Avichal Mehra, Ilya Rivin, Hendrik Tolman,
Todd Spindler and
Poster presented at
GODAE OceanView - GSOP - CLIVAR Workshop on
Evaluation and Intercomparisons,
Univ. of California Santa Cruz, CA,
13-17 June 2011.
In the July 2004 report of the NOAA Science Advisory Board on Ocean
Modeling, NWS/NCEP was charged to become the “computational backbone” for
operational physical ocean modeling within NOAA. In particular the response
to the report states that the charge is
To develop a national backbone capability for ocean, coastal ocean and
Great Lakes modeling as part of an integrated operational Earth System Model
… [to] serve as the foundation for operational environmental prediction for
a diverse array of customers and partners.
Within NOAA, the primary responsibility for (weather- and) basin-scale
physical modeling resides with NWS/NCEP, whereas the responsibility for
regional and coastal scales is shared by partners inside and outside NOAA
(NOS, OAR, IOOS Regional Associations, etc.), with relevant modeling efforts
to be transferred to NCEP operational super computing facilities. The
primary responsibility for the integrated Ecosystem modeling resides within
NOS, with individual responsibilities mainly residing within NOS and NMFS.
These efforts can only succeed as a part of a national effort, with a strong
partnerships with the Navy, NASA, USCG, USACE, academia and industry.
As a response to this charge and to build adequate ocean forecasting
capability at NCEP, an operational global eddy resolving system is needed to
provide initial and boundary conditions for other operational basin-wide,
regional and coupled forecast systems. Efforts are ongoing to implement
such a real-time global ocean forecast system in operations at
NCEP/NWS/NOAA. This system will be based on an eddy resolving 1/12°
global HYCOM model (Bleck, 2002; Chassignet et al., 2009) and will serve as
part of a larger national backbone capability of ocean modeling at NWS in
strong partnership with US Navy.
Real-Time Ocean Forecast System (RTOFS)
Based on the successful design of the existing operational RTOFS-Atlantic
system (Mehra and Rivin, 2010), the global ocean forecast system will run
once a day and produce a 8 day long forecast using the daily initialization
fields produced at NAVOCEANO using NCODA, a 3D multi-variate data
assimilation methodology (Cummings, 2005). The data types assimilated
include in situ profiles of temperature and salinity from a variety of
sources and remotely sensed SST, SSH and sea-ice concentrations. The
operational ocean model configuration has 32 hybrid layers and a horizontal
grid size of (4500 x 3298) . The grid has an Arctic bi-polar patch north of
47 deg N and a Mercator projection south of 47 N through 78.6 S ( Figure 1
below). The coastline is fixed at 10 m isobath with open Bering Straits. The
potential temperature is referenced to 2000 m depth (sigma-2) and the first
level is fixed at 1 m depth. The dynamic ocean model is coupled to a
thermodynamic energy loan ice model and uses the KPP mixed layer formulation
(Large et al., 1994). The forecast system is forced with 3-hourly momentum,
radiation and precipitation fluxes from the operational Global Forecast
System (GFS) fields.
Results include daily volume and 3 hourly surface fields in netCDF format
with CF conventions. Some surface fields in GRIB format are also generated
for internal use at NWS.
Figure 1: Grid for the global ocean forecast system with each cell
representing 54th row and 75th column of the grid.
Monitoring and Diagnostics
Preliminary monitoring and diagnostic evaluation of the results from this
forecast system have begun and include representation of the Gulf Stream
path with comparisons to dual US Navy analysis (see Figure 2). The Gulf
Stream path is well defined in the forecast system and agrees with the
envelope of the Navy’s frontal analysis based on IR imagery feature
extraction. Daily SSH anomaly comparisons with the independent SSH analysis
from AVISO’s SSALTO/DUACS (https://www.aviso.oceanobs.com) show some
differences in the regions of large sea surface height variability (see
Figure 3). For example, in the Indian Ocean basin, large differences are
seen in the Agulhas Current region and in Celebes Sea (west of Borneo) and
some of the coastal regions near the Arabian peninsula. To examine
sub-surface water masses, comparisons with WOCE sections
(https://www.nodc.noaa.gov/WOCE) can indicate drifts in the forecasts which
would need to be corrected. One such comparison is shown in Figure 4. The
low salinity cell centered around 30 N and between 500 and 1000 m depths is
well represented on the meridional WOCE section P-14 across the central
Figure 2: Location of the north wall of the Gulf Stream based on feature
detection from multiple sources including Global RTOFS.
Figure 3: Differences in SSH anomalies in the Indian Ocean between RTOFS and
SSALTO/DUACS SSH analysis.
Figure 4: Salinity along the meridional P-14 WOCE section (top) compared
with the salinities along the same section from RTOFS.
As a participant in GODAE OceanView program
(https://www.godae-oceanview.org), results from RTOFS will be used for
analysis of Class 4 metrics which are based on model-derived variables in
observation space. These will include ARGO profiles, remotely sensed SSH and
SST data and Sea-Ice concentration. A proposal has also been submitted to
assess the impact of SSS on model initialization in the context of short
term real time operational forecasts.
Current plans are to implement this system into operations by the end of
fiscal year 2011. In-house analysis and initialization of this system at
NCEP using a 3DVAR data assimilation scheme will be developed in time for
the next machine (hardware) upgrade expected in 2014. Long term plans also
include providing initial and boundary conditions to existing operational
regional and coupled hurricane forecast systems at NCEP. A coarser version
will also serve as the ocean component of a future climate forecast system.
Bleck, R., 2002: An oceanic general circulation model framed in hybrid
isopycnic-cartesian coordinates. Ocean Modeling, 4, 55-88.
Chassignet, E.P., H.E. Hurlburt, E.J. Metzger, O.M. Smedstad, J. Cummings,
G.R. Halliwell, R. Bleck, R. Baraille, A.J. Wallcraft, C. Lozano, H.L.
Tolman, A. Srinivasan, S. Hankin, P. Cornillon, R. Weisberg, A. Barth, R.
He, F. Werner, and J. Wilkin, 2009. U.S. GODAE: Global Ocean Prediction with
the HYbrid Coordinate Ocean Model (HYCOM). Oceanography, 22(2), 64-75.
Cummings, J.A., 2005: Operational multivariate ocean data assimilation.
Quart. J. Royal Met. Soc., Part C, 131(613), 3583-3604.
Large, W.C., J.C. McWilliams, and S.C. Doney, 1994: Oceanic vertical mixing:
a review and a model with a nonlocal boundary layer paramterization. Rev.
Geophys., 32, 363-403.
Mehra, A. and I. Rivin, 2010: A Real Time Ocean Forecast System for the
North Atlantic Ocean, Terr. Atmos. Ocean. Sci., (Accepted for publication).