer HYCOM/NCODA-3DVAR/MODAS synthetics/eloan ice
ESPC: T359 NAVGEM, 1/12.5º HYCOM, and CICE
Items in red are different from the preceding system
GOFS and Reanalysis outputs available at http://hycom.org
Large Scale Ocean Prediction
GOFS 3.5 Major Objectives & Milestones
MS Event/Action/Improvement Objective
Completion and/or Delivery
Quarter/FY
Description of Capability Completed and/or
Delivered
Expect completion 4QFY14
Calibrate NAVGEM 1.3 wind and heat flux forcing to
assure the underlying ocean model response is
consistent across the changeover from NAVGEM
1.2 to NAVGEM 1.3
Static calibration of NAVGEM 1.4 for
implementation in GOFS: assure consistent
ocean model response
Expect completion 4QFY15
Calibrate NAVGEM 1.4 wind and heat flux forcing to
assure the underlying ocean model response is
consistent across the changeover from NAVGEM
1.3 to NAVGEM 1.4
Time-evolving calibration of NAVGEM output
based on satellite data
Expect completion 4QFY15
Automatic on-the-fly calibration of NAVGEM winds
and heat fluxes
Implement infrastructure to separate tidal and
non-tidal forecasts
Expect completion 2QFY15
Separate tidal and non-tidal components of GOFS
forecasts for use in NAVOCEANO systems
Develop methods, software and scripts to
extract only required output as defined by
NAVOCEANO
Expect completion 2QFY15
Assist NAVOCEANO to better manage the
voluminous output from the 1/25° system
Complete GOFS 3.5 VTR and acceptance by VTP
Expect completion 2QFY16
Improvements over GOFS 3.1: 1/25° vs. 1/12°
horizontal resolution, implement tidal forcing
Static calibration of NAVGEM 1.3 for
implementation in GOFS: assure consistent
ocean model response
GOFS 3.1 vs. GOFS 3.0
GOFS 3.1: 41 layer 1/12° global HYCOM
NCODA-3DVAR
Los Alamos Community Ice CodE (CICE)
Improved Synthetic Ocean Profiles (ISOP)
GOFS 3.0: 32 layer 1/12° global HYCOM
NCODA-3DVAR
Energy loan ice
Modular Ocean Data Assimilation System (MODAS)
GOFS 3.1 vs. GOFS 3.0
• HYCOM changes
• Newer source code: v2.2.86 vs. v2.2.19
• Improved base bathymetry: 30” GEBCO vs ETOP05
• 0.1 m coastline vs. coastline at 10 m isobath
• Improved Equation of State: 17-term vs. 7-term
• Improved vertical structure: additional 9 layers near the
surface – mixed layers are typically better resolved
• Changes to surface momentum forcing: calculated in-line
in GOFS 3.1 using HYCOM SST and allowing for HYCOM
surface currents
• Improved ocean turbidity scheme: chlorophyll-based vs.
photosynthetically available radiation-based.
• Improved SSS relaxation for better river representation
GOFS 3.1 vs. GOFS 3.0
• Ice model changes
• Full two-way coupling between HYCOM-CICE via ESMF vs.
thermodynamic “energy-loan” ice model
• Ocean-ice coupling frequency is every hour
• Ocean-ice models share the same grid
• Full rheology in CICE vs. ice growth/melt in response to
temperature and heat fluxes and no ice advection by wind
or ocean currents
• GOFS 3.1 ice validation is compared against Arctic Cap
Nowcast/Forecast System (ACNFS), not GOFS 3.0
• Same version of CICE (v4.0) within GOFS 3.1 and ACNFS
GOFS 3.1 vs. GOFS 3.0
• NCODA-3DVAR changes
• NCODA analysis at 12Z using 24-hour HYCOM forecast
daily mean vs. NCODA analysis at 18Z using 24-hour
HYCOM forecast instantaneous field
• Daily mean will filter out the tidal signal when tides are
implemented in GOFS 3.5
• Data selection for assimilation at receipt time vs.
observation time
• Uses First Guess at Appropriate Time (FGAT) to account
for late arriving data
• Cap on the maximum Rossby radius of deformation
GOFS 3.1 vs. GOFS 3.0
• NCODA-3DVAR: downward projection of surface information
• Synthetic profiles are generated along altimeter tracks
• Use observed SSH and SST to create synthetic T & S
profiles based on historical relationships between surface
and subsurface observations
• GOFS 3.1 uses ISOP, GOFS 3.0 uses MODAS
• ISOP has been shown to be an improvement over MODAS
as it is able to better represent the vertical structure of
the ocean by also constraining vertical gradients of T & S
Hindcast Period - Ocean
• A GOFS 3.1 hindcast was integrated emulating a real-time
operational system
• It ran forward one day at a time but did not create multi-day
forecasts each hindcast day
• NAVGEM 1.1 forcing: 1 August – 31 December 2013
• NAVGEM 1.2 forcing: 1 January – 30 April 2014
• A series of 14-day forecasts were integrated to examine
medium-range forecast skill
• Initialized from operational GOFS 3.0 and hindcast GOFS 3.1 restarts and
forced with analysis quality NAVGEM 1.1 or 1.2 forcing
• The frequency of these forecasts was dictated by the availability of GOFS 3.0
restart files, typically on the 2nd and 16th of each month
• In total, fifteen 14-day forecasts were integrated (with no data assimilation)
Hindcast Period - Ice
• Important to span a complete year that covers both the melt
and freeze periods
• An ACNFS hindcast using NAVGEM 1.1 forcing existed that
spanned the period 1 June 2012 – 31 May 2013
• A GOFS 3.1 hindcast had also been integrated over this same
period but with an older ISOP configuration within NCODA
• While that particular GOFS 3.1 hindcast was deemed nonoptimal for the upper ocean response, ice nowcasts and
forecasts were not adversely affected and these two
hindcasts are used for validation
• A sequence of GOFS 3.1 5-day forecasts were also integrated
to examine ice error as a function of forecast length
Ocean Validation – Ocean Analysis Regions
Ocean Validation – Profile Data
• Only unassimilated observations are used in the analyses within
• For a given observation, both systems are sampled at the nearest
model grid point
• Both the observations and hindcast output are remapped in the
vertical to a common set of depths: 0, 5, 8, 12, 17, 25, 36, 50, 73,
100, 125, 150, 175, 200, 225, 250, 275, 300, 343, 400, 450, and 500 m
• Model-data differences that exceed 3 standard deviations are thrown
out, i.e. 99% confidence interval
• This is why the observation count may differ between the two systems
Ocean Validation – T vs. Depth – Nowcast
GOFS 3.1
GOFS 3.0
For the global
domain, mean error
is similar, but GOFS
3.1 has lower
RMSE than 3.0.
Regionally, the
results vary.
Mean error
Mean over
this depth range
RMSE
Number of
observations
Mean error
RMSE
Ocean Validation – S vs. Depth – Nowcast
Mean error
GOFS 3.1
GOFS 3.0
RMSE
Mean error
RMSE
For the global
domain, mean error
Is similar but GOFS
3.1 has lower
RMSE than 3.0.
Regionally the
results vary.
Ocean Validation – Mixed Layer Depth – Nowcast
Median
Bias
Error
GOFS 3.1
MdBE = -1.8 m
RMSE = 31.9 m
2° bins
Box size
indicates
observation
count
GOFS 3.0
MdBE = -3.1 m
RMSE = 36.7 m
Ice Validation – Ice Concentration and Thickness
GOFS 3.1
Concentration (%)
Thickness (m)
ACNFS
3 June 2014
Similar ice
thickness
except in the
Beaufort and
E. Siberian
Seas
Ice Validation – NH Ice Edge Location Error – Nowcast
Ice Validation – NH Ice Edge Location Error – Forecast
GOFS 3.1
ACNFS
Approximately 100 forecasts used in this analysis
Slow error
growth with
increasing
forecast
length
Ice Validation – SH Ice Concentration and Thickness
22 June 2014
Black line is independent NIC ice edge analysis
Ice Validation – SH Ice Edge Location Error – Nowcast
SH ice edge error is of similar magnitude to NH error
In-Line Wind Stress
 The bulk parameterization for wind stress
includes (Tair-SST) and 10m winds (U10)
 Usually calculated off-line on the NWP (e.g.
NAVGEM) grid using NWP SST
 Evidence that U10 should be replaced with (U10Uocn)
 HYCOM now has option to read in 10m winds
and calculate wind stress in-line
 Higher resolution SST
 (U10-Uocn)
 CICE also reads in U10 for ice-atm stress and gets
Uocn from HYCOM for ice-ocn stress.
1/12 Global 2003-2007 surface EKE (per unit mass)
standard stress
formulation
stress formulation
including wind-current shear
drifter observations
1/25 Global HYCOM+CICE Agulhas Rings
SSH (cm) from in-line stress
SSH (cm) from off-line stress
(U10-Uocn)
U10
Original Tides in Global HYCOM
• Tripole grid from 78.6S to 90N, at 1/12 and 1/25
• Tidal body forcing with 8 constituents
- Semidiurnal M2, S2, N2 and K2
- Diurnal
O1, P1, Q1 and N1
• Scalar self-attraction and loading (SAL)
• Topographic wave drag applied only to the tides
No drag (grey)
over 75% of the
world’s oceans
Global tide modeling
• Global tide modeling a very different art than regional/coastal tide
modeling. In the latter, tidal boundary forcing is often sufficient. In
the former, tides develop from astronomical forcing.
• “Self-attraction and loading” (SAL) accounts for the deformation of
the solid earth due to tidal loading, the resultant gravitational
perturbation due to self-attraction of the deformed solid earth, and
the gravitational perturbation due to self-attraction of the ocean
tide.
• SAL is traditionally calculated using spherical harmonics, which are
computationally expensive.
• Using global data-assimilative barotropic tide models, Egbert and Ray
(2000) demonstrated that significant tidal dissipation takes place
over rough topography in deep water.
• Inclusion of wave drag (due to internal wave formation and breaking
over rough topography) improves the accuracy of forward tide
models because tidal amplitudes are dependent on drag strength
(Arbic et al. 2004).
Improvements to Tides in HYCOM: SAL
• Scalar SAL = -constant*non-steric SSH
- always 180 out of phase with the tides
• HYCOM can now read in a full 8-component SAL
- allows iterating to actual SAL
- or use observed SAL (e.g., from TPXO8 atlas)
New Wave Drag Parameterizations
M2 SSH RMS Error vs TPXO 8 Atlas
(Shriver et al, 2012)
Existing 3D HYCOM; RMSEa = 7.5 cm
Nycander, red.sc, it.SAL; 2.75; RMSEa =
3.6 cm
RMSE
[cm]
THE END.