Revised July 20-22, 2004
                                           TABLE OF CONTENTS

1              Introduction ............................ Error! Bookmark not defined.Error! Bookmark not defined.

2          GODAE Components and Oversight .... Error! Bookmark not defined.Error! Bookmark not
  2.1      The GODAE common concept ............. Error! Bookmark not defined.Error! Bookmark not
  2.2      The role of the IGST .............. Error! Bookmark not defined.Error! Bookmark not defined.
3              Measurement Networks ......... Error! Bookmark not defined.Error! Bookmark not defined.
    3.1        Remotely sensed data ........... Error! Bookmark not defined.Error! Bookmark not defined.
       3.1.1   Sea Surface Topography (Altimetry).....Error! Bookmark not defined.Error! Bookmark not
       3.1.2   Sea Surface Temperature ......Error! Bookmark not defined.Error! Bookmark not defined.
       3.1.3   Surface wind ..........................Error! Bookmark not defined.Error! Bookmark not defined.
       3.1.4   Sea-ice ...................................Error! Bookmark not defined.Error! Bookmark not defined.
       3.1.5   Other remote sensing inputs ..Error! Bookmark not defined.Error! Bookmark not defined.
    3.2        In Situ ..................................... Error! Bookmark not defined.Error! Bookmark not defined.
       3.2.1   Argo Network .........................Error! Bookmark not defined.Error! Bookmark not defined.
       3.2.2   The Ship-of-Opportunity Program (SOOP) Network..... Error! Bookmark not defined.Error!
               Bookmark not defined.
      3.2.3    Global Tropical Moored Buoy Network .Error! Bookmark not defined.Error! Bookmark not
      3.2.4    Global time-series observatory Network ..... Error! Bookmark not defined.Error! Bookmark
               not defined.
      3.2.5    Global Surface Drifting Buoy Network ..Error! Bookmark not defined.Error! Bookmark not
      3.2.6    Surface marine and surface flux data ...Error! Bookmark not defined.Error! Bookmark not
      3.2.7    Tide gauge Network ...............Error! Bookmark not defined.Error! Bookmark not defined.
1.        Validation in delayed mode of the analysis, forecasts, and reanalysis produced by the
GODAE Partners. ........................................ Error! Bookmark not defined.Error! Bookmark not defined.

2.           Validation and evaluation, in quasi real time, of the "analysis products" produced on daily to
weekly basis, and for evaluation of the quality of the forecasts, which will extend 1 to 2 weeks ahead.
             Error! Bookmark not defined.Error! Bookmark not defined.

3.             Assimilation in the forecasting system. The data will need to be timely in a similar sense to
satellite altimetry data that are now delivered every week. This use of sea level gauge data in assimilation
mode where it represents a small, isolated point input, needs to be tested. The data could be of major
interest for monitoring the flows through straits. ......... Error! Bookmark not defined.Error! Bookmark not
      3.2.8 Other important in situ data sources .....Error! Bookmark not defined.Error! Bookmark not
   3.3         Observing System Challenges .............. Error! Bookmark not defined.Error! Bookmark not
4              Numerical Weather PredictionError! Bookmark not defined.Error! Bookmark not defined.

Near real-time and historical time-series of surface fluxes are available for GODAE partners from .... Error!
Bookmark not defined.Error! Bookmark not defined.
  4.1        Availability of surface fluxes ... Error! Bookmark not defined.Error! Bookmark not defined.

GODAE IP July 20-22, 2004                              Page i                                  11/06/10 19:06
     4.1.1    ECMWF Forcing Fields ..........Error! Bookmark not defined.Error! Bookmark not defined.
  4.2         Surface wind stresses ............ Error! Bookmark not defined.Error! Bookmark not defined.
  4.3         Surface heat fluxes ................ Error! Bookmark not defined.Error! Bookmark not defined.
  4.4         Sea-ice ................................... Error! Bookmark not defined.Error! Bookmark not defined.
4.5.       Sea Level Pressure ................ Error! Bookmark not defined.Error! Bookmark not defined.
   4.5     Feedback from GODAE to NWP community ................. Error! Bookmark not defined.Error!
   Bookmark not defined.
   4.6     Work plan/issues .................... Error! Bookmark not defined.Error! Bookmark not defined.
5             Data assembly and data products ....... Error! Bookmark not defined.Error! Bookmark not
  5.1         Data definitions ...................... Error! Bookmark not defined.Error! Bookmark not defined.
Figure 14. Schematic of data / product Level definitions .. Error! Bookmark not defined.Error! Bookmark
not defined.
   5.2       Data assembly centre roles and functions .. Error! Bookmark not defined.Error! Bookmark
   not defined.
   5.3       Altimetry ................................. Error! Bookmark not defined.Error! Bookmark not defined.
      5.3.1 SSALTO/DUACS (CLS/CNES) (altimetry/geoid) ........... Error! Bookmark not defined.Error!
             Bookmark not defined.
Jason-1, ENVISAT, GEOSAT Follow On, TOPEX/POSEIDON, ERS-2 ...................... Error! Bookmark not
defined.Error! Bookmark not defined.
     5.3.2 Altimeter Data Fusion Centre (ADFC) at the Naval Oceanographic Office (NAVO) ....... Error!
            Bookmark not defined.Error! Bookmark not defined.
  5.4       Sea Surface Temperature ...... Error! Bookmark not defined.Error! Bookmark not defined.
     5.4.1 Australia .................................Error! Bookmark not defined.Error! Bookmark not defined.
Contacts:     Ian Barton ian.barton@marine.csiro.au Error! Bookmark not defined.Error! Bookmark not

Neville Smith N.Smith@bom.gov.au ........... Error! Bookmark not defined.Error! Bookmark not defined.
      5.4.2 France / EUMETSAT .............Error! Bookmark not defined.Error! Bookmark not defined.
WWW site :   www.meteorologie.eu.org/safoError! Bookmark not defined.Error! Bookmark not defined.
    5.4.3    USA ........................................Error! Bookmark not defined.Error! Bookmark not defined.
    5.4.4    Climate-scale SST analyses ..Error! Bookmark not defined.Error! Bookmark not defined.
 5.5         Winds/fluxes from satellites ... Error! Bookmark not defined.Error! Bookmark not defined.
    5.5.1    O&SI SAF...............................Error! Bookmark not defined.Error! Bookmark not defined.
Coverage: Atlantic 100W-45E; 60S-60N on 0.1degree linear grids .... Error! Bookmark not defined.Error!
Bookmark not defined.

Coverage: Atlantic 100W-45E; 60S-90N on 0.1degree linear grids .... Error! Bookmark not defined.Error!
Bookmark not defined.
    5.5.2 CERSAT .................................Error! Bookmark not defined.Error! Bookmark not defined.
Jean-François Piollé email: jean.francois.piolle@ifremer.fr ................ Error! Bookmark not defined.Error!
Bookmark not defined.

Abderrahim Bentamy email: abderrahim.bentamy@ifremer.fr ............ Error! Bookmark not defined.Error!
Bookmark not defined.

Robert Ezraty email: Robert.Ezraty@ifremer.fr .......... Error! Bookmark not defined.Error! Bookmark not
     5.5.3 NOAA/NASA real-time, near-real-time .Error! Bookmark not defined.Error! Bookmark not

GODAE IP July 20-22, 2004                            Page ii                                11/06/10 19:06
  5.6         Sea Ice ................................... Error! Bookmark not defined.Error! Bookmark not defined.
     5.6.1    OSI-SAF: http://saf.dnmi.no ...Error! Bookmark not defined.Error! Bookmark not defined.
Processed data product characteristics:...... Error! Bookmark not defined.Error! Bookmark not defined.
     5.6.2 US National Ice Centre, Polar Science Team.
            http://science.natice.noaa.gov/scienceDisplay.htm ....... Error! Bookmark not defined.Error!
            Bookmark not defined.
Processed data product characteristics:...... Error! Bookmark not defined.Error! Bookmark not defined.
     5.6.3 IFREMER / CERSAT: http://www.ifremer.fr/cersat/ ....... Error! Bookmark not defined.Error!
            Bookmark not defined.
Processed data product characteristics:...... Error! Bookmark not defined.Error! Bookmark not defined.
     5.6.4 NSIDC(National Snow and Ice Data Center). http://www.nsidc.org ....... Error! Bookmark not
            defined.Error! Bookmark not defined.
Processed data product characteristics:...... Error! Bookmark not defined.Error! Bookmark not defined.
  5.7       In situ data.............................. Error! Bookmark not defined.Error! Bookmark not defined.
     5.7.1 Argo data assembly ...............Error! Bookmark not defined.Error! Bookmark not defined.
Figure 15. A schematic of data flow in the Argo Pilot Project.............. Error! Bookmark not defined.Error!
Bookmark not defined.
     5.7.2 CORIOLIS ..............................Error! Bookmark not defined.Error! Bookmark not defined.
Contacts: Sylvie Pouliquen Sylvie.Pouliquen@ifremer.fr . Error! Bookmark not defined.Error! Bookmark
not defined.

Loic Petit de la Villéon Loic.Petit.De.La.Villéon@ifremer.fr ................. Error! Bookmark not defined.Error!
Bookmark not defined.
     5.7.3 GTSPP ...................................Error! Bookmark not defined.Error! Bookmark not defined.
Email:   services@nodc.noaa.gov....... Error! Bookmark not defined.Error! Bookmark not defined.
    GOSUD ...............................................Error! Bookmark not defined.Error! Bookmark not defined.
Co-chairs: delcroix@notos.cst.cnes.fr or keeley@meds-sdmm.dfo-mpo.gc.ca........... Error! Bookmark not
defined.Error! Bookmark not defined.

Email:        Loic.Petit.De.La.villeon@ifremer.fr ....... Error! Bookmark not defined.Error! Bookmark not
     5.7.4    Global Tropical Moored Buoy Network .Error! Bookmark not defined.Error! Bookmark not
Email: atlasrt@pmel.noaa.gov ..................... Error! Bookmark not defined.Error! Bookmark not defined.

Mike McPhaden email: mcmphaden@pmel.noaa.gov. Error! Bookmark not defined.Error! Bookmark
not defined.
     5.7.5 Drifter and Ocean Currents ....Error! Bookmark not defined.Error! Bookmark not defined.
     5.7.6 Sea level from tide gauges ....Error! Bookmark not defined.Error! Bookmark not defined.
  5.8        Ancillary datasets ................... Error! Bookmark not defined.Error! Bookmark not defined.
     5.8.1 Climatologies ..........................Error! Bookmark not defined.Error! Bookmark not defined.
     Bathymetry ..........................................Error! Bookmark not defined.Error! Bookmark not defined.
1.            Land points have bottom depths > 0 ..... Error! Bookmark not defined.Error! Bookmark not

2.            Sea points have bottom depths = 0 ...... Error! Bookmark not defined.Error! Bookmark not
   5.9        Issues and challenges ........... Error! Bookmark not defined.Error! Bookmark not defined.

GODAE IP July 20-22, 2004                            Page iii                               11/06/10 19:06
      5.9.1    An Ocean Information Technology Project . Error! Bookmark not defined.Error! Bookmark
               not defined.
      5.9.2    Quality Control .......................Error! Bookmark not defined.Error! Bookmark not defined.
6              Data servers ........................... Error! Bookmark not defined.Error! Bookmark not defined.
    6.1        Global telecommunication system ........ Error! Bookmark not defined.Error! Bookmark not
    6.2      Specialist ocean data server (R/T)........ Error! Bookmark not defined.Error! Bookmark not
       6.2.1 GODAE Real-time Data Server in Monterey ................. Error! Bookmark not defined.Error!
             Bookmark not defined.
       6.2.2 CORIOLIS ..............................Error! Bookmark not defined.Error! Bookmark not defined.
    6.3      IPRC ....................................... Error! Bookmark not defined.Error! Bookmark not defined.
    6.4      Others .................................... Error! Bookmark not defined.Error! Bookmark not defined.
    6.5      Issues and Challenges ........... Error! Bookmark not defined.Error! Bookmark not defined.
7              Modelling/assimilation centres or groups .................................................................................6
    7.1        Australia ....................................................................................................................................6
       7.1.1   Background .............................................................................................................................. 6
       7.1.2   The BLUElink Forecasting System .......................................................................................... 6
       7.1.3   Seasonal-to-interannual prediction .......................................................................................... 9
       7.1.4   Short-range ocean forecasts .................................................................................................... 9
       7.1.5   Coastal forecasting .................................................................................................................. 9
       7.1.6   Computing resources ............................................................................................................. 10
    7.2        Europe - MERSEA ................................................................................................................. 11
    7.3        France .................................................................................................................................... 11
       7.3.1   Background ............................................................................................................................ 11
       7.3.2   MERCATOR Ocean Forecasting System .............................................................................. 11
    7.4        Canada ................................................................................................................................... 14
    7.5        Japan ..................................................................................................................................... 15
       7.5.1   Background ............................................................................................................................ 15
       7.5.2   Japan Meteorological Agency (JMA)/Office of Marine Prediction (OMP) and Meteorological
               Research Institute (MRI): COMPASS-K ................................................................................ 16
      7.5.3    JMA/Climate Prediction Division (CPD): ODAS ..................................................................... 17
      7.5.4    Kyoto University and Japan Marine Science Foundation (MSF) ........................................... 18
      7.5.5    Frontier Group (1): Frontier Research System for Global Change (FRSGC)/Integrated
               Modelling Research Program (IMRP) and Kyoto University .................................................. 18
      7.5.6    Frontier Group (2): Frontier Research System for Global Change (FRSGC)/ Climate
               Variations Research Program (CVRP): J-COPE ................................................................... 19
      7.5.7    Kyushu University/Research Institute for Applied Mechanics (RIAM)/Dynamics Simulations
               Research Centre (DSRC) ...................................................................................................... 20
       7.5.8   Data and Product serving ...................................................................................................... 20
    7.6        Norway/Europe ...................................................................................................................... 21
       7.6.1   Background ............................................................................................................................ 21
       7.6.2   The DIADEM/TOPAZ Ocean Forecasting System ................................................................ 21
    7.7        UK .......................................................................................................................................... 24
       7.7.1   Background ............................................................................................................................ 24
       7.7.2   Met Office Forecasting Ocean Assimilation Model (FOAM) System ..................................... 24
       7.7.3   OCCAM Ocean Forecasting System ..................................................................................... 26
    7.8        USA ........................................................................................................................................ 26
       7.8.1   Background ............................................................................................................................ 26
       7.8.2   Product Dissemination ........................................................................................................... 27
       7.8.3   NCEP/ Global Climate and Weather Modeling Branch (GCWMB) ........................................ 27
       7.8.4   U.S. Navy Operational System at FNMOC ............................................................................ 28
       7.8.5   U.S. Navy Operational Systems at the Naval Oceanographic Office .................................... 29
       7.8.6   U.S. Navy Research Systems at NRL/Stennis ...................................................................... 31

GODAE IP July 20-22, 2004                                             Page iv                                                   11/06/10 19:06
        7.8.7      ECCO Consortium ................................................................................................................. 34
        7.8.8      HYCOM Consortium .............................................................................................................. 36
        7.8.9      GMAO .................................................................................................................................... 40
        7.8.10     University of Maryland SODA ................................................................................................ 41
     7.9           China ...................................................................................................................................... 43
     7.10          Issues ..................................................................................................................................... 44
8            Standard products and product serving ..... Error! Bookmark not defined.Error! Bookmark
not defined.
  8.1        Product coherency and standardisation Error! Bookmark not defined.Error! Bookmark not
  8.2        From production to distribution ............. Error! Bookmark not defined.Error! Bookmark not
  8.3        Product selection, manipulation and visualisation ......... Error! Bookmark not defined.Error!
  Bookmark not defined.
9                  Application Areas ................................................................................................................... 50
     9.1           Ocean mesoscale and surface fields and short-range forecasts ........................................... 50
     9.2           Seasonal-to-interannual climate ............................................................................................ 52
     9.3           Medium to long-term climate.................................................................................................. 54
10                 Users and Benefits ................................................................................................................. 56
     10.1          Broad approach...................................................................................................................... 56
     10.2          Product Lines ......................................................................................................................... 58
     10.3          Reaching the user community ............................................................................................... 58
       10.3.1      GODAE Symposium .............................................................................................................. 59
       10.3.2      User forums............................................................................................................................ 60
       10.3.3      Regional alliances .................................................................................................................. 60
       10.3.4      Indian Ocean GOOS .............................................................................................................. 60
     10.4          Mechanisms for Feedback/Evaluation ................................................................................... 61
       10.4.1      Possible external measures ................................................................................................... 61
11                 GODAE Work Plan ................................................................................................................ 62
     11.1          Pilot Projects .......................................................................................................................... 62
       11.1.1      Argo ........................................................................................................................................ 62
       11.1.2      SST ........................................................................................................................................ 62
     11.2          On-going Actions and Work ................................................................................................... 66
       11.2.1      Data set characterization for assimilation .............................................................................. 66
       11.2.2      Forcing fields .......................................................................................................................... 67
       11.2.3      Quality control of observations ............................................................................................... 67
       11.2.4      Ocean Modelling .................................................................................................................... 68
       11.2.5      Data assimilation .................................................................................................................... 68
       11.2.6      Product Characterization and Internal Metrics ....................................................................... 69
     11.3          Running regional and global prototype systems and transition to operational systems ........ 70
     11.4          Intercomparison activities ...................................................................................................... 71
       11.4.1      North Atlantic.......................................................................................................................... 71
       11.4.2      North Pacific ........................................................................................................................... 73
       11.4.3      Equatorial Pacific ................................................................................................................... 73
       11.4.4      Global ..................................................................................................................................... 74
       11.4.5      Reanalysis activities ............................................................................................................... 74
12                 External Relationships ........................................................................................................... 75
     12.1          Liaisons with International Organizations .............................................................................. 75
     12.2          Relation to other programs .................................................................................................... 75
13                 References ............................................................................................................................. 76

Annex 1: URL List ........................................................................................................................................ 78

GODAE IP July 20-22, 2004                                                  Page v                                                   11/06/10 19:06
GODAE IP July 20-22, 2004   Page vi   11/06/10 19:06
7 Modelling/assimilation
  centres or groups

The GODAE modelling/assimilation centres or
groups as they are envisioned now are described.
Characteristics of the systems planned for the
GODAE operational phases (2003-2007) (input
data, model, assimilation method, anticipated
products, target applications…) are summarized.

7.1       Australia

7.1.1      Background
The Australian Bureau of Meteorology (BoM),
CSIRO (Marine and Atmospheric Research,
CMAR) and the Royal Australian Navy (RAN) are collaborating in the development of an ocean
forecasting system to support a GODAE Project within Australia. The project was launched in
September 2002 under the name of BLUElink – Ocean Forecasting Australia and is scheduled to
deliver a global operational system with a focus on the Australasian region by 2007. Both the BoM and
CSIRO provide the scientific expertise and technological infrastructure whereas the RAN is a major
funding source and beneficiary of the forecasting system.

BMRC and CMR developed the first operational coupled ocean-atmosphere model for ENSO
prediction (see http://www.bom.gov.au/bmrc/ocean/JAFOOS/POAMA). The ocean component is an
Australian version of the MOM code (ACOM, the Australian Community Ocean Model) with enhanced
resolution in the tropics but coarse resolution elsewhere. The atmospheric component of the model is
a T47L34 version of the BMRC Atmospheric model and the OASIS coupler is used to link the models.
Development of a short-range ocean forecasting system will also benefit further improvements of the
seasonal-to-interannual prediction system.
Contacts:      Neville Smith N.Smith@bom.gov.au
               Andreas Schiller A.Schiller@csiro.au

7.1.2      The BLUElink Forecasting System

Input Data
The project will have a heavy reliance on data and data products of other Partners in GODAE (e.g., via
the Monterey data server).
          The Joint Facility (JAFOOS) was established in 1999 to support the collaborative CMR/Bureau
           observational program (see http://www.bom.gov.au/bmrc/ocean/JAFOOS/). JAFOOS is
           coordinating Australian contributions to GOOS and has a significant effort in assembly and
           quality control of ocean data. The JAFOOS is likely to be the focus of attempts to enhance
           Australia‘s contribution. Contact:   Ann Thresher Ann.Thresher@csiro.au
          The forcing fields of the operational systems will come from the Bureau of Meteorology NWP
           systems For development and testing, both the NCEP and ECMWF re-analysis fields will be
          CMR presently has a near-real-time altimeter product. This will be used initially for the
           Australian region.
          BMRC and CMR are participating in the GHRSST Project. The goal is the development of
           unique high-resolution SST data sets and products for the Australian region.
          The data assimilation system (see below) will use both SST and SSH data from available
           satellite systems and from in situ data streams (mainly Argo floats).

GODAE IP July 20-22, 2004                        Page 7                               11/06/10 19:06
Data serving
The responsibility for operational data and product management as part of this project resides with
CBoM. It comprises the following components:

      A data base and server for incoming data (local, regional, global telecommunications), for
       provision of data to models, and for provision of data to the Integrated Product Service and
       scientists involved in the development of components:
           o Argo profiling floats;
           o Ship of Opportunity lines,
           o Surface drifters;
           o Fixed-point time series;
           o Tide gauge data from the National Tidal Centre and other Australian sites; etc.
           o The key satellite data sets include:
                     Altimetry: Topex/Poseidon, Jason 1, GFO, ERS-2, Envisat;
                     SST: NOAA AVHRR, MTSAT/GOES, Envisat, MODIS, AMSR-E;
                     Surface winds: QuikScat and other available remote measurements such as
                        DMSP, and/or NWP;
                     Sea-ice extent (an Antarctic Climate and Ecosystems CRC project).

      Implementation of an archive and retrieval product database and server for Bluelink fields and
       delivery of relevant products via the Australian Integrated Forecast System (AIFS).

      Implementation of a digital and graphical service for Bluelink products via the Internet and a
       DODS Server (Http://DODS.bom.gov.au).

   Contact: Graham Warren g.warren@bom.gov.au

   The forecasting component of the Bluelink project will be based on a nested modeling system.

      The global ocean data assimilation and prediction model (OFAM) is based on the latest GFDL
       MOM4 code. It has a horizontal resolution of (1/10)º within a ―rectangular‖ box between 90ºE
       and 180ºE and 17Nº and 75ºS. Within the remaining parts of the Equatorial and Southern
       Pacific and Indian Oceans the minimum resolution is 0.9º. Outside this area the resolution is
       as coarse as 2º. The model has 47 vertical levels with 35 levels in the upper 1000m. This
       model will be implemented and run operationally by the CBoM. Contact for global model:
       Andreas Schiller Andreas.Schiller@csiro.au
      A regional model, nested within the eddy-resolving region of the global model is being
       designed as a relocatable model which will require only minimum user input. The regional
       ocean model MECO is based on an in-house development by CMR (Model of the Estuaries
       and Coastal Oceans). It is one-way coupled to a nested relocatable atmospheric model
       (Colorado State University RAMS model); RAMS is nested with CBoM‘s GASP/LAPS
       atmospheric models. This model will be implemented at the operational centre of the RAN.
       Further details are:
              Automatic implementation via visual interface of hydrodynamic model with minimum
               user input
              Forecasts of ocean and atmosphere state out to 3 days
              Ocean domains of scales down to 100 km x 100 km with resolution down to 2km
              Surface atmospheric data for ocean from RAMS (currently just one-way coupling)
              Contact for regional model: Peter Craig Peter.Craig@csiro.au

Assimilation method

GODAE IP July 20-22, 2004                     Page 8                                 11/06/10 19:06
       The stand-alone operational analysis system run by the Bureau will be continued with
        enhanced global and regional versions, with fine resolution in the Australian region, and
        capable of ingesting both in situ temperature and salinity profiles and altimeter data (via
       A daily-updated nowcast of the three-dimensional temperature, salinity and current velocity
        field for the Australasian region based on in situ and satellite data has been developed and is
        run by CSIRO. Contact: David Griffin David.Griffin@csiro.au
       The forecasting system is based on a multivariate assimilation scheme that combines a model
        forecast with available in situ and satellite derived observations into the global ocean model to
        provide improved short-range model predictions with forecast skill of up to 4-6 days. The
        system will be run initially with a statistical interpolation data assimilation system, assimilating
        altimeter data, subsurface Argo, SOOP and XBT data (assembled into a single data set) and
        SST data.
       An enhanced data assimilation method will be developed in collaboration with the Bluelink
        project. It will be based on the ensemble Kalman filter but is unlikely to be implemented in the
        operational ocean forecasting system before 2008. Additional work will be done within the
        Antarctic and Climate Ecosystems CRC for assimilation of Antarctic data.
        Contacts: Oscar Alves O.Alves@BoM.gov.au, Peter Oke Peter.Oke@csiro.au

Prototype systems and transition to global systems
Pre-operational trials of the complete short-range forecasting systems are scheduled for 2005 (global
model with data assimilation, nested models, access to data input streams). It is planned to have the
system fully operational by the end of 2006/early 2007.
Contacts: Gary Brassington (G.Brassington@bom.gov.au), Andreas Schiller A.Schiller@csiro.au

Assimilation products and dissemination
As part of this joint Ocean Forecasting System project, a substantial effort is being devoted to the data
and product server. The Bureau is implementing the ECMWF Meteorological Archival and Retrieval
System (MARS) and this should greatly enhance the functionality for ocean applications. In
collaboration with ECMWF, MARS is being adapted to handle data types and products common to
ocean data assimilation and prediction and interfaces will be established to allow both intranet and
internet access to Bureau operational products, including both regional and global systems.
    Contacts: Graham Warren G.Warren@bom.gov.au, Neville Smith N.Smith@bom.gov.au
The principal source for products from the operational trials and fully operational systems will be the
external access point to the MARS server. The Partnership with the Navy potentially allows for
expanded requirements in terms of product serving, proper treatment of oceanographic fields,
ingestion and serving of data products, and modes for internal and external access. The OPenDAP
server DODS.bom.gov.au will be used for intercomparisons and product exchange.

To facilitate access to products via the intranet (e.g., for the RAN) and internet, a web-based interface
to an internal MARS client will be constructed. This interface will operate from an external proxy server
for security reasons. We anticipate this functionality will satisfy our requirement for ingesting large data
and product sets from outside and providing our own data and products for the GODAE Partners.

It is also anticipated that several products, particularly those from the experimental system, will be
available directly from the Australian Partnership, either through CSIRO or BMRC.

At this time not all details have been decided. A potential list is:

GODAE IP July 20-22, 2004                          Page 9                                  11/06/10 19:06
        Daily global and regional surface wind products from the NWP systems
        SST data products based on locally retrieved SST
        Regional and global SST and subsurface analyses (no model assimilation)
        Global model assimilation estimates (u, v, η, T, S), probably on reduced space-time grids
        Global and regional model forecasts, again on reduced space-time grids
        Various ocean state diagnostics and statistics as determined by GODAE metrics
        Various coupled model climate forecasts fields

7.1.3   Seasonal-to-interannual prediction
The BMRC coupled seasonal prediction model has been running operationally since 2002. The
application of climate products occurs through the Bureau's National Climate Centre and its Climate
Analysis Section. The monthly Seasonal Climate Outlook meetings provide and interpretation of
Bureau and other climate monitoring and prediction information. An internal forum involving
operational and research personnel provides advice on the performance of systems and on the
implementation of new systems.

The Climate Analysis Section, and contacts in the regions, provides a conduit from climate product to
application (consumption). There are direct links into many sectors but agriculture is the dominant
user. Other groups (e.g., the Queensland Department of Primary Industry) also provide value adding
of Bureau climate information.
Contacts: Oscar Alves O.Alves@BoM.gov.au

7.1.4   Short-range ocean forecasts
The Navy will be the principal customer for the new Ocean Forecasting System. The project is linked
to an effort to develop improved acoustical model systems for the Navy. Links to regional users are
being established through WAGOOS, a consortium of industry, private sector and Government users
in Western Australia, regional research initiatives (e.f., SRFME), and the regional Offices of the Bureau
of Meteorology.

7.1.5   Coastal forecasting
The regional forecasting model will be operated at a resolution that permits useful ocean state
estimates and forecasts for the coastal region. Discussions have been initiated with private and public
sector groups in Western Australia with the aim of developing several value-adding projects that use
the products of the ocean forecasting and data assimilation system.

Links with GODAE pilot projects (Argo, GHRSST)
Data from Argo floats are central to the data assimilation system. Australia (CBoM, CSIRO, ACE CRC)
is deploying an increasing number of floats in its regional oceans which will benefit the forecasting
system. Research on how to optimize information derived from Argo floats in the data assimilation
system is a core activity.

Both CBoM and CSIRO actively participate in the GHRSST pilot project. Apart from calibration and
validation activities the goal is the provision of real-time regional SST products with order 10 km spatial
resolution within 6 hours of data reception. The products will be based on both satellite and in situ
measurements. Satellite data will come from both infrared and microwave instruments and the
processing system will blend all data sources in to skin, sub-skin and bulk (depths of 1 to 5 m)
temperature estimates.

Contacts: Helen Beggs (H.Beggs@bom.gov.au), Ian Barton: Ian.Barton@csiro.au

Internal metrics and intercomparison plans

GODAE IP July 20-22, 2004                        Page 10                                  11/06/10 19:06
The forecasting system will be subject to ongoing validation with independent (non-assimilated)
observations and with products provided by other agencies (e.g. NAVOCEANO).

The Bluelink research partners have agreed to lead an intercomparison effort for the Australian region.
Because of the schedule for Bluelink, Australia will not be ready to provide products for at least 12
months, probably early 2005. A set of metrics for the Australian region, including the Antarctic sector,
will be developed. The standards and formats for presentation of data on servers will follow that
developed by the MERSEA project. The Bluelink partners are currently working on a schedule of
activities noting dates/times of (i) Agreement on metrics/classes, (ii) participation, (iii) set up of servers
and formats, (iv) planned/needed workshops, and (v) any other milestones. Initial focus will be on
metrics based on MERSEA class 1 and class 2 data sets. Expected participants are: Aust, NRL,

Targeted Users and envisioned external metrics
Initially, the Navy and research scientists will be the main users of the forecasting system. There is
potentially considerable interest by the marine user community such offshore industries, fisheries,
marine transport and search and rescue. Within Australia‘s Ocean Policy, the Bureau, CSIRO and
Navy, among others are actively pursuing the concepts of marine environmental prediction and
operational ocean services, with activities and actions during 2004-2007 that will enable engagement
of a broader user base. It is anticipated that once the system is fully operational many national and
regional users will incorporate data from the Bluelink system in their services.

Nowcasts of the three-dimensional temperature, salinity and current velocity field based on in situ and
satellite data are being used by the Australian Marine Safety Authority.

Reanalysis activities

A major reanalysis effort is scheduled to start in late 2004. The focus will be on the oceans around
Australia. Input data sets for the assimilation will be based on satellite SSH and SST, Argo, SOOP,
XBT and other sources. The integration period will cover the early 1990‘s until present. At a later stage
a longer reanalysis period might be considered. This effort will be based on the global prediction model
and the model results will be validated as part of the intercomparsion effort. It is anticipated that the
complete data set will be available by mid-2005.

Outside the BLUElink project, but within the Antarctic, Climate and Ecosystems CRC, a high-resolution
model with order 1/8 degree resolution is being run, specifically for research within the CRC.
Simulations from this model will be used from time for comparison and evaluation (it is not intended
that this model be made operational, but it will be used to guide evolution of the BLUElink system;
contact is Nathan Bindoff [n.bindoff@utas.edu.au].

7.1.6   Computing resources
The Bureau and CSIRO jointly operate the High Performance Computing and Communication Centre.
At present, the system is being upgraded to a NEX SX-6 machine with 18 nodes, each of 8 CPUs and
64GBytes of memory. In the timeframe of GODAE, an upgrade to 28 nodes is scheduled. The
operational systems have a limited window within which they must operate and ocean systems must
compete for time with NWP, climate and wave model systems.

It is anticipated the global operational system will require order 40 min of dedicated super-computer
time each day (with an option for spreading it over less CPUs and longer elapsed times).

The MARS system is supported by an IBM with good mass-store facilities.

GODAE IP July 20-22, 2004                         Page 11                                   11/06/10 19:06
7.2       Europe - MERSEA

7.3       France

7.3.1      Background
The MERCATOR project was launched in 1995 by the six major French agencies involved in
oceanography (CNES, CNRS, IFREMER, IRD, Météo-France and SHOM, with involvement of their
subsidiaries CLS and CERFACS) to develop a joint operational capacity for global high resolution
ocean monitoring and forecasting.

The project is lead by the MERCATOR OCEAN public company, created in 2002 by these six Patrons
agencies for that purpose, with commitments to prepare an operational centre by 2006.

MERCATOR Assimilation Centre is based in Toulouse (France).

Objectives are to:

          simulate the global ocean with a primitive-equation high resolution model, assimilating satellite
           and in situ data, to provide hindcasts and near-real time nowcasts and forecasts of the global
           ocean circulation,

          be operated on an operational mode (ie routine and near-real-time) to answer (i) research, (ii)
           national (military and civilian) state application, and (iii) commercial oceanography end-user

          then contribute to the development of a seasonal and climate forecasting system by providing
           ocean initial conditions for ocean/atmosphere coupled models.

MERCATOR is the French contribution to GODAE, and partner of the European GMES MERSEA
consortium, where in charge of developing the European High Resolution Global Ocean system.

Scheduling is based on GODAE, with an intensive experiment period in 2003-2005. Years 2000-2002
were dedicated to the system development, including pre-operational near-real-time experiments
(2001-2002) on the basis of system prototypes.

A series of ocean forecasting prototypes of growing complexity are progressively implemented by
Mercator through years 2001-2005. Three systems are currently running (status as June 2004),
delivering weekly ocean nowcasts on North and Tropical Atlantic, Med Sea, and Global Ocean.

Real-time outputs, as well as validation reports, are available at http://www.mercator.eu.org.

7.3.2      MERCATOR Ocean Forecasting System

MERCATOR uses the OPA z-coordinate primitive equation ocean code (Madec & al).

Two configurations are Mercator target configurations for the GODAE intensive phase, so that to cover
global ocean with an eddy-permitting resolution, and North Atlantic and Mediterranean Sea basins with
an eddy-resolving resolution :

-     a basin high resolution configuration (5 to 7 km horizontal resolution, 43 vertical levels) covering
      North Atlantic and Mediterranean sea; this configuration focuses on mesoscale processes and
      links with coastal modelling in European seas;

GODAE IP July 20-22, 2004                          Page 12                                 11/06/10 19:06
-   a global middle resolution configuration (1/4° horizontal resolution, 46 vertical levels) covering
    global ocean; this configuration aims at providing the best ocean state estimates for global ocean
    analysis and boundary conditions for regional models worldwide.
Two lighter configurations are also implemented in real-time by MERCATOR for demonstration and
testing of new algorithms:

1. a basin middle resolution configuration (1/3° horizontal resolution, 43 vertical levels) covering
   North and Tropical Atlantic;

2. a global low resolution configuration (2° horizontal resolution, 30 vertical levels) covering global
In June 2004,
       the Atlantic 1/3° model is validated and has been routinely run with assimilation in the
        operational chain since January 2001 ; multiyear (10-15) model simulations are available ; a
        reanalysis undergone ;
       the Atlantic/Mediterranean 1/15° model is validated and has been routinely run with
        assimilation since January 2003 ; multiyear model simulations are available ;
       the Global 2° model configuration is validated and has been routinely run with assimilation in
        the operational chain since July 2003 ; 10 year reanalysis with assimilation is available ;
       the Global ¼° model configuration is validated and integration with assimiliation is undergone ;
        some validation model simulations available.

Assimilation method

MERCATOR assimilation deals with altimeter sea level anomaly, sea surface temperature and
temperature and salinity in situ profiles data into its basin and global scale models. Assimilation is
considered for routine near-real-time nowcasts and forecasts issues, but also long term reanalysis

MERCATOR is developing a suite of assimilation tools (called "SAM" for Mercator Assimilation
System) of increasing complexity, from sequential to variational method:

       the first release, SAM-1, is based on the multivariate optimal interpolation SOFA scheme
        developed by LEGOS (De Mey, 1999);
       the second release, SAM-2, will add an Singular Extended Evolutive Kalman (SEEK) filtering
        analysis method, developed in LEGI (Brasseur, 1999)
       the third one, SAM-3, will consider advanced methods such as variational method.
The SAM suite uses the PALM software (algebric coupling concept) developed by CERFACS
(Piacentini, 1998) to couple in a generic way the different assimilation algorithms.

In June 2004,

       The SAM-1 version is fully operational: after 3 years of continous operation in its univariate
        ―altimetry only‖ version, it was upgraded in its multivariate version since January 2004,
        enabling joint assimilation of along-track SLA, SST and T&S vertical profiles. SAM1 is a
        reduced ordre optimal interpolation scheme, using a set of seasonal EOFs computed at each
        grid point, (Ψ, T, S) as state vector.
       The SAM-2 (SEEK filter) version is under development ; version 1 being available for tests ;
       The SAM-3 (variational) is under specification, served by research investigations conducted
        these last 3 years inside the Mercator Science Working Team.

GODAE IP July 20-22, 2004                      Page 13                                 11/06/10 19:06
Input Data

MERCATOR relies on existing data assembly centres to collect, process and validate its input real
time and delayed mode data. Input data for MERCATOR include:

      Altimetry (from SSALTO/DUACS center) : Along-track and intercalibrated sea level anomalies
       from Jason-1, Envisat and GFO, as well as Topex/Poseidon missions ; weekly retrieval ;
       Mean-Sea-Surface-Height combining gravity/insitu data (M.H. Rio & al).
      In-situ data (from CORIOLIS center) : ARGO profiling floats, XBT,TAO/PIRATA/TRITON,
       surface drifters... ; ; weekly retrieval ;
      Sea Surface Temperature: Global Reynolds SST product for operational assimilation ;
       Eumetsat / Météo-France SAF Ocean&Ice Atlantic high resolution SST product (10 km, daily)
       for routine validation ; Strong interest for a high resolution GODAE SST product
      Forcing data: 6-hour analyses and predictions from ECMWF for operational forcings ; Cersat
       real-time scatterometry winds for validation.
Data Serving
Altimetry (SSALTO/DUACS), In Situ (CORIOLIS), NWP Forcings (ECMWF), SST (SAF O&I)

Product Serving

URL: http://www.mercator.eu.org ; mailto://products@mercator-ocean.fr

The MERCATOR system provides a full 3D depiction of the ocean dynamics and thermohaline
circulation (T, S, currents, mixed layer depth, …), with a priority given to high resolution (eddy
resolving) scales.

Information is available on a near-real-time and routine basis, by providing weekly Near-Real-Time
Analysis and 2-week Forecasts ; and on a Reanalysis mode, with data assimilation

Each weekly ocean bulletins provides:

      near-real-time ocean nowcasts and forecasts issued from routine assimilation and modelling
      input data assessments and analysis (comparison & combination of data sets before
      a set of predefined maps and 3D files giving a complete depiction of the ocean (more than 800
       maps each week today)
       technical information on the assimilation run
A quarterly newsletter is edited and gives up-to-date Mercator system validation results. So far, 13
newsletters have been published : they give a good overview of the Mercator System Products
validation activity.

Dissemination of MERCATOR Products is made through www and FTP tools, both for real-time and
archived products.

The Live Access Server (LAS) tool is implemented and tested to be the standard interface for the
GODAE community, and the MERSEA group in Europe.

MERCATOR outputs are freely available for Research and Educational applications.

Internal Metrics and intercomparison plans

MERCATOR has been routinely provided on the www its ―technical bulletin‖ coupled to each weekly
ocean bulletin since January 2001, providing internal diagnostics of assimilation.

GODAE IP July 20-22, 2004                      Page 14                              11/06/10 19:06
The system was upgraded in the framework of the MERSEA Strand 1 GMES/EC project by the
implementation of the standard metrics adopted by all MERSEA centres on North Atlantic and Med
Sea, as well as GODAE.

Targeted Users and envisioned external metrics

Targeted users of the MERCATOR system are the six Mercator Patrons Agencies and their application
sectors, the National and European Policy Makers, GODAE partners, Research, and Commercial

Application Centres identified among the Mercator partners, are e.g. SHOM for Navy applications,
Météo-France for marine safety and oil spill monitoring, and seasonal forecasting (collab. with
ECMWF), or Ifremer for coastal and ecosystem monitoring.

If research is one of the leading application in the scope of the Mercator Science Working Team, more
than 50% of the users are today outside the research field. In the commercial sector, fisheries and off-
shore applications are considered.

Mercator serves today more than 100 users.

Reanalysis activities

Providing reanalysis with multidata assimilation (i.e. on the period 1993-present time for the availability
of accurate altimetry) is one of the objectives of MERCATOR.

First 11 year reanalysis is undergone on 1993-2004 years (MERA-11), on the basis of the 1/3° North
and Tropical Atlantic configuration and a reanalysis version of the SAM1 assimilation tool to assimilate
SLA, SST and T&S profiles.

Computing resources

The MERCATOR system is developed on a Linux network of workstations with a Fujitsu Linux Cluster
for local computing facilities. For supercomputing needs, the project team works on the Meteo-France
(FujiTsu VPP 5000) and ECMWF (IBM P690) centers facilities, and the CNRS IDRIS center (NEC) is
used by the Mercator Science Working Team for its research activity. Mercator uses Archiving Data
Center of the French Space Agency for its mass storage needs.

7.4   Canada
Canada is considering developing an operational global ocean forecasting system and active
discussions involving Environment Canada, Department of Fisheries and Oceans, Department of
National Defence, and Canadian universities are presently underway. Topics of discussion include the
choice of model, coupling of the ocean model to a global atmosphere and ice model and, most
importantly, the major allocation of human and computer resources. Decisions are expected to be
made in 2004.

In anticipation of developing a deep ocean operational capability, Canada has already funded some
research and development projects that involve the assimilation of Argo and satellite data into basin
scales models. One project is focused on the North Atlantic which we will denote by NA1 for the
purposes of this report. A parallel project that uses the same modeling and assimilation strategy has
been initiated for the North Pacific (NP1). These two projects focus on medium range ocean
forecasting. The third project is also for the North Pacific but focuses on longer time scales (seasonal
forecasting) and uses a different model and assimilation strategy (NP2). Details of the three projects
are given below. Note that these projects are research and development and so some of the
information requested cannot be given.

GODAE IP July 20-22, 2004                        Page 15                                  11/06/10 19:06
Input Data

         NA1: Temperature and salinity climatology from Ygor Yashaev, sea surface height anomalies,
          NCEP surface forcing, and Argo profiles.
          NP1: Levitus climatology, COADS surface fluxes, Argo data.
          NP2: TOPEX/Poseidon/Jason altimetry, Argo data.

          NA1: Parallel Ocean Program (POP)
          NP1: Parallel Ocean Program (POP)
          NP2: PEZ/IPEZ (i.e. Andrew Bennett's version of MOM and its inverse)

Assimilation method

         NA1: Frequency dependent nudging to ensure the model‘s temperature and salinity
          climatology is close to the observed climatology (while allowing eddies to evolve freely),
          Cooper-Haines for sea surface height anomalies, and multivariate optimal interpolation
          (planned) for Argo profiles. (Experiments underway with the ensemble Kalman Filter as a way
          of assimilating Argo trajectories).
          NP2: Essentially the same as NA1.
          NP2: The representer approach developed by Andrew Bennett.
Assimilation products and dissemination

         NA1: Hindcasts, and pre-operational nowcasts and forecasts of the North Atlantic with lead
          times up to 20 days.
          NP1: Nowcasts/forcasts for the Northeast Pacific but not in operational mode.
         NP2: Reconstruction of North Pacific circulation from 1992 to present and initial conditions to
          launch seasonal forecasts.
Links with GODAE pilot projects (Argo, GHRSST)

All three projects plan to assimilate Argo profile data. As part of NA1, an attempt will be made to
assimilate Argo trajectory data using the EnKF.

Internal metrics and intercomparison plans

Where possible we will use the metrics defined by MERSEA.

7.5       Japan

7.5.1      Background
Three working groups (WGs) were established and have discussed Japan-GODAE activity. The first
WG is under the Science Council of Japan. The WG discusses scientific directions and
recommendations to agencies towards operational developments. The second WG is under the Japan
CEOS/IGOS-Ocean Committee in the Japan Aerospace Exploration Agency (JAXA). The WG
discusses the activities with linkages between members, data flow/product and assimilation
experiments. The third WG is for J-GODAE High Resolution SST project (ADEOS-II New Generation
SSTs) in the Earth Science and Technology Forum (ESTF) under Earth Science and Technology
Organization (AESTO). It acts as a forum for developing Japan-GODAE High Resolution SST. The

GODAE IP July 20-22, 2004                        Page 16                                   11/06/10 19:06
WGs consist of university members (Kyoto, Tohoku, Hokkaido, Tokai, Kyushu, and Tokyo) and
Agency members such as JAMSTEC, JAXA, RESTEC, CRIEPI, Frontier groups (FRSGC/IGCR,

The activity of the groups will build on and develop links with existing activities, such as data centres at
the JMA of real time delivery, JAMSTEC of intermediate time delivery for research works, JODC as the
National Oceanographic Data Centre, and JAXA of satellite data delivery. The groups work with
ongoing research initiatives esp. Millennium Project (Japan-Argo), RR2002 Project, CLIVAR/UOP, and

The WGs will establish a ―Japan-GODAE Network‖ which activates links of research and technical
information, data delivery, and distribution of products to targeted users.

The main Japanese contribution, as assimilation centres, to GODAE will be the Japan Meteorological
Agency, Frontier groups, and University groups. Those groups have operational or research systems
in real time or delayed modes.

7.5.2   Japan Meteorological Agency (JMA)/Office of Marine Prediction (OMP) and
        Meteorological Research Institute (MRI): COMPASS-K

The ocean model used is an MRI eddy permitting model (MRI-EGCM). The model is a rigid-lid version
in the North Pacific from 12N to 55N. The grid has a variable mesh size: 1/4x1/4, and 21 vertical levels
around Japan. The model includes Takano-Oonishi scheme for treating steep bottom topography with
generalised Arakawa scheme for momentum advection terms. The model will be changed to a new
community model, MRI.COM for global and western North Pacific oceans for research mode.
Operational system (COMPASS-K) is for North PAcifc. Research system1 (MOVE: 3DVAR) is for
western North Pacific. Research system 2 (MOVE: 3DVAR) is for global ocean. Product line is from
1993 to 2007.

Assimilation Method

The assimilation system uses a multivariate, scale-dependent four-dimensional optimum interpolation
(4DOI) method with a preceding-nudging insertion (Kamachi et al., 2001). The OI employs
inhomogeneous, anisotropic space-time combined background error covariance analysed by
Kuragano and Kamachi (2001) for TOPEX/POSEIDON altimetry. The method will be changed to a
reduced space 3DVAR-IAU method, for salinity improvement, with a nonlinear descent method when it
becomes available.

Input Data

Forcing data: JMA-NWP real time analysis (NuSDaS) for operation and NCEP daily mean forcing for
research (reanalysis and different forcing comparison).

In-situ data: from the GTS with JMA QC procedures applied. ARGO float T-S data are used.

Altimetry: TOPEX/POSEIDON (and will be changed to Jason-1 in 2002)

SST: JMA OI-analysis with satellite and ship data. It will be changed to J_GODAE High Resolution
SST (GHRSST) produced by Tohoku University and JAXA for research version and by JMA for
operational version


JMA provides the state variables through NEARGOOS data server (RRTDB) and JMA GODAE
product server (http: godaelas.kishou.go.jp:8080/las/servlets/dataset). Access is permitted for
registered group only. JMA and MRI have a plan of reanalysis (version 1: 1993 to 2007; version 2:
1980‘s to 2007).

GODAE IP July 20-22, 2004                        Page 17                                   11/06/10 19:06
Targeted Users

The targeted users are, ocean research community and other agencies (e.g., Japan Coastal Guard,
Japan Fisheries Agency). Regional and coastal applications will be discussed.

Prototype system

The group already started to operate a near real time assimilation system. It also started to develop
the real time forecasting system. JMA and MRI groups participate intercomparison projects in the
North and Equatorial Pacific and Global Ocean. JMA and MRI groups will use the own
supercomputers (Hitachi sr8000 in JMA, and NEC SX6 in MRI).

7.5.3   JMA/Climate Prediction Division (CPD): ODAS

The ocean model used is a JMA-OGCM. The model is a rigid-lid version in the global ocean. It
includes a nonlinear dissipation scheme with deformation fields. Mixed layer scheme is the Mellor-
Yamada‘s turbulent closure scheme, level 2.5. The grid is a variable grid, 2.0x2.5, 20 vertical levels,
and 0.5x2.5 around equator. The model will be changed to a free surface version and then a new
community model, MRI.COM, when it becomes available.

Assimilation Method

The assimilation system uses Derber-Rosati type 3DVAR with ship and TAO-TRITON temperature
and TOPEX/POSEIDON altimeter data. The observation error covariance is proportional to the vertical
gradient of the temperature. Salinity assimilation will start in the next version. The method will be
changed to a reduced space 3DVAR-IAU method with a nonlinear descent method when it becomes

Input Data

Forcing data: JMA-NWP real time analysis (NuSDaS).

In-situ data: from the GTS with JMA QC procedures applied. TAO-TRITON data are used.

Argo‘s T-S data will be used.

Altimetry: TOPEX/POSEIDON (and Jason-1) altimetry will be used.

SST: JMA OI-analysis with satellite and ship data.


JMA provides the SST data through internet.

Targeted Users

Initial condition for seasonal-to-interannual prediction with a coupled GCM.

Temperature data are for climate research community.

Prototype system

JMA/DCI already started the near real time operation.

GODAE IP July 20-22, 2004                       Page 18                               11/06/10 19:06
7.5.4   Kyoto University and Japan Marine Science Foundation (MSF)

The ocean model used is a KYOTO OGCM (based on MRI and CCSR OGCM). The model is a free-
surface version in the global ocean. The sigma-z hybrid vertical coordinate is adopted. It includes
Takano-Oonishi scheme for treating steep bottom topography with generalised Arakawa scheme for
momentum advection terms and 3rd order tracer advection terms. Noh‘s mixed layer and Gent-
McWillams eddy parametarization schemes are adopted. The grid is 1x1, 34 vertical levels.

Assimilation Method

The assimilation system uses a four-dimensional variational (4DVAR) method.

Input Data

Forcing data: COADS dataset (da Silva, 1994) and OMIP dataset (based on ECMWF reanalysis,
Roske 2001).

In-situ data: World Ocean Atlas 94 (Temperature, salinity, SST and SSS) and WOCE Climatological
dataset (Gouretski and Jancke, 1999). Argo float data will be used.

Altimetry: TOPEX/POSEIDON (and will be changed to Jason-1 in 2002) altimetry will be used.


The group has a plan to provide the state variables forcing on Oyashio-Kuroshio confluence region, in
collaboration with MSF, which will be delivered through internet along with short-range forecast data.

Targeted Users

Targeted users are climate research community; initial condition for experimental prediction of
seasonal-to-interannual variabilities with a coupled GCM. Regional to coastal applications are
discussed in the group and agencies.

Prototype system

The Kyoto University group has a plan to operate the system as a research and delayed mode for the
GODAE period.

7.5.5   Frontier Group (1): Frontier Research System for Global Change
        (FRSGC)/Integrated Modelling Research Program (IMRP) and Kyoto University

The ocean model used is the MOM3. The model is a free surface version in the global ocean. The grid
is 1x1, 34 vertical levels. The Pacific ocean model with a resolution of 1/4x1/4, 34level has be
developed and will be nested to the global model.

Assimilation Method

The assimilation system uses a four-dimensional variational (4DVAR) method. The adjoint code is
derived with the Tangent-linear and Adjoint Compiler (TAMC) of Giering and Kaminsky (1997) and
partly by hand.

Input Data

Forcing data: NCEP-monthly mean wind stress.

GODAE IP July 20-22, 2004                     Page 19                                 11/06/10 19:06
In-situ data: Argo float. WOA01

Altimetry: TOPEX/POSEIDON (and will be changed to Jason-1 in 2002)


The above 4D-VAR ocean data assimilation system has been applied to seasonal climatological
experiments and obtained a comprehensive dataset with high accuracy and good dynamical
consistency. The group is now in the process of making a reanalysis dataset in the 1990‘s. The four
dimensional state vector will be provided through internet.

Targeted Users

Targeted users are climate research community for understanding of 1990‘s El Nino.

Prototype system

The group has a plan to operate the system as a research and delayed mode for the GODAE period.

7.5.6   Frontier Group (2): Frontier Research System for Global Change (FRSGC)/ Climate
        Variations Research Program (CVRP): J-COPE

The ocean model used is the Princeton Ocean Model (POM). The model has a vertical sigma
coordinate with 1/4x1/4, 21 vertical levels in the North Pacific ocean. The regional model has the grid
1/12x1/12, 35 vertical levels with a one-way nesting to the North Pacific model.

Assimilation Method

The altimeter data are converted to the subsurface temperature and salinity fields with a correlation
method. The assimilation system uses an optimum interpolation and IAU method.

Input Data

Forcing data: heat flux and wind stress calculated from the NCEP-6hourly Reanalysis data.

Altimetry: TOPEX/POSEIDON and ERS-2 (and will be changed to Jason-1 in 2002).


The four dimensional state vector in the Kuroshio region is provided through internet and CD-ROM.

Targeted Users

Targeted users are ocean research community for understanding of Kuroshio variability and
experimental coastal prediction.

Prototype system

The group already started to operate a near real time assimilation system.

GODAE IP July 20-22, 2004                      Page 20                                 11/06/10 19:06
7.5.7   Kyushu University/Research Institute for Applied Mechanics (RIAM)/Dynamics
        Simulations Research Centre (DSRC)

An eddy resolving model coded at RIAM, Kyushu University will be used. The model named RIAMOM
allows free surface and has Takano-Oonishi scheme for treating steep bottom topography with
generalised Arakawa scheme for momentum advection terms. The model covers entire Japan/East
China Sea with 1/12x1/12 degree and 40 vertical levels.

Assimilation Method

The assimilation method is a reduced-order Kalman filter with sufficient resolution for the mesoscale

Input Data

Forcing data: JMA-NWP wind stress and fluxes.

In-situ data: Ship data in the Tsushima Strait for boundary condition.

Remote-sensing: TOPEX/POSEIDON (and will be changed to Jason-1 in 2002), ERS, and PALACE.


RIAM/DSRC has a plan to provide the state variables through internet in future.

Targeted Users

RIAM/DSRC has a plan to develop an experimental prediction system in the Japan Sea. The targeted
users are fisheries, agencies, and university researchers.

Prototype system

RIAM/DSRC has a plan to operate the system (the RIAMOM and Kalman filter) as a research and
delayed mode for the GODAE period.

7.5.8   Data and Product serving
Japanese contributions GODAE data and product servers would be made by the Japan Meteorological
Agency (JMA), Japan Oceanographic Data Centre (JODC) in the Japan Coast Guard (JCG),
International Pacific Research Centre (IPRC), Japan Marine Science and Technology Centre
(JAMSTEC) and National Space Development Agency of Japan (NASDA). Those groups are
candidates of the data and product servers, because they have operational or research based data
processing, delivery and analyses systems both in real time and delayed modes.

The activity of the product servers will build on and develop links with existing data centres, such as
data centres at the JMA of real time delivery, JAMSTEC and IPRC of intermediate time delivery for
research works, JCG/JODC as the National Oceanographic Data Centre, and NASDA of satellite data
delivery. The data centres work with ongoing data processing and management initiatives esp.,
NEARGOOS Reagional Real Time and Delayed Mode Databases operated by JMA (RRTDB) and
JODC (RDMDB) and national and regional data centres of Argo operated by JMA and JAMSTEC. JMA
is establishing a GODAE product server, which provides the outputs of an ocean data assimilation
system (COMPASS-K) and the result of global high resolution sea surface temperature analysis

GODAE IP July 20-22, 2004                       Page 21                               11/06/10 19:06
7.6     Norway/Europe

7.6.1     Background

The EC MAST-III DIADEM and EC FP-5 TOPAZ projects coordinated by the Nansen Centre in
Norway, have developed a pre-operational data assimilation system for monitoring and forecasting of
the North Atlantic.

The participants in the TOPAZ project were:

-     Nansen Environmental and Remote Sensing Centre, Norway.

-     Collecte Localisation Satellites, France.

-     Universite Joseph Fourier, Laboratoire des Ecoulements Geophysiques et Industriels, France.

-     Alfred Wegener Institute, Germany.

The system was originally developed through the DIADEM project where the major objective was to
implement and demonstrate novel sophisticated data assimilation methods with an OGCM and a
marine ecosystem model. The DIADEM assimilation system is completed and has been operated in a
real time operation experiment since October 2000 with predictions of marine parameters issued on
the project web page http://diadem.nersc.no. The TOPAZ real-time experiment restarted in January
2003 and bulletins are found on the web-page http://topaz.nersc.no. The further developments and
operations are secured until the end of 2008 through the EC-FP6 MERSEA integrated project.

7.6.2     The DIADEM/TOPAZ Ocean Forecasting System
Input data

In both the DIADEM and TOPAZ projects, we have relied on data processing and delivery from the
exising data centres in Europe. In TOPAZ, Sea Level Anomaly (SLA) and Sea Surface Temperature
(SST) data were assimilated into the physical model. The SLA data are processed and delivered
through the SSALTO/DUACS system at CLS. For SST the global Reynolds product has been replaced
by the CLS SST with higher resolution 1/4 degree. The latter product does not use proxies to fill cloudy
regions but the EnKF handles this lack of information consistently.

In the TOPAZ project we also assimilate in the coupled ocean-ice system satellite measured ice
concentrations (SSM/I from The National Snow and Ice Data Center, Colorado, USA) since September
2003 and we will assimilate in situ data from e.g. the ARGO floats program, which are processed and
distributed through the French Coriolis program. It is expected that the latter data types will be
assimilated in operational mode from spring 2004. The atmospheric forcing data (hindcast and
forecast) are obtained from ECMWF.

Data serving

      1. TOPAZ bulletins and animations: http://topaz.nersc.no displays horizontal maps of all model
         hybrid layers. A second version of the web-page showing ice parameters and sections will be
         released during spring 2004. Open access.

      2. TOPAZ standard NetCDF files (MERSEA                  Class1   to     Class3   products)    are   on
         http://mersea.eu.org in open access.

      3. High     resolution monthly    forecasts       for    the     Gulf     of     Mexico      are    on
         http://www.oceannumerics.com (restricted)

GODAE IP July 20-22, 2004                         Page 22                                  11/06/10 19:06

In the DIADEM project the Miami Isopycnic Coordinate Ocean Model (MICOM) was used in a
configuration for the North Atlantic, the Nordic Seas and the Arctic Ocean. A stretched grid is used
with enhanced resolution in the Gulf Stream Extension and the European coastal seas, and then
gradually coarser resolution when moving northward into the Arctic Ocean and southward into the
South Atlantic. In September 2001 the MICOM model was replaced with the Hybrid Coordinate Ocean
Model (HYCOM) which improves the major weaknesses in MICOM, e.g. by introducing vertical
resolution in the upper mixed layer and by using the KPP vertical mixing scheme. In January 2003, a
new grid with more even resolution (from 18 to 36 km) has been introduced, covering the Arctic and
full Atlantic basins to approx. 60 degrees South. A new thermodynamic ice model has been coupled to
HYCOM. Further, the system now introduces a downscaling to the coastal zones by nesting of high
resolution regional versions of the model (4km on the North Sea, 5km in the Gulf of Mexico). Three
ecosystem models with increasing complexity have been coupled to HYCOM but are not currently run
in near real-time. The MPI parallelization of HYCOM is being implemented on the supercomputer in

Assimilation method

The system is currently being operated using the Ensemble Kalman Filter (EnKF). The numerical cost
associated with the use of these schemes is justified by the benefit of having time evolving error
statistics and a fully multivariate analysis step. In a practical application, this implies that less time
needs to be spent on calibration or engineering of the assimilation scheme, but the CPU requirements
becomes higher than for optimal interpolation schemes. Wall-clock times however can remain
relatively small because ensemble model integration is naturally parallel.

Prototype system

The current TOPAZ system covers the Arctic and full Atlantic domains down to 60 degrees South.
HYCOM runs with 22 hybrid layers. The horizontal resolution is relatively homogeneous (between 18
and 36 km) and avoids a singularity at the North Pole. The model resolution will be doubled in 2006
and the ecosystem model will be run in real time assimilating ocean colour data.

The assimilation cycle is weekly, with reference days on Wednesday 00:00 UTC. The processing
times for TOPAZ are following:

    1. Analysis: assimilating SLA, SST and ice concentrations (~250.000 observations) takes 1 hour
       parallelized on 4 CPU.

    2. Propagation: takes between 1 or 2 days depending on machine load. Parallelization adapts
       dynamically to available computer resources.

There are currently no plans for extending the system to the global scale. A major proposal to do so
was submitted October 2001, but did not get funded. A global model will not be set up before the
beginning of 2005.

Assimilation products and dissemination

TOPAZ is an operational monitoring and prediction system for the North Atlantic and will be tailored to
meet the needs for marine forecasts from the off-shore oil industry operating in the harsh deep-water
areas in the Gulf of Mexico and other regions in the Atlantic. The system will be developed to deliver
real time products of ocean currents and other physical variables, to support specific user needs in the
off-shore oil industry. Monthly high-resolution forecasts for the Gulf of Mexico will be marketed to the
oil industry and their access is restricted.

General     products  are    displayed     on   Internet     server   (http://topaz.nersc.no   and
http://www.oceannumerics.com) for open dissemination to the science community and other interested
users. HYCOM and the EnKF will be implemented at the Norwegian Met Office (Met.no, aka. DNMI),
for operational use.

GODAE IP July 20-22, 2004                       Page 23                                  11/06/10 19:06
Links with GODAE pilot projects

The assimilation of Argo data in TOPAZ with the EnKF has been evaluated and will be used in the
real-time experiment from March 2004. The error variance maps posterior to the assimilation of T/S
profiles in the EnKF will contribute to an assessment of the present network and recommendations for
optimal placement of future in-situ observations.

Assimilation of non-linear measurements (Temperature Brightness as from SMOS) has been studied
and proved the efficiency of Monte-Carlo methods for assimilating non-linear measurements.

Internal metrics and intercomparison plans

TOPAZ is one of the four components of the EC Integrated Project MERSEA and complies to the
standards defined within MERSEA Str1. Standard interpolated fields for the North Atlantic are open
access on the NERSC Live Access Server (LAS), accessible through the MERSEA Str.1 project page
http://mersea.eu.org and are being compared to NRL, MERCATOR and FOAM output.

Targeted users

Linked to the TOPAZ project is a user group consisting of 13 international oil companies. The TOPAZ
system is regularly being presented to the user group and there is an ongoing interaction on the
optimal tailoring and presentation of products to meet the oil industry's needs. Within the ESA project
EMOFOR, an oil company operating in the Gulf of Mexico has been approached and a service trial will
begin in Feb. 2004 to evaluate monthly forecasts of a high resolution (5km) HYCOM model of the Gulf
of Mexico, nested into TOPAZ.

The links towards other user groups (ship routing in high latitudes, oil spills) will be addressed within
the MERSEA IP EC project and the ROSES project funded by ESA.

Reanalysis activities:

No reanalysis activities planned, efforts concentrate on the forecasting activities.

Computing resources:

NERSC has a privileged access to the computing resources of the University of Bergen: an IBM
Regatta administrated by Parallab.

       3 eSeries p690 Regatta SMP nodes

       96 Power4 1.3 Ghz processors (32 cpus per node)

       320 Gigabyte memory (64 + 64 + 192 Gigabyte)

       UNIX operating system (AIX 5.1)

Detailed description on http://www.parallab.uib.no/resources/regatta/

A major upgrade is planned for 2006, the machine will be replaced by a distributed memory


The system developed has similarities with the other major initiatives in GODAE and will be
complementary in many respects. Further, the system is one of the major initiatives contributing to the
EuroGOOS task teams, in particular the Atlantic Task Team, the North West Shelf Task Team and the
Arctic Task Team.

GODAE IP July 20-22, 2004                        Page 24                                11/06/10 19:06
7.7       UK

7.7.1     Background
The Natural Environment Research Council (NERC) is the main funding body for oceanographic
research within the UK. In addition to University departments, they fund work at Southampton
Oceanography Centre (SOC), Proudman Oceanography Laboratory (POL), and Plymouth Marine
Laboratory (PML).

The Met Office has produced operational surface wave and storm surge forecasts for 2 decades. In
the last decade, it has started to provide daily operational forecasts for the deep ocean using the
Forecasting Ocean Assimilation Model (FOAM) system, and forecasts for the shelf seas using a model
developed by POL. It is also developing an operational system for seasonal prediction.

The Inter Agency Committee for Marine Science and Technology coordinates the activities of the
above agencies.

FOAM, the main UK contribution to GODAE, is described below. In addition, some background
information on the OCCAM global 1/4° data assimilation system is given, as OCCAM will contribute to
GODAE for research applications.

7.7.2     Met Office Forecasting Ocean Assimilation Model (FOAM) System
1. Input Data

         Forcing data: six-hourly Met Office real-time full resolution 6-day forecasts of wind stress, wind
          mixing energy, penetrating heat flux, non-penetrating heat flux and precipitation minus
          evaporation. The NWP system calculates fluxes over sea-ice and open water ("leads")
          separately and combines them using sea-ice concentration analyses generated by NCEP.
          Steps are being taken to archive higher frequency fluxes for the first 6 hours of each forecast.
         In situ profile data: BATHY, TESAC and BUOY messages from the GTS (these message
          formats are used to report VOS XBTs, Argo and TAO/Triton data respectively). Both
          temperature and salinity data are assimilated. Quality control checks on the data include track,
          stability, background and buddy checks.
         Altimeter data: from CLS presently only Jason-1, Envisat and GFO twice a week.
         Surface temperature data: Observations from ships and drifting and moored buoys are
          assimilated. At present only AVHRR data on a coarse grid (2.5 spacing) are assimilated. We
          will upgrade to using GHRSST satellite data products when they become available.
         Sea ice concentration data: fields from Canadian Met Centre (CMC) based on SSM/I data.
2. Data Servers
In situ data and some satellite data are currently accessed through the Global Telecommunications
System (GTS). Satellite altimeter data and sea-ice concentration data are accessed by ftp. The Data
and Products Distribution System (DPDS) supplies full resolution global model fields by ftp in GRIB
format on an operational basis to national meteorological centres and customers. For more details see
http://www.metoffice.com/research/ocean/operational/dpds/dpds_foam.html . Products are also
distributed through a Live Access Server at ESSC (see section 6).

3. Model

The FOAM ocean and sea-ice model is similar to HadOM3 (Gordon et al. 2000), except in the
following respects. A third order accurate upwind interpolation scheme is used for advection of tracers
(Holland et al. 1998) and the velocities used to advect momentum are calculated using the method of
Webb (1995). A rigid-lid is used with a formulation which avoids the Killworth instability (Bell 2000,
appendix A). The Brown & Campana (1978) pressure averaging technique is used in some
configurations to increase the model timestep. River inflow is based on climatological data from the
Global Rivers Data Center (GRDC)..The nesting of models is one-way. It uses the Flow Relaxation

GODAE IP July 20-22, 2004                         Page 25                                  11/06/10 19:06
scheme (Davies 1976) with the bathymetries of the models in the nesting region prescribed to be as
similar as possible. In models of the Atlantic, a 1 Sv exchange flow through the Gibraltar Straits is
specified using the scheme developed by Roberts (2003). Storkey (2004) provides a self-contained
description of the model formulation.

4. Assimilation method

Data assimilation is based on a new version of the analysis correction scheme originally devised by
Lorenc et al. (1991) and implemented for FOAM by Bell et al. (2000). The new version (Bell et al.
2004a) provides a sub-optimal approximation to a variant of 4D variational assimilation. Analysis steps
are performed once per day. Each observation makes its full impact on the model on the day it arrives
and on subsequent days is taken into account by giving additional weight to the model at the
observation‘s location. Each analysis step consists of a number of iterations. On each iteration the
observations are separated into groups which are easily related (thermal profiles, saline profiles,
surface temperature, surface height), For each group of observations (e.g. the temperature profile
data), increments are calculated first for the directly related model variables (e.g. the temperature
fields). These increment fields are then used to calculate increments for less directly related model
variables (e.g. the velocity fields) using hydrostatic and geostrophic balance relationships, water
property conservation or statistical relationships. These balancing increments make the analysis
multivariate. Increments are also made to the observations (Bratseth 1986) so that the iterations
converge towards the statistically optimal analysis. The univariate components of the model error
covariance are specified as the sum of two 3D error covariances, one describing the ocean
mesoscale, the other large scales including atmospheric synoptic scales. These and the observation
error covariances are estimated from statistics of observation minus model values obtained from
hindcast assimilations. Altimeter data are assimilated by displacement of isopycnal surfaces (an
extension of the Cooper & Haines 1996 scheme). A pressure correction technique (Bell et al. 2004b)
is employed to improve the dynamical balance near the equator.

5. Prototype Systems
A global version of FOAM on a latitude-longitude grid with 1 spacing and 20 levels has produced 5-
day forecasts daily in the Met Office operational suite since 1997. A nested model covering the Atlantic
and Arctic with a grid spacing of 35 km and 20 levels was introduced into the operational suite in
January 2001. Five-day forecasts have been made daily by a nested model covering the North Atlantic
with a 12 km grid since April 2002. Nested models covering the Indian Ocean with a 35 km grid and
the Arabian Sea with an 11 km grid have produced 5-day forecasts on a daily basis since 1 September
2001. FOAM forecasts for the Mediterranean on a 12 km grid have also been produced since October
2002. Forecasts for the Antarctic on a 27 km grid have been produced since Feb 2004.

All of these models can also be integrated in hindcast mode (from May 1997 to the present) using
archived six hourly fluxes and observations.

6. Assimilation Products and Dissemination

Full resolution FOAM data from the N Atlantic model is being made available on the Environmental
Systems Science Centre (ESSC) web site http://www.nerc-essc.ac.uk/las for the duration of GODAE in
real-time. This server provides an on-line disk archive of these data with access using the live access
server (LAS) software. The surface fluxes used by the FOAM system are also being delivered to
ESSC and will be made accessible to GODAE partners.

The Navy access FOAM data through the Met Office's Horace forecaster workbench system originally
developed for numerical weather prediction.

The FOAM operational analyses and forecasts are retained in the operational archive on the Met
Office's MASS archive and retrieval system.

7. Links with GODAE pilot projects (ARGO, GHRSST)

Jon Turton of the Met Office is manages the UKARGO project. The FOAM system assimilates the
temperature and salinity data generated by the Argo system.

GODAE IP July 20-22, 2004                      Page 26                                 11/06/10 19:06
Craig Donlon of the Met Office is the leader of the GHRSST project and Director of the GHRSST
Project Office. The FOAM system will assimilate GHRSST products when they are available.

8. Internal Metrics and Intercomparison plans

FOAM is using the internal metrics agreed for the Mersea Strand-1 project.

In addition, many observation minus forecast statistics are calculated for selected areas and by station
identifier for profile data.

9. Targeted Users and envisioned external metrics

The primary user of FOAM is the Royal Navy.
The 1/3 Atlantic FOAM configuration is used to drive the NW European shelf seas model developed
by POL and operated by the Met Office.

High-resolution hindcasts and forecasts of currents by FOAM configurations in the Intra-Americas
Seas are being evaluated. Their value for oil companies search and rescue, and oil/chemical spill
forecasting will be assessed.

Developments for hindcasts and forecasts of phytoplankton, zooplankton and carbon air-sea fluxes are
being made in collaboration with NERC through the CASIX (Carbon Air-Sea Interface flux) project.

British Maritime Technology Marine Information Systems (BMTMIS) have developed tools for search
and research (SARIS), oil spill (OSIS) and chemical spill (CHEMSIS) management and interfaces to
the FOAM and shelf seas data (http://www.bmtmis.com/).

10. Reanalysis activities

A re-analysis using the FOAM global, Atlantic and North Atlantic models for the period 1997-2007 is
planned to be made in 2007. A 40-year reanalysis is being undertaken by the seasonal forecasting
group at the Met Office using the GloSea system which is similar to the global FOAM system.

11. Computing Resources

The Met Office has an NEC SX6 with 30 nodes. The FOAM system currently uses about 1% of this

7.7.3   OCCAM Ocean Forecasting System
The OCCAM model was developed at SOC and UEA. It is a z-coordinate model based on the MOM
code with a free surface and with a rotated grid in the N Atlantic to permit full coverage of the arctic
basin. Data assimilation code for this model has been developed at Edinburgh University and now at
the Reading University Environmental Systems Science Centre (ESSC). The model has so far been
run at ¼ 36 level resolution in an analysis of the period 1992-1996 with ECMWF wind forcing and in
situ and altimeter data assimilation Fox et al (2000) and Fox and Haines (2002). The data from this run
is available for browse and download at www.nerc-essc.ac.uk/las. The model is still in development
and future runs will have improved physical parameterisations for interior mixing, sea ice, and the
bathymetry, as well as improved data assimilation schemes. A new version of the code, which can run
at 1/12o global resolution, is also being developed at SOC. There are currently no plans to run this
model for real time forecasting.

7.8     USA

7.8.1   Background
The U.S. contributions to GODAE are focused towards developing the next generation operational
global forecast capabilities at short-term (mesoscale) and seasonal-to-decadal climate scales for U.S.

GODAE IP July 20-22, 2004                       Page 27                                11/06/10 19:06
operational Agencies: the U.S. Navy and NOAA (National Oceanographic and Atmospheric
Administration). The goals are to:

       Develop products and improved assimilation methodologies that integrate diverse data
        streams for
       Real-time Navy and NOAA operational ocean activities such as maritime safety and forecasts
        of the coastal environment;
       Initialisation of seasonal-to-decadal climate forecast models; and
       Descriptions of the ocean state, including historical estimates akin to the atmospheric
        reanalyses, in support of research investigations such as CLIVAR.
       Contribute to the design of an integrated observing system for mesoscale and climate
        applications through the assessment of observations and surface forcing fields in the context
        of ocean data assimilation.
The approach is to build upon existing operational capabilities, using them as the baseline against
which to measure improvements. These operational capabilities constitute the core US real-time

7.8.2   Product Dissemination
The real time products will be disseminated through the Live Access Server (LAS) on the GODAE
server in Monterey. The NOAA operational products from the National Centres for Environmental
Prediction (NCEP) will also be available from their server. Other groups will disseminate their own
products, with links to them provided through the GODAE server in Monterey.

The existing capabilities, prototype plans for the GODAE development phase, and anticipated
capabilities for the GODAE operational phase are summarized below.

7.8.3   NCEP/ Global Climate and Weather Modeling Branch (GCWMB)
NCEP/GCWMB's goal in ocean data assimilation is to maintain and improve their system for producing
ocean initial conditions for seasonal to interannual forecasts with a coupled ocean-atmospheric model.
The current version of the NCEP Global Ocean Data Assimilation System (GODAS) became
operational in September 2003.


The ocean model used by NCEP in GODAS is MOM.v3 from GFDL. The NCEP version includes the
isoneutral scheme of Gent and McWilliams (1990) for mixing tracers and the KPP boundary layer
mixing scheme of Large, McWilliams and Doney (1994). It also has an explicit free surface and uses
partial bottom cells to better resolve the topography. The grid is quasi-global (no Arctic Ocean) and
                   O       O
extends from 75 S to 65 N. The resolution is 1 degree globally with enhanced meridional resolution
in the tropics. There are 40 levels in the vertical and each of the top 20 levels is 10 meters thick. NCEP
will transition to MOM4 for its next implementation of GODAS.

Assimilation method

The assimilation system uses a three-dimensional variational (3DVAR) method that is an extension of
the work of Derber and Rosati (1989). The background error covariances are represented by Gaussian
functions in the horizontal and vertical. The background error variance is allowed to vary by grid point
and is represented by a function that is proportional to the vertical gradient of temperature or salinity.
The system assimilates temperature and salinity profile data. The salinity data is currently in the form
of synthetic profiles computed from temperature profiles and the local climatological TS relationship.
Testing is underway to include directly observed salinity profiles from Argo.

GODAE IP July 20-22, 2004                       Page 28                                  11/06/10 19:06
Input Data

         Forcing data: NCEP atmospheric analyses from the Global Data Assimilation System
         In situ data: from the GTS with NCEP QC procedures applied.
         SST: NCEP weekly SST OI product.
Assimilation products and dissemination

GODAS currently produces averages of the state variables and forcing fields on the full model grid in
the NetCDF format. Because GODAS runs daily, this represents a considerable amount of data. It
has not yet been decided how much of this data will be served outside of NCEP and in what format it
will be served.

Targeted Users

GODAS products are generated for use as ocean initial conditions for NCEP‘s seasonal-to-interannual
forecasts. They will also be made available in the near future to outside users as estimates of the
ocean state.

Operational system

The NCEP GODAS has been operational since September 2003. It runs daily and produces daily
averaged analyses of the ocean state. Because GODAS uses observations in a time-window
extending from two weeks in the past to two weeks in the future, the date of the most recent analysis
lags the current date by 14 days.

7.8.4   U.S. Navy Operational System at FNMOC
Plans for GODAE Intensive Period (2003-2007)

Oceanographic Forecast Model

FNMOC will make output from the global ocean model component of its coupled air-sea system
available for GODAE. The current candidate model for this component is the Los Alamos Parallel
Ocean Program (POP) model, which is of GFDL heritage with an explicit free surface. The model will
be run in a coupled mode with the Navy Operational Global Atmospheric Prediction System
(NOGAPS). The model will evolve throughout the course of GODAE, with a target configuration of
1/10˚ latitude-longitude resolution with 30-40 levels in the vertical. Towards the end of the GODAE
intensive period, the model will also be coupled to a sea ice prediction system. POP may be replaced
with another similar ocean model during the course of GODAE.

Forcing fields

In 2006, FNMOC plans to increase the resolution of their global atmospheric prediction model to T479
(spectral modes), which corresponds to about 27 km horizontal grid spacing.

Ocean Data Assimilation

The FNMOC ocean data assimilation system uses a 3D multivariate optimal interpolation analysis
(NCODA, NRL Coupled Ocean Data Assimilation).                The subsurface analysis variables are
temperature, salinity, geopotential, and the u, v velocity components. The surface analysis variables
are sea ice concentration, SST, and sea surface height anomalies (SSHA).

Input Data:
       In situ data: temperature, salinity, and velocity observations from the GTS with FNMOC QC
        procedures applied.

GODAE IP July 20-22, 2004                     Page 29                                11/06/10 19:06
       Altimetry: near real time Jason, Topex/Poseidon, ERS-2, ENVISAT, and GFO data from the
        Altimeter Data Fusion Centre (ADFC) at NAVOCEANO.
       Satellite SST: AVHRR, GOES, microwave retrievals from NAVOCEANO.
       Sea Ice: DMSP SSM/I retrievals from FNMOC.
Surface Wave Model

The WaveWatch III model will be run globally on a 1/2 degree latitude-longitude grid (or finer) with 15-
degree angular resolution (or finer) for the directional spectra. Surface wind stress fields generated by
NOGAPS provide the atmospheric forcing. Model outputs are directional wave spectra from which a
number of parameters, including significant wave height, sea height, swell height, peak wave period
and peak wave direction, are derived. The global grid will increase by a factor of 2 (1/4 degree
horizontally) following the upgrade of NOGAPS to T479.

Assimilation Products

Unclassified FNMOC products will be available from the GODAE server in Monterey. Products will be:
       3D fields of T, S, and ocean currents along with surface fields of sea ice, SST and SSHA on a
        Mercator grid (65S to 65N) with 1/3 degree horizontal spacing at the Equator.
       2D fields of sea ice and SST at the same resolution as the NOGAPS fields described above
       Global surface wave products
       Quality controlled ocean observations in real-time.
Targeted Users

The U.S. Navy products are primarily targeted toward the Fleet; however, unclassified products are
available routinely on the Web (https://www.fnmoc.navy.mil/PUBLIC/). Many FNMOC products are
also available on the GODAE server, as described in Section 6.2.1.

7.8.5   U.S. Navy Operational Systems at the Naval Oceanographic Office
The Naval Oceanographic Office runs a suite of operational ocean prediction systems from very high-
resolution coastal through regional and global domains. The coastal systems are particularly targeted
for specific Navy applications with the global domain primarily required for initial and boundary
conditions for the smaller area systems. The operational, GODAE-related global ocean prediction
systems, developed at and transitioned by the Naval Research Laboratory/Stennis Space Center
(NRL/SSC), allow the nowcasting and forecasting of the mesoscale, large scale and upper ocean. The
present global systems using NLOM in combination with NCOM and the MODAS data analysis system
comprise the present GODAE-related operational capability. NAVOCEANO anticipates the next-
generation, operational, global system to be based on HYCOM.

Input data

         Atmospheric Forcing: from the FNMOC Navy Operational Global Atmospheric Prediction
          System (NOGAPS)
         SSH: sea surface height from real-time altimetry via the NAVOCEANO Altimeter Data
          Fusion Center (ADFC): ENVISAT, GFO, Jason-1
         SST: sea surface temperature from satellite infrared
         3D synthetic T and S fields: generated from SSH and SST fields using the MODAS
          system algorithms and statistics.
         In Situ data: real-time hydrographic data, including ARGO
Data Serving

GODAE-related, operational, digital output from NLOM and NCOM surface nowcast and forecast fields
are available via NRL anonymous ftp. Data serving via an NRL LAS is planned. (See NRL Section

GODAE IP July 20-22, 2004                       Page 30                                 11/06/10 19:06

Two global ocean models are in operational or near-operational use at NAVOCEANO: the Navy
Layered Ocean Model (NLOM); (Hurlburt and Thompson, 1980; Wallcraft et al., 2003) and the Navy
Coastal Ocean Model (NCOM); (Barron et al., 2004a). Designed around the available operational
computational capabilities, NLOM and NCOM are coupled as a system to take optimal advantage of
the respective horizontal and vertical resolutions of each. (See NRL Section for more detail)
NLOM has been operational since fall 2001. Global NCOM is currently running in real-time,
undergoing operational testing and expected to be fully operational by fall 2004

Assimilation Methods

The Modular Ocean Data Assimilation System (MODAS) is an optimal interpolation-based analysis
using the previous analysis as the first guess. It currently runs at 1/8˚ globally. The SSH and SST are
derived from NAVOCEANO operational altimeter and AVHRR MCSST products respectively.
Covariance functions have been derived from several years of satellite-based SST and SSH
observations. MODAS also computes synthetic 3D grids based on real-time, gridded SST and SSH
analyses using statistical relationships, derived from the historical databases, between subsurface
temperature, SST and SSH. Salinity is computed from the derived temperature using local
climatological relationships between T and S. The specific assimilation procedures for global NLOM
and NCOM and how each use MODAS-derived fields are detailed in NRL Section

Prototype systems

The Naval Oceanographic Office does not specifically run prototype systems. However, within a year
or two of a global system‘s official transition from NRL, operational queues are provided to allow the
real-time execution of planned transitional systems. NLOM and NCOM have each been run in this
real-time, pre-operational mode prior to their transition. Similarly, global HYCOM is expected to run in
pre-operational mode in the operational job queue before its scheduled 2007 transition to

Assimilation products and dissemination

NAVOCEANO provides its real-time, operational ocean prediction system results via unclassified and
classified channels to its military users. Publicly available examples of the MODAS, NLOM and
NCOM fields are best viewed on the NRL web page http://www.ocean.nrlssc.navy.mil/global_nlom.
(See NRL Section

Links with GODAE pilot projects (ARGO, GHRSST)

ARGO data is assimilated in NAVOCEANO regional analysis and prediction systems today. The initial
delivery of the GODAE-related, global NCOM system does not assimilate in situ profile data in order
that that sufficient independent comparison data might be available for verification during its initial
transition. However in 2005, the in situ profile data (including ARGO) will be incorporated into the
global MODAS and NCOM systems.

NAVOCEANO is represented on the GHRSST Steering Team and is a participant in a GHRSST-
associated multi-institutional project on developing and improving high resolution SST products, a
project funded by the National Ocean Partnership Program (NOPP). The 1/8° MODAS SST analysis
of satellite AVHRR data is an operational product of NAVOCEANO originally developed by NRL.

Internal metrics and inter-comparison plans

NAVOCEANO operational metrics and inter-comparisons typically derive from specific measurement
programs associated with operational ship and aircraft surveys as well as military exercises and
operations. Regular real-time operational evaluations of the GODAE-related MODAS, NLOM &
NCOM products are planned with evaluation tools expected as part of the model transitions from NRL.
Descriptions and examples of these are detailed in NRL Section

GODAE IP July 20-22, 2004                      Page 31                                 11/06/10 19:06
Targeted users and external metrics

NAVOCEANO provides ocean prediction products and their interpretation to military users for military
applications. The prime external metric is user feedback on the operational effectiveness of the
environmental products provided.

Reanalysis activities

NAVOCEANO doesn‘t participate in GODAE-related ocean reanalysis activities. The closest related
activity is the use of historical NCOM runs to create an initial global climatological currents data-base
for evaluation purposes.

Computing resources

NAVOCEANO operates one of four U.S. Defense Department Major Shared Resource Centers
(MSRC) for high performance computing under the auspices of the High Performance Computing
Modernization Office (HPCMO). The bulk of these resources are devoted to research but a fraction of
the total capability of the NAVOCEANO MSRC is dedicated to operational applications. The HPCMO
program is designed to provide a major upgrade to MSRC computer resources every two years. The
existing ~10 TFLOP capability at NAVOCEANO is currently being upgraded with the 2004 upgrade to
~30 TFLOP. The primary source of this increased computing capability consists of two IBM Power 4+
machines, one ~3000 processors and the other ~500 processors. These are complemented by an
additional ~1400 processor IBM Power 4.

7.8.6    U.S. Navy Research Systems at NRL/Stennis
GODAE-related global and basin-scale ocean prediction systems developed at the Naval Research
Laboratory/Stennis Space Center (NRL/SSC) are designed for nowcasting and forecasting of the
mesoscale, large scale and upper ocean. Providing boundary conditions for even higher resolution
coastal and regional models and accurate SST for atmospheric models are important parts of this
effort. The Naval Oceanographic Office (NAVOCEANO) (see Section 7.8.5) and Fleet Numerical
Meteorology and Oceanography Center (FNMOC) (see Section 7.8.4) are the operational customers
for these R & D products. Earlier global systems using NLOM in combination with NCOM and the
MODAS data analysis system were largely developed in house, but a next generation system based
on HYCOM is under development in collaboration with a community effort (see Section 7.8.8 HYCOM

Input data

          atmospheric forcing from the FNMOC Navy Operational Global Atmospheric Prediction
           System (NOGAPS)
          real-time altimetry via the Altimeter Data Fusion Center (ADFC) at NAVOCEANO:
           ENVISAT, GFO, Jason-1
          daily operational 1/8° MODAS SST analyses derived from satellite AVHRR
          3D synthetic T and S fields generated from SSH and SST fields using the MODAS system
           algorithms and statistics.
          Hydrographic profiles, including ARGO
Data Serving

Digital output from NLOM and NCOM surface nowcast and forecast fields are available on anonymous
ftp. Serving via LAS is planned. HYCOM output is served via the University of Miami LAS server (see
Section 7.8.8 HYCOM Consortium).


Three ocean models are in use: the NRL Layered Ocean Model (NLOM) (Hurlburt and Thompson,
1980; Wallcraft et al., 2003), the Navy Coastal Ocean Model (NCOM) (Barron et al., 2004a) and the

GODAE IP July 20-22, 2004                       Page 32                                 11/06/10 19:06
HYbrid Coordinate Ocean Model (HYCOM) (Bleck, 2002). NLOM and NCOM were developed at NRL
and NRL is a major partner in the development of HYCOM (see Section 7.8.8 HYCOM Consortium).
NLOM was designed for use in a first generation global ocean prediction system by allowing high
horizontal resolution at much lower computational cost than other models (Hurlburt, 1984), 1/16° (~7
km mid-latitude) resolution in the system transitioned to NAVOCEANO in October 2000 (Smedstad et
al., 2003) and 1/32° in a planned upgrade that has been running in near real time since November
2003 (Shriver et al., 2004). However, NLOM has only 7 Lagrangian layers in the vertical, including the
mixed layer, and excludes the Arctic and most shallow water. NCOM is designed to complement
NLOM by covering these regions and using much higher vertical resolution, but with coarser 1/8° (15 -
16 km mid-latitude) horizontal resolution. NCOM has a mixed  - z vertical coordinate where 
coordinates are used when the bottom depth is <137 m in the current global configuration. This
configuration has 40 levels in the vertical with vertical resolution ~1 m very near the surface for fine
resolution of the mixed layer. NLOM and NCOM are currently used as components of a single system
(Rhodes et al., 2002, and see Assimilation Methods below).

Assimilation Methods

Several data assimilation techniques have been developed at NRL, including capabilities to assimilate
satellite altimeter data, SST, and to project surface information downward. The 1/16˚ and 1/32 global
NLOM-based systems (Smedstad et al., 2003; Shriver et al., 2004) assimilate altimeter data using an
OI deviation SSH analysis with the model field as the first guess and mesoscale covariances
calculated from T/P and ERS-2 data by Jacobs et al. (2001). A statistical inference technique
(Hurlburt et al., 1990) updates all layers of the model based on the analysed SSH deviations, including
geostrophic updates of the velocity field outside an equatorial band. The global model is then updated
to produce a nowcast using slow insertion to further reduce gravity wave generation. The model is
updated daily with real-time altimeter data from NAVOCEANO. In addition, model deviations from full
field MODAS SST analyses (Fox et al., 2002a) are assimilated in the form of heat fluxes.

The 1/8 global NCOM system (Rhodes et al., 2002; Barron et al., 2004b) uses nudging to assimilate
3-D synthetic T and S fields generated from MODAS SST analyses (Fox et al., 2002a) and NLOM
SSH fields using the MODAS system algorithms and statistics (Fox et al., 2002b). Hydrographic
profiles are added using OI, but an upgrade to the NCODA MVOI scheme (Cummings, 2003) is
planned. The nudging in NCOM is weak from below the surface to the base of the mixed layer to allow
greater model impact on the mixed layer. For HYCOM data assimilation plans, see Section 7.8.8
HYCOM Consortium.

Prototype systems

The 1/16 global NLOM system began running in real time in October 2000 and became an
operational system at NAVOCEANO in September 2001 (See Section 7.8.5 U.S. Navy Operational
Systems at NAVOCEANO). The 1/32 global NLOM system has been running in near real time since
November 2003 and should replace the 1/16 system as soon as NAVOCEANO‘s new IBM P4+ is
ready to run it operationally (late 2004 or early 2005). The 1/8 global NCOM system has been
running in real time since October 2001 and should become operational at NAVOCEANO in the latter
part of 2004.

A prototype 1/12 (~7 km mid-latitude resolution) Atlantic HYCOM system (28S - 70N including the
Mediterranean Sea) has been running in near real time since July 2002, as discussed in Section 7.8.8
HYCOM Consortium. A near real time 1/12 HYCOM Pacific system is planned in the latter part of
2004. These are intended for semi-operational use (not officially operational but used for operational
purposes) by late 2005 and should be replaced by a 1/12 global HYCOM system in 2007 with near
real time running planned to begin in late 2006. Plans are to upgrade to a 1/25 global HYCOM
system by the end of the decade. In addition, a global HYCOM system is planned for FNMOC for use
in a coupled atmosphere-ocean prediction system with FNMOC‘s operational NOGAPS as the
atmospheric component of the system. Substantial effort is devoted to making SST from HYCOM as
accurate as possible. All of the existing systems provide forecasts at least once a week (on
Wednesday), 30 days for the global NLOM systems, 5 days for the global NCOM system and 14 days
for the 1/12° Atlantic HYCOM system.

GODAE IP July 20-22, 2004                      Page 33                                 11/06/10 19:06
Assimilation products and dissemination

Real-time (or near real time) and archived results from the 1/16 and 1/32 global NLOM systems, the
1/8 global NCOM system, the model-independent global MODAS 1/8 SST and 1/4 SSH analyses
and      satellite  altimetry    can     all    be     viewed     on     the    NRL      web   page
http://www.ocean.nrlssc.navy.mil/global_nlom. For all, this includes nowcasts and many zoom regions
for SSH, surface layer currents, SST, and for NCOM, sea surface salinity. In addition, the NLOM web
pages include results of 30-day forecasts, forecast verification statistics, subsurface temperature
cross-sections and profiles, the amount of altimeter data used for each nowcast from each satellite
and nowcast comparisons with unassimilated data. Digital data for nowcast and forecast surface fields
from 1/16 global NLOM and 1/8 global NCOM are provided via anonymous ftp and serving via LAS
is planned. Results from the 1/12 Atlantic HYCOM system can be seen on the HYCOM Consortium
web page http://hycom.rsmas.miami.edu and digital output is available via LAS at the University of
Miami (see Section 7.8.8 HYCOM Consortium).

Links with GODAE pilot projects (ARGO, GHRSST)

ARGO data will be assimilated by the 1/8° global NCOM system and the HYCOM-based systems and
is used for evaluation of the HYCOM, NCOM and NLOM based systems. ARGO and other T,S profile
data will also be used in improving the MODAS (Fox et al., 2002b) synthetic T & S profiles derived
from SSH and SST.

NRL is represented on the GHRSST Steering Team and is a major participant in a GHRSST-
associated multi-institutional project on developing and improving high resolution SST products, a
project funded by the National Ocean Partnership Program (NOPP). The 1/8° MODAS SST analyses
of satellite AVHRR data are an operational product of NAVOCEANO that was developed at NRL (Fox
et al., 2002a).

Internal metrics and intercomparison plans

Verification studies for the 1/16° and/or 1/32° global NLOM systems are discussed in Hurlburt et al.
(2002), Rhodes et al (2002, 2003), Smedstad et al. (2003), and Shriver et al. (2004) and for the 1/8°
global NCOM system in Rhodes et al. (2002), Barron et al. (2004b, 2004c) and Kara et al. (2004). In
addition, the web pages for the 1/16° and 1/32° NLOM systems contain real-time (or near real-time)
verification results plus archives and continuously up-dated summaries of the results. Real-time
forecast verification metrics for SSH and SST include rms error, anomaly correlation, skill score and
for the Gulf Stream and the Kuroshio, axis error. These are plotted daily as a function of forecast
length for each 30-day forecast for numerous subregions of the world ocean. Nowcast SST time
series are compared to daily time series of unassimilated SST from TAO/TRITON, PIRATA and NDBC
buoys and are summarized by error statistics for each time series. As an example, the median of rms
difference in comparison to 84 2-year buoy time series is .31°C for the operational 1/16° global NLOM
system, attesting to the quality of AVHRR SSTs provided by NAVOCEANO and the 1/8° MODAS SST
analyses. Nowcast subsurface temperature profiles are compared to TAO/TRITON and PIRATA buoy
data. T & S profiles from ARGO and CTDs and T profiles from BT data are also used for subsurface
verification. As applicable, the NLOM and 1/12 Atlantic HYCOM web pages include SSH and SST
(for HYCOM) comparisons to real-time operational IR frontal analyses by NAVOCEANO in the Gulf of
Mexico, the Gulf Stream region, the Japan East Sea and the Kuroshio region. The HYCOM and
NLOM web pages include a 5-way nowcast product intercomparison using three 4-panel animations of
nowcast product SSH overlain on SeaWiFS ocean color imagery in the Gulf of Mexico. Unassimilated
daily mean detided SSH from tide gauges are used in arrears in evaluating nowcast SSH from all the
systems and the NLOM 30-day forecasts.

Participation in product intercomparisons is planned. The 1/12° Atlantic HYCOM system is already
part of the European MERSEA ocean prediction system intercomparison effort and the NCOM and
NLOM systems are scheduled to participate as soon as the NRL/SSC LAS is ready to serve the

GODAE IP July 20-22, 2004                     Page 34                                11/06/10 19:06
Targeted users and external metrics

Ocean products developed at NRL are typically targeted for transition to operational use at some
combination of NAVOCEANO, FNMOC, regional centers and shipboard environmental systems. The
NRL web pages are designed to make selected U.S. Navy products available to the world community
of potential users. NRL Oceanography Division web pages received 18.3 million hits during 2003,
most to real-time ocean products. Of these, 25.0% were U. S. military (which generally receives
environmental products from operational centers), 38.7% U. S. civilian and 36.3% from outside the
U.S. There were > 1000 hits from each of 59 different countries. Feedback indicates a wide range of
applications by users around the world (Hurlburt et al., 2002; Smedstad et al., 2003). The applications
include some that were never envisioned during development and some users are located in
developing nations. User product requests, questions, and feedback on value/performance help drive
improvements in web site product presentation, data serving, understanding of the results and
increased appreciation for the value of eddy-resolving ocean products.

Reanalysis activities

A 1/16° global NLOM reanalysis has been run for 1993-2000 and real-time results since then have
been archived. A 1/12° global HYCOM reanalysis covering 1993-2006 is planned in 2007.

Computing resources

NRL R&D for ocean products use grants of high performance computer time from the Defense
Department High Performance Computing Modernization Office. The grants are on a variety of
computing platforms at several different locations, including the NAVOCEANO Major Shared Resource
Center (see Section 7.8.5).

7.8.7   ECCO Consortium
The "Estimation of the Circulation and Climate of the Ocean" (ECCO) Consortium is a partnership
between groups at the Scripps Institution of Oceanography (SIO), the Jet Propulsion Laboratory (JPL),
and the Massachusetts Institute of Technology (MIT). It is formed by the US National Ocean
Partnership Program (NOPP) with funding provided by the Office of Naval Research (ONR), the
National Aeronautics and Space Administration (NASA), and the National Science Foundation (NSF).
ECCO's primary goal is to obtain physically consistent estimates of the time-varying global ocean
circulation for understanding the climatic state and changes of the ocean. The Consortium activities
and progress are documented at http://www.ecco-group.org .

Input Data

       Forcing data: NCEP reanalyses and operational real-time analyses.
       Altimetry: absolute and time-varying sea surface heights from TOPEX/Poseidon, Jason-1,
        ERS-1 and -2.
       Scatterometry: wind stress from ERS-1, and –2, NSCAT, and QSCAT.
       SST: Reynolds, and TMI.
       In-situ data: global XBT, TOGA-TAO temperature profiles, P-ALACE and ARGO temperature
        and salinity profiles, WOCE and pre-WOCE hydrographic sections of temperature and salinity,
        surface salinity observations, drifter 15-m mean velocity.

GODAE IP July 20-22, 2004                      Page 35                                11/06/10 19:06
Data serving None.


The ECCO ocean state estimation is based on the MITgcm (MIT General Circulation Model; Marshall
et al., 1997a, b; http://mitgcm.org). MITgcm is an isomorphic general circulation model for studying the
ocean and atmosphere. The model has a non-hydrostatic capability, uses the finite volume method,
employs the so-called ―lopped cells‖ to accurately represent the bottom boundary position, and
supports a wide range of advanced physical parameterizations. MITgcm has a tangent linear and
adjoint code maintained alongside the forward model which facilitates model sensitivity analysis and
variational data assimilation.

Assimilation method

The ECCO Consortium employs the adjoint method and approximate Kalman filters and Rauch-Tung-
Striebel (RTS) smoothers. The adjoint model is constructed by the Tangent-linear and Adjoint
Compiler (TAMC; Giering and Kaminsky, 1997) and its successor, Transformation of Algorithms in
Fortran (TAF; http://fastopt.de).    The sequential filter and smoother employ a hierarchy of
approximations including time-asymptotic, reduced-state, and partitioning methods (e.g., Fukumori
2002). Controls of the assimilations include open boundaries, initial conditions, surface forcings, and
internal model uncertainties (e.g., mixing coefficients). Differences and synergism among different
methods are being explored and capitalized on to identify the most efficient and effective optimization
method for future analysis.

Prototype systems and transition to global systems

ECCO presently has two near-global (without the Artic Ocean) analysis systems producing products
on an on-going basis. One of them (ECCO-1) is a reanalysis system based on the adjoint approach,
with a 1º uniform resolution and 22 vertical levels. ECCO-1 synthesizes all the data listed under ―input
data‖ above and presently covers the period of 1992-2002. Another product is a near-real time
analysis system (ECCO-2) that has a similar model domain as ECCO-1 but with higher resolution
(0.3º meridional resolution in the tropics and 46 vertical levels). This system uses a Kalman
filter/smoother to assimilate altimetric sea level anomalies (TOPEX/Poseidon and JASON-1), in-situ
temperature profiles (XBT, Argo, and TAO), Levitus climatological temperature and salinity, and NCEP
surface fluxes. ECCO-2 covers the period of 1993 to present, and is updated every 10 days. Adjoint
assimilation with an identical model configuration as the ECCO-2 system is also performed for the
period of 1997-2001. Embedded within these near-global systems are regional eddy-permitting
systems (e.g., North Atlantic) with 1/6º resolution. ECCO is currently implementing a global system
with a 1/4º and 50-level resolution. A cubed-sphere grid configuration is also being tested to have a
truly global system (Artic Ocean included) and to have an eddy-permitting resolution at mid-latitudes
(about 1/6º). Assimilation products with this high-resolution model will be available in mid to late 2004.

ECCO is undertaking an extended effort to further advance ocean state estimation and to facilitate the
initialization of seasonal-to-interannual prediction. This effort, funded under NOPP/GODAE, involves
partners from JPL, MIT, GFDL, NCEP, and GSFC.

Assimilation products and dissemination

ECCO assimilation products are openly available and are being disseminated through the LAS server
at http://ecco-group.org/las. These include near-global, 3-D state of the ocean (temperature, salinity,
currents, sea level, bottom pressure) at monthly and shorter time intervals (up to 12-hourly sea level
and bottom pressure) from 1992 to present.

Links with GODAE pilot projects (Argo, GHRSST)

In the ECCO systems, Argo data are being assimilated along with other in-situ profile data (e.g., XBT).
Separate evaluation of the impact of Argo data or model-data misfit for Argo data are currently not
available. However, a separate ―cost function‖ for Argo data constraint is being implemented in the
adjoint-based synthesis, making it possible to evaluate Argo data constraint separately from other data

GODAE IP July 20-22, 2004                       Page 36                                  11/06/10 19:06
constraints in the future. Due to the relatively coarse resolution of the current ECCO systems,
GHRSST data are not assimilated.

Internal metrics and intercomparison plans

ECCO assimilation products obtained with different assimilation methods and different resolutions are
being compared with one another and with independent observations (including float data and
geodetic observations). ECCO also plans to participate in comparison with other GODAE products.

Targeted users and envisioned external metrics

ECCO products are focused towards the CLIVAR as well as the GODAE communities. Current users
include physical oceanographers as well as scientists in the biogeochemical and geosciences
communities. No formal external metrics are presently implemented.

Reanalysis activities

The smoothed estimates (both the adjoint-based and RTS-smoothers-based), by their nature, revise
earlier estimates as the assimilation period is extended in time. Estimates are also further reanalysed
associated with advances in assimilation system, such as the expansion of the assimilation‘s suite of
control variables (e.g., inclusion of mixing coefficients as control variables).

Computer Resources

Computer resources are presently provided through San Diego Supercomputer Centre, National
Centre for Atmospheric Research, JPL Supercomputing Project, NASA Ames Research Center, and
NAVO. These resources are necessary for the continuation of ECCO, but their future availability is

7.8.8   HYCOM Consortium
A broad partnership of institutions is collaborating in developing and demonstrating the performance
and application of eddy-resolving, real-time global and basin-scale ocean prediction systems using the
HYbrid Coordinate Ocean Model (HYCOM). These systems will be transitioned for operational use by
the U.S. Navy at both the Naval Oceanographic Office (NAVOCEANO), Stennis Space Center, MS,
and the Fleet Numerical Meteorology and Oceanography Center (FNMOC), Monterey, CA, and by
NOAA at the National Centers for Environmental Prediction (NCEP), Washington, D.C. The systems
will run efficiently on a variety of massively parallel computers and will include sophisticated, but
relatively inexpensive, data assimilation techniques for assimilation of satellite altimeter sea surface
height (SSH) and sea surface temperature (SST) as well as in-situ temperature, salinity, and float
displacement. The Partnership addresses the Global Ocean Data Assimilation Experiment (GODAE)
goals of three-dimensional (3D) depiction of the ocean state at fine resolution in real-time and
provision of boundary conditions for coastal and regional models. It will also provide the ocean
component and oceanic boundary conditions for a global coupled ocean-atmosphere prediction model.
It will make these results available to the GODAE modeling community and general users on a 24/7
operational basis via a comprehensive data management strategy. The Consortium activities and
progress are documented at http://hycom.rsmas.miami.edu.

The Partnership represents a truly broad spectrum of the oceanographic community, bringing together
academia, federal agencies, and industry/commercial entities, spanning modeling, data assimilation,
data management and serving, observational capabilities, and application of HYCOM prediction
system outputs. The institutions participating in this Partnership have long histories of supporting and
carrying out a wide range of oceanographic and ocean prediction-related research and data
management. All institutions are committed to validating an operational hybrid-coordinate ocean model
that combines the strengths of the vertical coordinates used in the present generation of ocean models
by placing them where they perform best. This collaborative Partnership provides an outstanding
opportunity for NOAA-Navy collaboration and cooperation ranging from research to the operational

GODAE IP July 20-22, 2004                      Page 37                                 11/06/10 19:06
NAVOCEANO: HYCOM is planned as the model component of the next generation eddy-resolving
(1/12 grid spacing, ~7 km mid-latitude resolution) operational global ocean nowcast/forecast system
at NAVOCEANO. Transition of the HYCOM-based global system from Research and Development to
NAVOCEANO is planned for FY07 and the resolution should increase to 1/25 (~3-4 km at mid-
latitudes) by the end of the decade. These systems would include an embedded ice model and the
capability to host nested littoral models with even higher resolution. The early 2007 date for transition
is timed to coincide with the planned FY06 High Performance Computer (HPC) upgrade, the next
major upgrade after NAVOCEANO‘s existing 2944 processor IBM Power 4+ system installed in 2004.
NAVOCEANO‘s mission encompasses a broad range of oceanographic activities supporting the U.S.
Navy, with global numerical ocean prediction a key element. Scientists at NAVOCEANO provide
specialized, operationally-significant products and services for military ―warfighters.‖ NAVOCEANO
scientists also acquire global ocean and littoral data from sources such as: shipboard surveys, routine
data collected by both operational Navy and commercial ships, remotely-sensed data available from a
variety of airborne and satellite sensors, and drifting and fixed-location buoys. NAVOCEANO is the
U.S. provider of operational real-time satellite altimeter and MCSST data, including to FNMOC and
NCEP. The analyses of these data range from the creation of historical climatologies and real-time
maps of specific measured parameters to the assimilation of real-time data into ocean prediction
systems for temperature, salinity, tides, waves, and currents. The latter systems focus on coastal and
regional nowcast/forecast systems nested within the global prediction systems.

FNMOC: A loosely-coupled and then a tightly-coupled global atmosphere/ocean data assimilation
system based on the Navy Operational Global Atmospheric Prediction System (NOGAPS) and
HYCOM will be evaluated by NRL/Monterey. Transition of this system to operations at FNMOC is
planned in FY09. The ocean component will be addressed under this project and the atmospheric
component and coupling under partnering projects. In addition, the global HYCOM ocean data
assimilation system will be used for initialization and lateral boundary conditions for the globally
relocatable ocean model in the Coupled Ocean/Atmosphere Mesoscale Prediction System
(COAMPS™). FNMOC is the Department of Defense (DoD) primary central production site for
operational meteorological and oceanographic analysis and forecasts products worldwide.

NCEP: HYCOM is planned as the next generation operational ocean model at the NOAA/NWS/NCEP
Marine Modeling and Analysis Branch (MMAB). The MMAB is responsible for the development of
improved numerical marine modeling prediction and analysis systems within the National Weather
Service (NWS). The ocean forecast system will include eddy-resolving basin-wide configurations for
the North Atlantic and Pacific with forecasting capability for US coastal ocean areas, including Alaska
and Hawaii. The operational North Atlantic Ocean Forecast System (NAOFS) (to be in place in 2005)
will provide ocean forecasts for the entire eastern seaboard of the U.S. from the Gulf of Maine to the
Gulf of Mexico, including the Caribbean Seas area. Evaluation of a Pacific Ocean Forecast System will
also take place during this NOPP effort (years 4-5). The nowcast and forecast products will include
sea levels, currents, temperature and salinity. The suite of ocean products will be provided to the
National Weather Service‘s Ocean Prediction Center (OPC), the Tropical Prediction Center (TPC), the
National Ocean Service (NOS) (see attached letters of support from these three centers), and the
project Partners for critical evaluation. The products will be also distributed to forecast offices with
marine responsibilities, and external research and applications oriented communities.

Input data

Available from NAVOCEANO, Monterey GODAE server, CORIOLIS

       Forcing data: 6-hourly NOGAPS, ECMWF
       In situ data: XBTs, Argo
       Altimetry: MODAS SSH, Altimeter tracks
Data serving

The HYCOM products are daily 3D fields of the ocean state available once a week. Data serving is
taking place via a Live Access Server (LAS) (http://hycom.rsmas.miami.edu) for sub-sampled fields
and via ftp for the full fields. HYCOM outputs are freely available.

GODAE IP July 20-22, 2004                       Page 38                                 11/06/10 19:06

HYCOM development is the result of collaborative efforts among the University of Miami, the Naval
Research Laboratory (NRL), and the Los Alamos National Laboratory (LANL) (Bleck, 2002;
Chassignet et al., 2003; Halliwell, 2004). Traditional ocean models use a single coordinate type to
represent the vertical, but recent model comparison exercises performed in Europe (DYnamics of
North Atlantic MOdels - DYNAMO) (Willebrand et al., 2001) and in the U.S. (Data Assimilation and
Model Evaluation Experiment - DAMÉE) (Chassignet et al., 2000) have shown that no single vertical
coordinate -- depth, density, or terrain-following sigma -- can by itself be optimal everywhere in the
ocean. Isopycnal (density tracking) layers are best in the deep stratified ocean, z-levels (constant fixed
depths) are best used to provide high vertical resolution near the surface within the mixed layer, and
sigma-levels (terrain-following) are often the best choice in shallow coastal regions. HYCOM
combines all three approaches by choosing the optimal distribution at every time step. The hybrid
coordinate is one that is isopycnal in the open, stratified ocean, but that makes a dynamically smooth
transition (via the layered continuity equation) to terrain-following coordinates in shallow coastal
regions, and to pressure coordinates in the mixed layer and/or unstratified seas.

The capability of assigning additional coordinate surfaces to the oceanic mixed layer in HYCOM gives
us the option of implementing sophisticated vertical mixing turbulence closure schemes [see Halliwell
(2004), for a review]. The latest release of HYCOM has five primary vertical mixing algorithms, of
which three are non-slab models and two are slab models. The three non-slab models govern vertical
mixing throughout the water column and are the nonlocal K-Profile Parameterization (KPP) model of
Large et al. (1994), the NASA-GISS model of Canuto et al. (2001), and the level 2.5 turbulence closure
algorithm of Mellor and Yamada (1982) (MY). KPP is the default.

 HYCOM has been configured globally, on basin scales, and regionally. The fully global configuration
is currently being integrated with ~60 km mid-latitude resolution and has been coupled to the Los
Alamos ice model (CICE). North Pacific and Atlantic basin-scale simulations have been integrated with
~7 km mid-latitude resolution, our target resolution for the global configuration.

Assimilation method

Several techniques for assimilating data into HYCOM are either in place or under development. These
techniques vary in sophistication and computational requirements and include: Optimum Interpolation
(OI/Cooper-Haines), MVOI/3D-VAR, SEEK filter, Reduced Order Information Filter (ROIF), Ensemble
Kalman Filter (EnKF), Reduced Order Adaptive Filter (ROAF) (including adjoint), and the 4D-VAR
Representer Method. For the first prototype 1/12 North Atlantic HYCOM system, mostly because of its
simplicity, robustness, and low computational costs, we selected the OI technique with Cooper and
Haines       (1996)     for    downward       projection   of    SSH       from    altimetry   (see
http://hycom.rsmas.miami.edu/ocean_prediction.html for details). The multivariate OI (MVOI) is
expected to replace the Cooper-Haines OI in summer 2004 and the SEEK filter should be in place in
early 2005.

Prototype systems

        i)   Basin-Scale Domains

The present prototype of the near real time HYCOM Atlantic Ocean prediction system
                                                                            o             o
(http://hycom.rsmas.miami.edu/ocean_prediction.html) spans from 28 S to 70 N, including the
Mediterranean Sea. The latest version of this domain is an exact subset of the global HYCOM grid,
                                        o               o
with an Arctic dipole patch above 47 N. Below 47 N, the grid is Mercator with a grid aspect ratio of
one, but above 47 N, the grid aspect ratio can be as high as 1.5. The vertical resolution consists of 26
hybrid layers, with the top layer typically at its minimum thickness of 3 m (i.e., in fixed coordinate mode
to provide near surface values). In coastal waters, there are 15 sigma-levels, and the coastline is at
the 5 m isobath. The northern and southern boundaries are treated as closed, but are outfitted with 3
buffer zones in which temperature, salinity and pressure are linearly relaxed toward their seasonally
varying climatological values. Six-hourly wind and thermal forcing is presently provided by the FNMOC
Navy Operational Global Atmospheric Prediction System (NOGAPS). A similar forecasting system will
                               o       o
be in place for the Pacific (20 S to 65 N) in fall 2004 (same horizontal and vertical resolution).

GODAE IP July 20-22, 2004                        Page 39                                  11/06/10 19:06
A second North Atlantic configuration forms the backbone of the NOAA/NCEP/MMAB North Atlantic
Ocean Forecast System (NAOFS). By comparison to the first system, it will allow us to evaluate the
impact of a) NCEP-based wind and thermal forcing and b) a different grid. By taking advantage of the
general orthogonal curvilinear grid in HYCOM, the MMAB group is using a configuration which, for the
same number of grid points as a regular Mercator projection, has finer resolution in the western and
northern portions of the basin and on shelves (3-7 km) than toward the east and southeast (7-13 km).
The forcing will consist of a) atmospheric fields derived from analyses and predictions of NCEP Global
Model System (T540, about 30 km resolution) and b) eight tidal constituents (body force and boundary
forcing, the latter derived from the GOTT99.2 global tidal model). As in the first North Atlantic
configurations, the northern and southern boundaries are outfitted with 3 buffer zones in which
temperature and salinity are linearly relaxed toward their seasonally varying climatological values. A
similar projection will be used for the MMAB Pacific Ocean configuration later on.

        ii)   Global Domain

For the HYCOM global configuration, we use an Arctic dipole patch matched to a standard Mercator
          ○                             ○
grid at 47 N. Locating the dipoles at 47 N gives good resolution in the Arctic and Hudson Bay (7 km at
mid-latitude vs. 3.5 km at the North Pole). For our target resolution (1/12 at the equator), the array
size is 4500 by 3298 with 26 hybrid layers in the vertical. The complete system includes the LANL
CICE sea-ice model on the same grid.

Assimilation products and dissemination

Near real time and archived results can be viewed on the HYCOM web page
(http://hycom.rsmas.miami.edu). This includes nowcasts and forecasts with many zoom regions for
SSH, surface layer currents, SST, and SSS, subsurface temperature cross-sections, transports, and
nowcast comparisons with unassimilated data. Dissemination is taking place via a Live Access Server
(LAS) (http://hycom.rsmas.miami.edu) for sub-sampled fields and via ftp for the full fields.

Global HYCOM will at least marginally resolve the global coastal ocean [7 km at mid-latitudes, with 15
terrain-following (sigma) coordinates over the shelf], and is therefore an excellent starting point for
even higher resolution coastal ocean prediction systems. To increase the predictability of coastal
regimes, several partners will develop and evaluate boundary conditions for coastal prediction models
using the global and basin HYCOM real-time prediction system outputs. These partners will have real-
time operational coastal nowcast and forecast systems in place during part or all of the FY04-08 time
frame. These systems include nesting of HYCOM in HYCOM, fixed vertical coordinate models in
HYCOM, and unstructured grid/finite elements models in HYCOM. The vertical coordinate of the
various coastal and regional systems will typically be different from HYCOM‘s hybrid coordinate.

Links with GODAE pilot projects (ARGO, GHRSST)

ARGO data will be assimilated by the HYCOM-based system and is presently being used for its
evaluation. The HYCOM Consortium is also a participant in a U.S. GHRSST-associated multi-
institutional project on developing and improving high resolution SST products, a project funded by the
National Ocean Partnership Program (NOPP).

Internal metrics and intercomparison plans

The HYCOM Consortium is already a participant in the MERSEA North Atlantic intercomparison
projects. The Consortium will also be an active participant in the North Pacific and global comparison

Targeted users and external metrics

Application centres are identified through the HYCOM Consortium partners, viz., the U.S. and French
Navies for naval applications, NOAA, universities, and private companies for coastal applications, the
U.S. Coast Guard for marine safety, ….

GODAE IP July 20-22, 2004                      Page 40                                11/06/10 19:06
Reanalysis activities

A 1/12° global HYCOM reanalysis covering 1993-2006 is planned in 2007.

Computer resources

Computer time is currently (2001-2004) provided for our existing Atlantic ocean prediction system
through a DoD High Performance Computing Grand Challenge award. We anticipate receiving another
award project for 2004-2007 associated with the delivery of global HYCOM to NAVOCEANO for
operational testing in 2007.

7.8.9   GMAO
Ocean data assimilation efforts of the NASA Seasonal-to-Interannual Prediction Project (NSIPP) are
now undertaken in the Global Modeling and Assimilation Office (GMAO) at NASA/Goddard Space
Flight Centre. The efforts are directed towards seasonal-to-interannual climate prediction using
coupled ocean-atmosphere-land surface general circulation models. The primary ocean data
assimilation goal is to provide the best possible ocean initialisation for climate prediction. GMAO
activities and progress are documented at http://nsipp.gsfc.nasa.gov/.


GMAO uses the Poseidon quasi-isopycnal ocean model. The model is run quasi-globally (no Arctic
Ocean) from Antarctica to 72˚N with a buffer zone at the northern boundary. The model resolution is
1/3˚ meridionally and 5/8˚ zonally with 27 coordinate surfaces.

Assimilation method

GMAO's assimilation is based both on multivariate optimal interpolation and on an Ensemble Kalman
Filter (EnKF).

Input Data

         Forcing data: Near real time SSMI and Scatterometer-based surface wind analyses;
          relaxation to Reynolds SST, GPCP precipitation.
         In situ data: ARGO data from GODAE server in Monterey; mooring data from PMEL; other
          profile data from NCEP; synthetic salinity profiles from Levitus 94 climatology.
         Altimetry: Near real time data from JPL PODAAC.
         SST: Reynolds SST product.
Assimilation products and dissemination

Assimilation products will be global, monthly averaged estimates of T, S, ocean currents and sea
surface height. Products will be disseminated through the LAS at http://nsipp.gsfc.nasa.gov/las/.
Access will be open for all products distributed through the LAS.

Intercomparison Plans

GMAO plans to participate in the North Pacific Intercomparison Projects. GMAO has participated in the
Equatorial Pacific intercomparison experiments, as a leading member of the Consortium for Ocean
Data Assimilation for Seasonal-to-Interannual prediction (ODASI). Experiments have been undertaken
to evaluate different assimilation systems using the GODAE diagnostics metrics. Using the same input
data streams and quality-controlled observations as other participants, a common set of output data
has been generated for evaluation. Details are documented at http://nsipp.gsfc.nasa.gov/ODASI/.

GODAE IP July 20-22, 2004                     Page 41                                11/06/10 19:06
Targeted users and external metrics

GMAO products are focused towards its S-I forecasts. The external metric is 3, 6 and 12-month
forecasts of Niño-3 SST and Equatorial Pacific thermocline anomalies.

Computer resources

Computer resources are provided through the NASA Centre for Computational Sciences in support of


7.8.10 University of Maryland SODA

The SODA ocean model is built on the POP
(http://www.acl.lanl.gov/climate/models/pop/current_release/UsersGuide.pdf) 1.4 ocean numerics with
free surface and a 0.25x0.4 degree horizontal grid with displaced north pole and 40 vertical levels.
Mixing schemes include KPP and nonlinear horizontal mixing. We expect to transition to the recently
released POP2 code to exploit several improvements including partial filled bottom cells and NetCDF
output tools. The new POP2.0 formulation does allow anisotropic horizontal viscosity, which has
certain advantages as well (Large et al., 2001). River inflow is based on climatological data from the
Global Rivers Data Center (GRDC). Bottom topography has been obtained from the 1/30 degree
GTOPO30 with modifications for certain passages provided by Julie McLean. Under the Arctic ice
surface salinity is relaxed to the monthly Polar science center Hydrographic Climatology 2.1 (Steele et
al., 2001) in order to account for seasonal melting/freezing. Monthly polar sea ice coverage is based
on satellite estimates of ice concentrations (Parkinson, 2001) for the period 1979-pres edited by John
Weatherly of ERDC-CRREL.

Assimilation method

The current approach is based on a computationally efficient multivariate sequential estimation, a
development of the approach described in Carton et al. (2000a,b). This approach has been
augmented to allow for an empirically-based bias-correction model (Chepurin et al., 2004). Analysis of
forecast minus observation statistics are used to specify the (bias-corrected) forecast error
covariances in a reasonable. This preliminary analysis allows us to introduce a number of features of
the observed ocean including high vertical correlations within the mixed layer, geostrophic
relationships, latitudinal and flow-dependence, and rather steady relationships between temperature
and salinity forecast errors.

Input Data

Forcing data: two reanalyses have been produced to date, SODA1.0, 1948-present, based on
NCEP/NCAR daily reanalysis winds and SODA1.2 based on ECMWF ERA40 winds (1958-2001). The
disadvantage of the NCEP/NCAR wind product is the presence of bias, particularly the weakness of
the equatorial winds. We address bias in the stress by adding a steady spatially-dependent term to
each stress component to correct for time-mean bias and multiplying by a steady spatially-dependent
term to properly scale the stress variance. Both of these terms are determined by comparison with
Quickscat scatterometer wind stress during the years of overlap. Surface heat flux boundary
conditions are provided by a bulk formula for heat flux (which is reduced under polar ice), while
precipitation is provided by the Global Precipitation Climatology Project.

The basic subsurface temperature and salinity data sets consist of approximately 7  10 6 profiles, of
which two-thirds have been obtained from the World Ocean Database 2001 (Boyer et al., 2002) and
are extended by operational temperature profile observations from the National Oceanographic Data
Center\NOAA temperature archive, including observations from the TAO/Triton mooring thermistor

GODAE IP July 20-22, 2004                      Page 42                                  11/06/10 19:06
array and ARGO drifters. The profile data is concentrated along commercial shipping lanes. Mixed
layer temperature observations are available from the COADS surface marine observation set (Diaz et
al., 2002). Satellite altimeter sea level from GEOSAT, ERS/1-2, TOPEX/POSEIDON and JASON is
used beginning in the mid-1980s. Data checking for this analysis includes checks for duplicate reports
and errors in the recorded position and time of observations, for static stability, for deviation from
climatology, and checks on the relationship between temperature and salinity. Substantial quality
control is already in the WOD2001. Our additional quality control (including buddy-checking,
examination of forecast-minus-observation differences, and vertical stability) eliminates roughly 5% of
the profiles.

Assimilation Products and Dissemination
                                                  o    o
Reanalysis fields are mapped onto a uniform 0.5 x0.5 x5-dyx40-level in NetCDF format. They will be
distributed through our OpenDAP (formerly DODS) server dods.atmos.umd.edu and shortly through
our LAS server. We are also distributing a revised version of our original reanalysis (beta23) through
the same mechanisms.

Internal Metrics and Intercomparison plans

We are conducting an ongoing series of comparisons with independent observations (for example,
station time series and moored velocity) as well as comparisons with alternative reanalysis products.
We also collect a complete set of observation-minus-forecast statistics to monitor bias in the forecast.
From time to time these comparisons will be made available through our website,

Targeted Users and envisioned external metrics

SODA reanalysis is directed toward the climate research community although it has proven to be of
interest to other communities as well such as biological and chemical oceanographers.

Links with application centres or service providers

Aspects of this research have been funded by the National Aeronautics and Space Administration, the
National Oceanic and Atmospheric Administration, and most notably, by the National Science

Prototype Systems
An original prototype system was developed in the mid-1990s (Carton et al., 2000a). The 23 revision
of this original reanalysis (beta23) as recently been released.


Boyer, T.P., C. Stephens, J.I. Antonov, M.E. Conkright, , L.A. Locarnini , T.D. O'Brien, H.E. Garcia,
        2002: World Ocean Atlas 2001, Volume 2: Salinity. S. Levitus, Ed., NOAA Atlas NESDIS 49,
        U.S. Government Printing Office, Wash. D.C., 165 pp..

Carton, J.A., G. Chepurin, X. Cao, and B.S. Giese, 2000a: A Simple Ocean Data Assimilation analysis
        of the global upper ocean 1950-1995, Part 1: methodology, J. Phys. Oceanogr., 30, 294-309.

Carton, J.A., G. Chepurin, and X. Cao, 2000b: A Simple Ocean Data Assimilation analysis of the
        global upper ocean 1950-1995 Part 2: results, J. Phys. Oceanogr., 30, 311-326.

Chepurin, G.A., J.A. Carton, and D. Dee, 2004: Forecast model bias correction in ocean data
       assimilation, Mon. Wea. Rev., submitted.

Large, W.G., G. Danabasoglu, JC McWilliams, PR Gent, and FO Bryan, Equatorial circulation of a
        global ocean climate model with anisotropic horizontal viscosity, J. Phys. Oceanogr., 31, 518-

GODAE IP July 20-22, 2004                       Page 43                                11/06/10 19:06
Parkinson, C.L., 2002: Trends in the length of the southern Ocean sea-ice season, 1979-99, Annals of
       Glaciology 34, 435-440.

Steele, M., R. Morley, and W. Ermold, PHC, 2001: A globalocean hydrography with a high quality
        Arctic Ocean, J. Climate, 14, 2079-2087.

7.9       China
China is developing an operational ENSO forecasting system based on CGCM jointly by State
Oceanic Administration of China (SOA) and Institute of Atmospheric Physics (IAP)/Chinese Academy
of Sciences. The system is now in operational at IAP as a research project. The system will be in full
operational in 2006 at the National Marine Environmental Forecasting Center of SOA. In the following
description we refer this as ENSO.

The data assimilation system is a 3DVAR-based scheme (recently named OVALS, ocean variational
analysis system) which is a bivariate (T and S) , can assimilate XBT, ARGO, TAO and altimetry data,
and use nonlinear T-S diagram (derived from WOA climatology) as constraints . For background
error, vertical correlations of T and S fields are considered , and the horizontal correlation length
scales are calculated from model results .

To enhance the China‘s capability for operational forecast of its adjunct waters, IAP is developing a
                                                                      o         o         o       o
comprehensive ocean data assimilation system for China Waters(99 E to 143 E and 0 N to 51 N).
So far a coarser resolution (0.5 degree in both directions) temperature assimilation run from 1982-
1993 has been finished . The assimilation scheme is OVALS (without altimetry assimilation code). In
the following description we refer this as CODAC. In next two years we plan to increase resolution to
1/8 degree with HYCOM model (working with Nansen Center, Norway) and assimilate altimetry and
ARGO data.

[1] Han, G., J. Zhu and G. Zhou, 2004: Salinity estimation using T-S relation in the context of
variational data assimilation. J. Geophys. Res., 109. doi:10.1029/2003JC001781.
[2] C. X. Yan, J. Zhu and G. Q. Zhou, 2004: The roles of vertical correlation of the
background covariance and T-S relation in estimation temperature and salinity profiles from
surface dynamic height. J. Geophys. Res. 109. doi:10.1029/2003JC002224.
[3]G. Zhou, W.Fu, and J. Zhu, 2004: The impact of location dependent correlation length
scales of background covariance on a ocean data assimilation system. Geophy.Res.Letters. In
[4]You X, G. Zhou, J. Zhu, R. Li and Q. Zeng, 2003: Sea Temperature Data Assimilation System for
the China Sea and Adjacent Areas, Chinese Science Bulletin , 48 Supp.II 70-76.

Input Data
          ENSO: Reynolds SST, Levitus climatology, COADS surface fluxes, Argo data, T/P and Jason-1..
          CODAC: Reynolds SST, Levitus climatology, GTS XBT.
          ENSO: IAP CGCM
          CODAC: IAP Nested regional OGCM.
Assimilation products and dissemination
          ENSO: 1998-2003 T and S assimilated fields. The assimilated results are compared to
           withholding TAO T and S observation.
          CODAC: assimilated 1982-1993 T and velocity fields.

GODAE IP July 20-22, 2004                          Page 44                                 11/06/10 19:06
Links with GODAE pilot projects (Argo, GHRSST)

All two projects assimilated and plan to assimilate Argo profile data.

Internal metrics and intercomparison plans

All the two projects are willing to joint any suitable intercomparison plans.

7.10 Issues
We list here several issues that GODAE needs to address. Some of these are developed further in
section 11 of this implementation plan.

         Model improvements for ocean state estimation (see Section 11.2.4).
         Data assimilation improvements (see Section 11.2.5).
         Interconnection and interoperability of centres and intercomparison of products
         Routine operation
         Transition to operational systems
         Climate-GODAE interface: how are we joining/cooperating with seasonal and other
          climate prediction systems?
         Computing resources – is there a common requirement, e.g dedicated versus shared
         How do we address the uneven global effort which is skewed toward the Northern
          Hemisphere? Is it an issue from the user perspective or simply responding to reality?
         How do we extend capacity to developing countries and get broader participation?
         Should we be developing guidelines for reanalysis – not prescriptive but to ensure even
          and systematic evaluation and best possible inputs/data set preparation?
         Approach to integrating regional/local with GODAE (reference Section 9) – are we
          properly planning for the infrastructure needed to routinely support such applications?
         Maintaining a balance between regional and global approaches – this needs to be re-
         Need to more clearly articulate the measures of skill (Class 4 metrics).
         Need to close the feedback loop from GODAE models and applications to the observing
          system, including through OSSEs and simple metrics.

GODAE IP July 20-22, 2004                        Page 45                              11/06/10 19:06
8 Standard products and
  product serving

Product serving is a critical issue for GODAE.
Product servers are needed for:

            Intercomparing assimilation
             products between the different
             modelling / assimilation centres.
            Promoting GODAE for potential
             applications (including research).
            Serving application centres, service
             providers and/or end-users in an efficient way.
These functions have different requirements in terms of type of products and availability, products
quality and timeliness. Product intercomparison will require large sets of products for agreed variables
and (minimal) spatio-temporal sampling that have been generated by several assimilation centres to
be accessible to tools which can make simple, and perhaps in time sophisticated, intercomparisons.

Only a subset of the assimilation products (i.e. the 4D description of the ocean state including
analyses and forecasts) can be archived and compared. The main archive for some centres will not be
directly accessible by other partners and somewhat more limited subsets will be accessible to the
GODAE. This facility will also be useful for the research community, for some applications (e.g. those
requiring climatological statistics) and for promoting GODAE. Near real time applications are likely to
be served through additional dedicated or tailored services (e.g., extracting boundary conditions for
coastal/regional models) because of the timeliness constraint.

Serving real-time ocean monitoring system products to meet GODAE goals required:

            definition of metrics to be systematically used for assessing the quality and consistency
             of ocean monitoring system outputs (see section 10.5.6).
            product coherency and standardisation at many levels to facilitate the exchange and
             joint use of multi-disciplinary data (e.g. regional data sets and product organisation, times
             series concept and product classification, parameters definition, metadata with revisited
             convention, NetCDF format, model documentation)
            to rely on a decentralised but compatible system architecture from production to
             distribution, open on Internet, since data are managed independently at participating
             model sites, and information are collected at one level and shared at all levels,
            to develop ocean portal(s) to ease product selection, manipulation and
             visualisation (ie with 1- data discovery function or the ability to search for and find
             datasets of interest, 2- data transport function or the ability to access the data in an
             interoperable manner from client applications, 3- on-line browse function or the ability to
             evaluate the character of the data through common web browsers, 4- a user feedback
             function or the ability to trace the user and have feedback on the system

8.1       Product coherency and standardisation
Products from different assimilation systems should be harmonised and standardised, in particular, to
facilitate intercomparison experiments. The following requirements should apply:
      •    Model documentation: Each center listed and documented similarly their operational system
           and the distributed data (model overview, history file).
      •    Convention and format issues:     The data should be portable, self-describing.
           Conventions should be applied for metadata (information on data). They have been

GODAE IP July 20-22, 2004                         Page 46                                  11/06/10 19:06
           developed for metadata readable by human and program. However, a higher degree of
           product coherency / harmonisation may be required to reduce the complexities of setting
           up the whole integrated system and to run intercomparison exercises.

There are today several parallel but as yet unconnected activities relating to developing standards for
data and product exchange, many involving definition of syntactic metadata for NetCDF and like file
formats. These include the so-called "CF" standard developed for exchange of climate model results,
the standard being used for Argo data exchange, and standards developed for the WOCE data set
(conformant with COARDS definitions). Within the meteorological community there have been several
meetings to discuss metadata standards, mostly within the context of using XML-based tools (see
http://www.wmo.ch/web/www/reports.html#WDM). There have also been several discussions, mostly
at the national level, in relation to the development of interoperability and standards to support specific
components of the observing system and model product exchange. There is not, however, a
systematic process for creating these standards. The "standards" are emerging from small groups of
interested individuals (bottom-up), who provide uncertain levels of broad public review and no
community-wide way to advertise that the standards-writing is even going on. It remains unclear
whether the ponderous level of formality typical of ANSI/ISO computer standards is well suited to the
needs of the ocean data community. But perhaps other more modern standards-generating activities
in the IETF, WC3, etc. might serve as a model. Might we be able to develop a standards process for
the ocean/atmosphere community that could help with this predicament ?

8.2       From production to distribution
The requirements for the data distribution (data discovery and transport) are as follows:
          To distribute / transport data via Internet (push/pull integration)(with minimum constraint)

          To manage large data sets, numerous data flows and numerous users
          To allow synctatic and semantic operability (consistent format representation across data sets,
           consistent semantic representations of the data)

          To preserve data and products

          To rely on open source technology associated to reliable, sustainable, efficient operations

The underlying philosophy is that data should be managed independently at participating sites. The
volumes of data that can actually be exchanged are relatively small. Subsampling can be performed
by selecting:

            Class of products and subsets of the model variables
            Limited area regions or reduced spatial and vertical resolution or particular sections
            Periods (and reference dates) for time sampling for instantaneous or time-mean fields and
            Particular forecast ranges (e.g. best estimate time series, 6 -day forecast time series)

Several combinations of sub-sampling should be used to enable different aspects of the analyses to
be intercompared. Depending on the state of development of the tools that will be available, it may be
necessary to reinterpolate some of the data onto common grids.

Hankin et al. (2002) describe a number of data and product serving concepts and capabilities which
have been developed over the last decade. The key concepts which are being employed are
described in the following points.

1) A very general data model or ‗format‘ is used to associate metadata with data. The data model is
   based upon fundamental computer science data structures and mechanisms. Two of the data
   models and their associated software are the Open Source Project for a Network Data Access

GODAE IP July 20-22, 2004                          Page 47                                   11/06/10 19:06
    Protocol (OPeNDAP, formerly known as DODS, cf. http://www.opendap.org) and the Abstract
    Data Delivery Environment (ADDE) protocol. Data from a wide range of formats can be read into
    the general formats. The middleware codes which do this are known as servers. For example
    HDF-EOS, NetCDF, Matlab and GRIB can be read into the OPeNDAP format and a server which
    can read BUFR data into it is being developed. Most applications which can read NetCDF files can
    be quickly adapted to use the OPeNDAP format and some other applications such as Matlab and
    IDL have been individually adapted to use it.

2) The volumes of data transferred across the internet can often be reduced (by several orders of
   magnitude) by performing processing at the site where the data are stored. The simplest example
   of this is to request data for a particular variable for a given area or section. Additional processing
   at the data server, such as re-gridding and aggregation of data from a large number of data sets
   into a single product, can reduce some data transfers by further orders of magnitude. This idea
   has been demonstrated by the GrADS-DODS server (GDS) which combines the technologies of
   OPeNDAP and the Grid Analysis and Display System (GrADS). Users are able to embed a GrADS
   analysis expression in their request specifying the remote data set. The requested analysis is
   performed at the server producing an intermediate output data set. Subsets of that data set and
   aggregation in time can then be retrieved using the OPeNDAP protocols

3) Data push and data pull mechanisms are required to meet the needs of users within the capacity
   of the internet system. The Internet Data Distribution (IDD) system is an example of a
   modern ‗data push‘ system. It delivers up to 10GB of data per day to over 130 institutions by
   relaying data from site to site using hierarchical distribution trees. The data request functions use
   the http protocol, sending an enhanced URL to the server. Opendap also allows server-side
   function evaluation and advanced computational issues, as well as user traceability and user
   feedback (via logs analysis) for system improvements and end-users knowledge.

4) Additional tools are needed to help users to find the most suitable products. The National Virtual
   Ocean Data System is being developed as a single portal for many distributed OPeNDAP
   servers. Catalog/directory systems for finding data are also being developed by the Thematic Real
   time Environmental Data Distributed Services (THREDDS) project. A concensus on metadata
   standards for these catalogues has not yet emerged.

The ultimate goal is to allow end users, whoever they may be, to access immediately whatever data
they require in a form they can use, all while using applications they already possess and are familiar

Opendap has been serving the marine community since 1995 and serves already a broad range of
data, including oceanographic, atmospheric, and even astronomical data (cf. the Opendap master data
set list, http://www.opendap.org/data/ ).

Depending upon the nature of the data to be provided, it is recommended that data providers of:
       Gridded data, have their data available on Internet through Opendap data access protocol
        (Opendap is an operational component of IOOS for access to gridded data,
         Opendap servers are available for download without licensing costs)
       Complex data collections in a relational data base (SQL), have their data available on
        Internet through an Opendap relational data base server. (Opendap is a pilot component of
        IOOS for access to unstructured data collections). (Full operational support for relational data
        bases will be developed early in the evolution of IOOS DMAC, cf. http://dmac.ocean.us .)
       Geographic Information System (GIS) data collections, use gateways that provide
        translation from the GIS network protocols to the Opendap protocol. Gateways will be
        developed by IOOS.
       Large collections of files that comprise a single (logical) data set, have their data available
        on Internet through an Opendap ―aggregation server‖ and participate in a DMAC data
        aggregation pilot project. The main purpose of this distribution server is to read NetCDF data
        files and allow the "aggregation" of multiple datasets into one (virtual) dataset: a time series
        data set accessible by a unique URL to the users with or without sub-setting characteristics
        (like geographical extraction, one variable).

GODAE IP July 20-22, 2004                       Page 48                                  11/06/10 19:06
Once the distribution tool has been selected, it remains configuration issues to homogeneously
present product data sets to the end-users and operational issues to have it run operationally with data
release awareness.

8.3       Product selection, manipulation and visualisation
The requirements for data manipulation and visualisation were:
          To allow selection (and download) of chosen fields from each individual system and plot them
           in a very simple and convenient way.
          To unify the access to the model outputs (cf. file naming conventions, how long to keep files
           on line, whether variables are stored as individual files or consolidated).

          To manage large data sets, numerous data flows and numerous users.

          To rely on open source technology associated to reliable, sustainable, efficient operations.

Above the distribution technology, they are direct client application tools which allowed us to
manipulate and visualise the data distributed on Internet. We noted NCBrowse, Excel, Matlab, IDL,
Ferret, GRADS, IDV, Las, Map Server as Opendap client applications. We should finally identify a list
of web services which exists and could be helpful to use for the GODAE intercomparison experiment.

            The Live Access Server (LAS) is emerging with a central role in GODAE
             (http://ferret.wrc.noaa.gov/Ferret/LAS/ferret_LAS.html). LAS uses the OPeNDAP protocol
             and servers. Within the US, considerable effort has already been devoted to using LAS to
             enable the different contributions to GODAE to serve data and products in a similar way
             and to provide matching graphical products for intercomparisons. This effort was recently
             used by European ocean monitoring systems within Mersea project and should be
             extended to all international partners.
            We noticed also the Unesco Bilko commissioned educational image processing software
             package (cf. http://www.bilko.org) which has been developed to support training in coastal
             and marine remote sensing through a series of computer-based learning (CBL) modules.

With a Live Access Server (LAS), you can discover, browse and access data products.

LAS is an interoperable mapping tool. It federates and unifies the access to the model outputs
regardless of differences of individual Opendap servers. LAS is a configurable, interactive and inter-
operable scientific data "product" server, a on-request, on-the-fly mapping tool. A user can quickly
obtain products such as plots, images, and formatted files, generated on-the-fly, from custom
subsets of variables – any t, z, y, x combination – (with possibility there to customise the plot).
The description here is taken from Hankin et al. (2002). Figure 15 illustrates how LAS functions as a
―traffic cop‖, directing data requests. Clients submit requests for products to LAS through a standard
protocol. The protocol is defined in the eXtensible Markup Language (XML) (LAS on-line, 2002). LAS
dispatches those requests to back end applications, which produce the requested products in the form
of graphics (such as lat-long plots, vertical sections or Hovmuller diagrams) or data sets. The design of
the user interface (UI) is decoupled from the server, so that distinct UIs can be designed that are
friendly to users of differing scientific sophistication. One particular LAS UI assists high school
students learning about relationships between climate variables; another so-called ―batch UI‖ (in which
other computers are ―users‖) allows GODAE operational models to request the latest data available for
assimilation automatically via a Unix script.

GODAE IP July 20-22, 2004                         Page 49                                  11/06/10 19:06
                          Figure 1. "Traffic cop" design of the Live Access Server

The most advanced capabilities of LAS address the needs of collaborative model development. LAS
offers the ability to compare gridded fields as graphical overlays or computed difference plots (which
may require automated regridding). Modeling sites may choose to configure their LAS servers as
―sisters‖, such that the collective model holdings appear as a unified virtual data base of
intercomparable model runs. An example of a sister server network can be seen at the US GODAE
modelers‘ site (http://ferret.wrc.noaa.gov/godae/) on which outputs from U.Md., JPL, MIT, SIO, and
GFDL may be intercompared and compared to reference fields. Current development efforts on LAS
are addressing i) enhanced support for scattered data (in-situ measurements) and model-data
comparison; ii) transformations (e.g. averaging, extrema); and iii) standardized model diagnostics.

The Ferret application as the default back-end supplied with LAS but Matlab, GrADS, IDL and custom
systems are all in use or have been prototyped. The ―operations‖ which can be performed on each
variable within LAS depend on the individual installing the system who has to configure each variable


GODAE IP July 20-22, 2004                        Page 50                              11/06/10 19:06
9 Application Areas

The section describes existing and/or
potential application areas for GODAE
products. In each case a brief
description of the applications is given
with reference to the character of
GODAE products for the application
areas and the nature of the services
provided by these centres/application
areas. End-users are identified as well
as metrics envisioned for assessing the
impact of GODAE products.

We miss the regional/coastal part.

9.1   Ocean mesoscale and surface fields and short-range forecasts
Most applications for state estimates at mesoscale resolution are regional, though the number of these
applications is so large and their spatial distribution so varied and unanticipated that the generation of
global products at the mesoscale will be one of the important, demonstrable outcomes of GODAE.
The high frequency nature of the product places a premium on product delivery and automated push
or pull systems. The high resolution and regional nature of the applications mean that for many
applications provision of tools for users to select data for specific regions would be very valuable. For
some applications (e.g. Navies, marine safety, coastal forecasts) there is a demand for short-range
forecasts as well as ocean state estimates; for others (e.g., oil companies) accurate statistics of mean
state, variability and extreme states are also of value. Some applications (e.g., coastal/littoral zone
forecasts) require a close connection between the GODAE assimilation centre and the application
centre because of the importance of the details of product generation (e.g., consistency in bathymetry,
forcing, boundary configurations). GODAE products at the mesoscale will also have research
applications for both physical and biogeochemical oceanography.                Although some research
applications will be interested in merely the statistics of the mesoscale, interest in fields at the highest
spatial and temporal resolution will provide challenges in terms of product archive and access or

This sub-Section provides a general description of the applications. Section 9.2 provides a more
detailed discussion for coastal applications.

Characteristics of products delivered to application centres

         Typically, the upper ocean matters more to the users than the deep, consistent with time
          scales (though ocean current variations to 1000 m depth are important for oil industries)
         Emphasis is on upper ocean physics as well, particularly the mixed layer
         Some applications (e.g., ship routing, search & rescue) are interested only in surface
          currents, or sea state rather than temperature.
         The forecasts are initiated from the ocean state estimate, usually without any coupling,
          although there is interest in coupled forecasts for some applications, e.g., hurricane
          forecasting (and also coastal forecasting)
         Typical lead times for a forecast application are 0-20 days, with highest premium on the
          shorter range.
         In some cases, there is direct application of the forecast; in others there is middle-person
          interpretation (analogous to global and regional weather prediction)
         Applications are run in real time
         Demand is for routine (regular) products with automatic push or pull of products

GODAE IP July 20-22, 2004                        Page 51                                   11/06/10 19:06
         The Assimilation Centre and Application Centre are sometimes the same organization
         Research users will not always have a timeliness constraint, but will often desire access to
          several products – the provision of web-based tools for ease of product extraction and
          intercomparisons will be essential for engaging a diverse set of researchers and gaining
          their feedback.
Internal metrics:

         Model and assimilation intercomparison metrics as per the North Atlantic Intercomparison
          Project (Section 11.4.1).
         Statistics of model minus observation errors for observations generated by the application
          centre or end users (e.g. current measurements by oil companies)
         Ocean hindcasts forced with observed (c.f. forecast) winds, historical skill scores
Application metrics:

         Impact on skill of forecasts by application centres
         Impact on value of forecasts by application centres to end-users
         Improvement in the quality of real-time data stream through feedbacks to those collecting
          observations and to data assembly centres
         Feedback from the research community in terms of product reliability, consistency, etc.
         Other metrics to be established in concert with application centres or directly with the end
          users. Establishment of these metrics or feedback on product utility will be an activity as
          part of developing the GODAE Common (see Sections 10 and 11).
Examples of application centres:

         Public good services
         Navies require real-time products. Some will generate their own products, for example
                       FNMOC -- http://www.fnmoc.navy.mil/
                       NAVOCEANO -- http://www.navo.navy.mil
         Some Navies will utilize products from other centres. Examples are:
                       The Royal Navy will use products from the Met Office (http://www/metoffice.com)
                       SHOM will use products from Mercator (http://www.mercator.com.fr)
         Private enterprise companies such as Fugro-Geos serving industry (e.g. oil companies,
          design, etc.)
         Software companies providing tools for use by operators at sea (e.g. BMTMIS tools for
          search & rescue, and oil and chemical spills)
         Marine Prediction Centres (e.g., NCEP‘s Marine Prediction Centre,
          http://www.mpc.ncep.noaa.gov; Meteo-France,) providing warnings and outlooks.
         Companies providing services for ship-routing, yachting
         Agencies with responsibilities for monitoring marine pollution (e.g. IFREMER, CEFAS) and
          fisheries (ICES, CEFAS)
         Coastal Forecast centres (see Section 9.2).
         Centres with responsibilities for monitoring of sea-ice (e.g., DNMI)
         Regional Forecast centres (e.g., JMA, NAVOCEANO) JMA will provide model variables
          forecasted by a real time operational assimilation-prediction system (COMPASS-K)
          through JMA product server. The forecasting period is not decided yet. MRI will continue
          prediction experiments and contribute to the operation. MRI also started to develop a
          nested Kuroshio forecasting system with a new version of MRI-OGCM community model
          (MRI.COM). Issues of nesting of models are almost similar to the above coastal forecast

GODAE IP July 20-22, 2004                         Page 52                                11/06/10 19:06
          NWP groups providing surface wave and atmospheric forecasts
Examples of end users:

          Coastal and marine ecosystem monitoring, marine resource planners, Coastguard,
           fisheries, insurance industries, oil companies, deep cable laying, recreation industries,
           research (physical and biogeochemical oceanography) communities.

9.2     Coastal applications
Coastal applications will use GODAE ocean state estimates as boundary conditions for coastal/ littoral
zone hindcasts and forecasts and analyses. It is not yet clear what the accuracy requirements are.
Development of GODAE products for these applications will represent a significant research effort
between centres. Issues of nesting of models of different resolution, importance of regional wave
models, consistency in bathymetry, forcing, boundary configurations, and input to ecosystem models
are critical elements for collaboration. The issue of two-way coupling wherein GODAE products may
benefit from input of higher resolution and better representation of coastal processes is also an area of
potential research collaboration. The end users will include regional/local governments responsible for
coastal management, as well as coastal industries such as fishing, recreation.

9.2.1    General strategy for coupling to coastal/regional subsystems
i)       GODAE approach provides efficiency because the systems can provide information/BCs to
         multiple users, in a variety of ways

ii)      In some prototypes the regional/local modelling (high-resolution in coastal areas, specific
         regions) is in-built to the modelling systrem

iii)     In other cases the coastal modelling is part of the same project so the interface issues are
         being solved as part of the project
iv)      In other cases, BCs/information is being provided to 3 parties who may have knowledge of
         the source model (and vice versa) but otherwise are running completely independent

9.2.2    Interface to the GOOS Coastal Ocean Obs Panel
       From Keith Thompson outline:
        The modelling chapter [of the COOS IP] has 3 parts:
           (i) guidlines for setting up a modelling and analysis for a new GOOS Regional Alliance
           (ii) specific action items for the follow-on COOP (may be called POCO) including
           specifications of metrics, inventories of models and non-COOP data streams, setting up
           Community Modelling Networks etc, all aimed at linking the regional modelling activities into
           something approaching a global system, and finally
           (iii) possible pilot projects that will enhance coastal modelling at the global scale (e.g.
           satellite products for the coastal ocean, global surge model).
        o     Like to coordinate the COOP implementation plan with the GODAE plan.
        o      Given our (the COOP) very tight deadlines, there is a limit on how specific they can be in
            their implementation plan (Thompson is aiming for 5-10 pages for the modelling chapter).
        o      Thompson planned to outline a possible pilot project under (iii) above on the downscaling
            of deep ocean models to coastal seas. He will make statements about the need to include
            contrasting regions (e.g. narrow vs wide shelves - North Pacific vs European shelf possibly),
            the need for active shelf observation and modelling programs upon which to build, and
            possibly the establishment of a joint COOP/OOPC-GODAE working group that could look at
            limits to predicatibity (i.e. how much do the shelf modellers really gain in forecast skill by
            including offshore BC's from the present generation of deep ocean models, and what can
            we expect over the next 5 years etc).
        o     The Med work will also be mentioned, and the recent MERSEA study of coupled physical
            and biogeochemical modelling, as the sort of projects that could develop into new COOP

GODAE IP July 20-22, 2004                        Page 53                                  11/06/10 19:06
               pilot projects aimed at increasing skill of models for nonphysical variables in the short term,
               and working towards community models that can eventually deal with ecosystems in the
               longer term.

9.2.3    Global/basin-scale model interface to multiple local/regional associations
         i)          GODAE can be seen as providing one part of the infrastructure supporting a vibrant
                     regional/coastal modelling program.

         ii)         In part this may come from organised approaches to data assembly, rather than
                     having this focussed locally – GODAE will contribute to a data service to the
                     community, from in situ and satellite networks, with local enhancements

         iii)        In part this will come from GODAE modelling products which will be to a common
                     agreed standard: there will be metrics that can be accessed by the user community to
                     see whether model X has the attributes they desire. There will be consistency in the
                     way products are provided (standards).

         iv)         In part it will come from cooperation on the development of a GODAE-coastal
                     modelling interface which, initially might take the form of data servers that can
                     efficiently provide needed subsets of data, and ultimately might take the form of
                     agreed protocols for exchange of data between models (the PRISM approach).

         v)          Joint working group to study predictability as above

9.2.4    Case studies
The idea here is to demonstrate in the IP how this is being implemented now. It is not just an idea.

For example (also see Keith Thompson notes):

          NRL nested model approach

          HYCOM and/or Mersea project(s) with multiple modelling subsystem partners

          MFSTEP?

          BLUElink and its ROAM relocatable model

          FOAM and POLCOMS

          [Gulf of Maine, west coast N America – from IGST VIII]

          Try to get global coverage (of coastal subsystems).

9.3     Seasonal-to-interannual climate
The most common application for the GODAE ocean state estimate is an initial condition for a coupled
model forecast (seasonal-to-interannual prediction: SIP). One of the primary issues to be faced by this
community is how best to use the state estimate, i.e., because of the initialisation shock and climate
drift of coupled models, the best initialisation state may not be the best state estimate. There is a
demand from the research community for state estimates to help with descriptions/understanding of
the ocean variability at these times scales and for assessing prediction skill. There is also some
demand for ocean monitoring, including information relevant to fishing industries on shifts in the warm
pool boundary, coastal upwelling, etc.

          Characteristics of products delivered to application centres
          Primary emphasis is on the upper ocean content in the tropical oceans
          The role of salinity and currents in prediction skill is yet uncertain, so estimates of these
           variables are needed

GODAE IP July 20-22, 2004                            Page 54                                   11/06/10 19:06
         Description of surface currents (convergences) useful for some applications
         SST information at higher frequency and better resolution than from Reynolds may be
          important for monsoon prediction. In addition, there is emerging interest in extratropical
          SST prediction at subseasonal timescales.
         Time and space scales (minimal) are probably around 3 days and 1000 x 200 km
         Accuracy requirements are uncertain (because of coupled model drift), however, there is
          some indication the accuracies of 0.1 C for SST, and 4cm for sea level are needed.
          Uncertainty estimates for the product are needed (either from or for multiple realizations of
          the ocean estimates) for use in ensemble forecasts,
         Applications are run in near-real time (a few days to several weeks)
         Demand is for routine (regular) products
         Often the Assimilation Centre and Application Centre are the same organization
         Internal metrics:
         Model and assimilation intercomparison metrics as per Equatorial Pacific Intercomparison
          Project (Section 9.10)
         Forced ocean hindcasts
         Application metrics:
         Impact on skill of forecast of phase, amplitude and structure of tropical SST anomalies as
          a function of lead-time.
         Provision of products to contribute to assessment of observing system requirements for
         Examples of application centres:
         Public good services, such as Climate Services (e.g., NOAA‘s Climate Prediction Centre,
          http://www.cpc.ncep.noaa.gov; Australian Bureau of Meteorology‘s climate centre,
          http://bom.gov.au/climate/ahead/). The requirement is probably monthly for seasonal
          outlook. Other centres for research products would not have a timeliness requirement.
         SIP Centres require near real-time products. Examples of centres that will generate their
          own products are
         ECMWF -- http://www.ecmwf.int/products/forecasts/seasonal/
         NCEP -- http://www.cpc.ncep.noaa.gov/
            IRI – http://iri.ldeo.columbia.edu/
         JMA -- http://ddb.kishou.go.jp/climate/ElNino/ddbelmon2.html#outlook
         BMRC/CMR -- http://www.bom.gov.au/bmrc/ocean/staff/fzt/CM/page1.html
         Hadley Centre – http://www.met-office.gov.uk/research/hadleycentre/index.html
         NSIPP – http://nsipp.gsfc.nasa.gov/ /exptlpreds/exptl_preds_main.html
         GFDL – http://www.gfdl.noaa.gov/experimental.html
         SIP Centres that will utilize products from other centres. Examples are:
         COLA -- http://www.iges.org/ellfb/Mar02/Huang/huang.htm
         CDC -- http://www.cdc.noaa.gov/
         Examples of end users: agriculture, energy, and water resource planners, fisheries,
          insurance industries, research (physical and biogeochemical oceanography, atmospheric)

9.4   Medium to long-term climate
Although currently there is no medium (decadal) to long-term climate prediction activities, research
strategies are starting to emerge to start to address this requirement. The operational imperative is
the evidence that the impact of the PDO, AO, etc on seasonal predictability of precipitation and surface
temperature over continents is just as important as ENSO, and of course that the thermohaline

GODAE IP July 20-22, 2004                          Page 55                             11/06/10 19:06
circulation is central to global change scenarios. Nevertheless, most applications for GODAE products
at this timescale are in the realm of research for understanding the long-period variability of the ocean
and for defining the requirements and strategy for an ocean observing system for climate monitoring.
Questions about climate changes and underlying mechanisms are subtle and require dynamically
consistent, high quality state estimates, including the time-varying surface fluxes.

         Characteristics of products delivered to application centres
         High emphasis on quality and coherence (self-consistent estimates of u, T, S) over the full
          ocean depth
         Error estimates are important
         Accuracy requirements are not yet precisely defined but are significant because of the
          subtle climate change signals.
         Sea-ice is an important element of the climate state
         Multi-decade historical re-analyses are in demand by researchers
         Often a product of 4D VAR because of premium placed on long period, large-scale modes
         No timeliness constraints at present, however routine products of the state of the ocean
          climate each season are of interest,
         Products include ocean transports, ocean ventilation (water mass formation), global and
          regional budgets
         Internal metrics: model and assimilation intercomparison metrics from the GODAE
          projects described in Section 9.
         External metrics: not yet established, though an important metric will be feedback from
          the research community concerning product reliability.
         Most application centres will be separate from the assimilation centres. For example,
          Global change or climate prediction groups such as GFDL, NCAR, GISS, Hadley Centre,
          MPI, FRSGC, would look to products from assimilation groups such as ECCO
          (http://www.ecco-group.org for ocean initial fields because of the high computational
          burden associated with ocean state estimates with subtle accuracy requirements.
         Examples of application centres:
         ECCO – will use its own products for research
         Some central Climate Services Group will need to be established to support analyses for
          CLIVAR science, and climate assessment. However, such facilities do not yet exist.
         Climate assessment
         Examples of end users: CLIVAR science community, biogeochemical research
          community, global change scenario groups and policy makers; Climate centres that
          provide outlooks to other users

GODAE IP July 20-22, 2004                       Page 56                                 11/06/10 19:06
10 Users and Benefits

10.1 Broad approach
Ultimately, the success of GODAE will be
judged by the benefits delivered to users. To
this point, most of the activity has focussed on
developing core capabilities (data streams, data
and product servers, models, data assimilation
systems, applications), as detailed in the
previous sections. These activities are the focus
of the Demonstration Phase.           However, in
parallel with this activity there is an increasing
number of projects with the user community
dedicated to exploring and extending the use of
GODAE data and products.

GODAE will deliver both short-term and long-term (sustained) benefits. The short-term benefits will
often be realised through established activities of the application centres. For example, the operational
and experimental seasonal-to-interannual prediction activities have established mechanisms for
interpreting and transmitting useful information to the user community (see the previous Section and
references therein). There will also be short-term benefits arising from the Pilot Projects and the data
and product serving activities, independent of core GODAE modelling and data assimilation activities.

The longer-term benefits will require specific activity from GODAE to cultivate and nurture value-
adding activities (Fig. Error! Reference source not found.Error! Reference source not found.).
This will require dialogue with end users, either through feedback mechanisms and involvement in
external metrics (external evaluation of GODAE products and services), or through fora and
workshops that allow direct dialogue between the providers of information and those seeking to exploit
it. The detail of some of these activities is emerging now and, for others, GODAE must initiate action.

 Figure 2. A schematic illustrating the role of value adding "middle men" in exploiting the GODAE product line for
                                                       specific end-users.

GODAE IP July 20-22, 2004                           Page 57                                      11/06/10 19:06
                                        Linkages to Users

                         Safety &                                          farmer’s
                        Rescue                                             gate
                                      distribution           product

                                                                                    The
                       The
                                                                                   cellar door
                       basket    observations       production
                           The                                             The ship-
                    scientist’s kit               R&D                      master’s chart

   Figure 3. A schematic showing the core "business" of GODAE (observations, data distribution, processing,
                        products) and linking via value adding centres / businesses to the users.

Error! Reference source not found.Error! Reference source not found. illustrates the line of
products from GODAE to the value adding ―middlemen,‖ many of which will be involved in application
centres. The conceptual advantage of this approach is that GODAE maintains its focus on its main
―business‖, a strong, dependable and rich set of products. The middle-men in turn provide specialist
skills for translating these products for particular users or for downscaling to particular regions. Error!
Reference source not found.Error! Reference source not found. provides another variation on this
concept, with GODAE focussing on the ―content‖.

In many cases there is existing or emerging groups that have established, or are establishing working
relationships with the end users. Section 9 discussed these ―lines of action‖ from the broad perspective
of application areas. Here we wish to provide more specific illustrations of the links to users, either
through collaboration with application centres or through more direct interaction with value-adders
and/or users.

A very broad indication of the potential extent of this user community is provided by statistics from the
Navy Research Laboratory (courtesy H. Hurlburt; Figure 4). Though a detailed analysis of the
character of these accesses has not been completed (e.g., how many were related to data and
product downloads; how many are routine and regular; how many from the education sector)
anecdotal evidence suggests interest is very broad and includes a broad public use/service
component, mainly in secondary and tertiary education. Such figures, and experiences from other
groups, give great encouragement that GODAE products are both relevant and useful.

This analysis also illustrates a principle that has been fundamental in the development of GODAE
related to efficiency. It is clear that a broad range of users, spread globally, is finding the product suite
of the NRL Oceanography Division useful and, we can surmise, very few of these users were
specifically targeted in the development of these products. Rather, a comprehensive real-time global
ocean system is able to simultaneously meet the particular needs of the Navy and the unspecified
needs of multiple external users. Such multi-purpose use delivers efficiency that is lost if individual
users, or even groups within a sector, are forced to develop such functionality, replicating many times
over the systems and complex infrastructure that GODAE groups are able to develop.

The analysis also reminds us that the benefit of GODAE is likely to be realised in a variety of ways. In
this particular case, the sponsor of the system (the Navy) derives direct, substantial benefit and is
intimately involved in shaping the system to meet its needs. However, there are likely many users who
draw only marginal or occasional benefit but, through shear weight of numbers, represent an

GODAE IP July 20-22, 2004                         Page 58                                    11/06/10 19:06
aggregated benefit that is substantial and enduring. GODAE has good knowledge of the first group
since they are very directly involved in the patronage and support of GODAE. The second group, and
all those users in between, is less well known and GODAE must work with various sectors to try to
better understand its role and relevance.

               Figure 4. An analysis of the number of accesses to the Navy web site during 2003.

10.2 Product Lines
A range of products was described in Sections 7, 7 and 7 and Section 9 described the main
application areas and products associated with these areas. At this stage, it is not clear which of these
products will form a basic global product set (i.e. be common to most of the centres) and which will
take on a more specialist nature. These ―product lines‖ will be the foundation upon which the benefits
of GODAE will be built.

Annex I contains a list of URLs for the present activities.

In line with the comments of Pierre B and others, this section needs to be extended to provide
a concise description of the GODAE product stream. It might in the form of a product table, or a
list of products from each of the GODAE partners. A list of URLs misses this.

10.3 Reaching the user community
In this sub-Section, we provide examples of initiatives within GODAE that are building a relationship or
identifying user needs. It is important to note that many of these relationships are not exclusive and
that the role of GODAE and its products may vary from pivotal to marginal.

It is important to note that GODAE is not seeking exclusive relationships with many of the users but is
part of a broader effort to extend the benefits of operational oceanography to the broad community in
both the private and public sectors. The development of a Coastal Ocean Observing System and the
associated Implementation Plan (COOP, 2004) has been central in identifying user needs in the

GODAE IP July 20-22, 2004                         Page 59                                     11/06/10 19:06
coastal community and in developing systems that will satisfy these needs. Similarly the Ocean
Climate Observing System and associated implementation plans (e.g., GCOS 2004) provide the
leading guidance for the climate community. GODAE plays a bridging role, developing infrastructure
and capacity that, for now, is accessible only to a relatively small and specialised sector of the
community but, through the promotion of principles of product sharing (GODAE Common) and multiple
use, is able to reach and contribute to the capacity of many. GODAE needs to work with those who
have special knowledge and relationships with these users but also be prepared to nurture and
development uses more directly.

10.3.1 GODAE Symposium
The GODAE Symposium represents the first attempt to create a global forum for discussion of
products and for learning in greater detail about the broad user requirements. Each of the GODAE
projects (see Section 7) have undertaken such analyses for their region or sector. The Symposium
provides an opportunity to explore more deeply the common needs and to develop actions for GODAE
that will ensure more efficient and effective services.

I suggest we then include paragraphs introducing each of these sectors (including climate!)
with a specific example in each case, drawn from GODAE partners, that illustrates in concept
or more directly such use/benefit. For example, MERCATOR has a live example of use of
products for oil spill response. BLUElink has an example of products being used search and
rescue. NRL has an example relate to biogeochemistry. HYCOM is developing relationships
with coastal managers. ECCO has developed products for monitoring climate change in the
ocean. JMA have products used for Seasonal-to-Interannual prediction.


Oil Spill Forecasting

Ocean forecasts provide important inputs to models for forecasting the movement and dispersion of oil
spills. Data on spills from satellite SAR and aircraft surveillance can be used to initialise these
forecasts. Attribution of minor spills to particular vessels by inverse modelling acts as a deterrent to
illegal deliberate discharges of oil.

Oil spill forecasting in the Baltic was stimulated by a HELCOM resolution in 1990 that every country
should have access to an oil drift forecasting system by 1993. It was specified that these systems were
to be fully operational, using advanced models and continuously updated information and to be
accessible through user-friendly interfaces which could be operated by the teams combating the slicks.
A first response to oil spills is expected within 2 hours of the first alert and actions combating the spill
are expected within 12 hours. The Seatrack system was set up with a web interface
(http://produkter.smhi.se/seatrack/) to enable Municipal Rescue authorities and the Swedish
Coastguard to specify the location and type of spill and receive forecasts of the track of the spill within
a few minutes.

Boundary data for coastal and shelf-seas systems

Forecasts for coastal waters and shelf-seas can benefit from information on deep ocean currents,
temperatures, salinities (and biological variables) propagating into their domains through their open
Kindle (communication to 6 IGST) has demonstrated the impact of Kelvin waves propagating into
models of the California Current System. The benefits of nesting coastal models within larger area
models in European waters will be studied within the Mersea Integrated Project.

The POLCOMS (Proudman Oceanographic Laboratory Coastal Ocean Model System) configurations
operated on a daily basis by the Met Office are nested within suitable deep ocean FOAM

[Further material to be added.]

GODAE IP July 20-22, 2004                        Page 60                                   11/06/10 19:06
10.3.2 User forums
This subsection needs to describe the various user forums that have been developed to guide project
development. This will include links to the research community.

10.3.3 Regional alliances
We are using Regional Alliances, established to coordinate regional participation in GOOS, as brokers
for GODAE products. PIRIOS in the South Pacific is an example where we are reaching out to the
small island developing states.

IOGOOS is an example where we are developing joint projects.

In the US, it will be via the regional associations.

Potential Applications of Ocean Observations for the Pacific Islands

Needs to be updated. Wilson and Smith.

The Argo Science Team initiated discussion on a Workshop to showcase the value of ocean
observations, and Argo in particular, for small island states in the South Pacific region. For Argo,
collaboration and cooperation of this region is critical since their collective Exclusive Economic Zones
occupy much of the western Pacific thereby posing a potential problem for global deployments.

The overall objectives of the Workshop are:

    1. Review potential applications of ocean observations—including those to be collected by new
       sources such as Argo profiling floats—that can be of benefit to the Pacific Islands.
       Applications include seasonal-to-interannual climate forecasting, monitoring and predicting
       sea level change, operational ocean and marine services, assessment/prediction of the health
       of coral reefs, fisheries population modeling, basic research in ocean variability and air-sea
       interaction, and secondary education.

    2. Identify data product and timeliness requirements that are needed for each of these
       applications, and assess the extent to which these requirements are presently being met. In
       the case of needed improvements, especially those to which new data sources such as Argo
       can contribute, establish linkages between those who will develop the enhanced products and
       services, those who will routinely create them, and those who will use them.

The Workshop will focus on several applied themes including seasonal prediction, climate change and
ocean and marine forecasting. For the latter, it is self-evident that the Pacific Islands and their
Exclusive Economic Zones occupy a vast area of the Pacific Ocean. Sustainable development and
management of exploitation and risk are important considerations for the region, and each is
dependent to some extent upon access to ocean data and information. There are also common
issues concerning marine biodiversity, regulation of shipping and transport, and safety — particularly
for recreation and tourism. Access to timely ocean and marine products and forecasts is important.
GODAE will develop a system capable of meeting regional requirements.

This Workshop is scheduled for August 2002 (Contact: S. Wilson Stan.Wilson@noaa.gov).

Indian Ocean GOOS

Desperately needs to be updated.

In 2000, a Workshop was convened in Perth to look at the existing and potential sustained observation
networks for the Indian Ocean. Unlike, say, the North Atlantic, the Indian Ocean does not have access
to large resources for observation nor, in most cases, access to sophisticated model and data

GODAE IP July 20-22, 2004                         Page 61                              11/06/10 19:06
assimilation system. The so-called SOCIO Workshop developed a broad-based rationale for sustained
observations under the assumption that the combine needs and demands of this broad base would
provide the impetus for action and implementation. A critical element of the approach was that any
action plan must involve the region, both in their capacity as potential providers of data and in their role
as users. GODAE was prominent in the discussions and plans.

A second Workshop is planned for Mauritius in late 2002. Once more, GODAE and its objectives will
be prominent. The Workshop provides an opportunity to showcase available products for the region
and to describe likely product lines that will be available in 2005. The challenge is to develop
partnerships of the form depicted in Error! Reference source not found.Error! Reference source
not found. in order to promote uptake
of GODAE products for local use and

10.4 Mechanisms          for
See Strategic Plan for initial outline

Measuring the utility of GODAE
products and the components of the
end-to-end system is central to the
objectives. Data and product servers
should have software interfaces to
enable ease of feedback from
application centres and end users,
and to facilitate feedbacks between
the providers of data and assimilators,
etc.     To provide the basis for
evaluation, GODAE will:

         monitor access to the data and product servers,
         ask users to register their use and provide feedback,
         ask application centres to identify users.

10.4.1 Possible external measures
         Availabaility and accessability of products (e.g. number of bytes per day downloaded by
          users - .gov, .edu, .com)
         Utility of products
         Feedback from science community, FNMOC, NOAA, and nternational partners,
          International Linkages Task group
         E.g. number of scientific papers published using GODAE products; comparison of
          assimilation products with observation-only products
         Quantified improvement in operational products through GODAE partnerships
         Predictability enhancement, cross-validation statistics
         Impact on other disciplines
         E.g., number of partnerships established with, e.g. biogeochemical or coastal communities

GODAE IP July 20-22, 2004                        Page 62                                   11/06/10 19:06
11 GODAE Work Plan

Over the past two years, the main GODAE components have been developed and most of the needed
infrastructure at the national and international levels has been set up. The previous sections of this
document summarize the main outputs of this development and pre-operational phase. Several
important issues have been raised, however, to ensure that the GODAE components will support in
combination a coherent demonstration of routine monitoring and prediction of the ocean.

The main identified issues are (see Sections 3 to 8):

         Observing system
         Specific data products
         Data characterization
         Forcing fields
         Quality control
         Feedback on the observing system
         Model and Assimilation method improvements
         Product Intercomparison and internal metrics
Issues related to application centres and end-users are analysed in sections 10 and 11.

These issues will be best addressed through international collaboration and sharing of experience and
knowledge between GODAE partners. They will be mainly realized during the demonstration phase
through the continuation of prototype systems and pilot projects and through directed tasks or working
groups. The regional prototype systems have been designed to test and promote the end-to-end
system (i.e. from the observing system through to the end users) while pilot projects have been
focused on specific components that are considered to be of major importance for GODAE and require
significant and non-trivial developments and contributions.

In the following, we start with a status of the two pilot projects that mainly address the first two issues.
An outline of the work to be carried out for the other issues is then given. The IGST will have to
elaborate a more detailed work plan and associated schedule in the coming months. The section ends
with the description of prototype system activities (North Atlantic, North Pacific and Equatorial Pacific)
and the envisioned intercomparison experiments.

11.1 Pilot Projects

11.1.1 Argo
See Argo implementation plan (link)
11.1.2 SST
The primary aim of the GODAE High Resolution Sea Surface Temperature Pilot Project (GHRSST-PP)
is to develop and operate an operational demonstration system that will deliver high-resolution (better
than 10 km and ~6 hours) global coverage SST data products for use in operational
ocean/atmosphere modelling systems and the wider scientific community. A new generation of SST
data products will be derived and served to the user community by combining complementary satellite
and in situ SST observations in near real time. A dedicated part of the GHRSST-PP is dedicated to
reanalysis of the near real time SST data streams for use in climate applications and construction of
Climate Data Records (CDR). In particular, through the GHRSST-PP reanalysis effort it is expected
that a harmonisation between satellite and in situ SST data sets can be reached and that satellite data
sets can be used with confidence within the long term climate record. A full description of the
GHRSST-PP project can be obtained from the GHRSST-PP project web server located at
http://www.ghrsst-pp.org hosted by the MetOffice as part of the GHRSST-PP International Project

GODAE IP July 20-22, 2004                        Page 63                                   11/06/10 19:06
General scope

The primary user requirements applicable to the GHRSST-PP have been established as follows:
      A common data format for all SST data products is required to facilitate application
       development, data inter-comparison and data management of SST data sets.
      SST observations are required for assimilation by numerical models.
      SST analyses are required for diagnostic/R&D studies
      Uncertainty estimates and observation time are required for each observation and analysis
       grid point
      SST data products should be developed at high resolution in space (10km or better) and in
       time (NRT 6hours to 24 hours) with an accuracy of 0.4K or better
      Timely data delivery is required in an operational manner (i.e. reliable and sustainable)
      Supporting data for quality control and interpretation of uncertainty estimates are required.
The main objectives for the GHRSST-PP are:
      To develop and provide a new generation of high resolution operational SST products
       including NRT and delayed Climate Data Records, that address the needs of national and
       international projects.
      Ensure that the international duplication of SST activities are minimized by synchronization of
       data merging/processing procedures, techniques, algorithms and data formats.
      To Implement an operationally efficient methodology for real time fusion of SST data providing
       increased efficiency and cost-effectiveness of SST product generation and delivery.
      Develop and foster considerable scientific and operational knowledge during the lifecycle of
       the GHRSST-PP by increasing the network capacity within international and national projects
       of differing scope and budget
      To implement an internationally distributed data processing and data management framework
       that can deliver GHRSST-PP data products in NRT to the user community.
The major scientific challenges for the GHRSST-PP are to quantify the uncertainties due to:
      The interpretation of in situ SST relative to the skin and sub-skin measurements provided by
      Specification of the optimal data analysis/merging system to combine complementary
       microwave, infrared and in situ SST data.
      Inaccuracies and differences between SST algorithms/Atmospheric transmission models
      Cloud and aerosol contamination of IR data sets
      Rain/RF/sidelobe contamination of microwave SST data sets
      The extent of sea Ice extent and SST in the marginal ice zone
      Diurnal variability (including sampling alias)
      Satellite and in situ instrument stability and calibration

The main concept underpinning all activities within the GHRSST-PP is that in principle, the merging
and analysis of complementary satellite and in situ measurements can deliver SST products with
enhanced accuracy, spatial and temporal coverage. The emphasis is to capitalise on the synergy
benefits resulting from this approach. For example, MW satellite data sets are able to ‗see‘ through
clouds at a much coarser resolution than IR imagers which are more accurate but opaque to the ocean
surface in the presence of clouds. Figure 5 summarises the general concept of the GHRSST-PP:

GODAE IP July 20-22, 2004                     Page 64                                11/06/10 19:06
Figure 5. The general concept of data merging and analysis underpinning the GHRSST-PP.

Five GHRSST-PP RDAC (regional data assembly centre) projects are currently in preparation as
indicated in Figure 6:
    1. The New Generations SST project (NGSST) serving the western Pacific area replaced by a
       approximately defined by the Geostationary Meteorological Satellite (GMS, now replaced by
       GOES-9) footprint. The NGSST project is based in Japan.
    2. The European Medspiration Project (MSP) serving the Atlantic area and European shelf seas.
    3. The BLUELink> Ocean Forecasting Australia project serving the regional needs of the
       Australian region. This project is hosted by the Bureau of Meteorology (BoM Australia)
    4. The Survey of the Environment Assisted by Satellite (SEASnet) program of the IDD serving
       the tropical oceans. The SEASnet project is based in France.
    5. A project serving the SST needs of the USA under the US National Ocean Partnership
       Program (NOPP) under the general title of ‗SST for GODAE‘. Currently waiting for
       confirmation of funding but includes the US National Hurricane Prediction Centre as a key

              Figure 6. RDAC regional projects and their relationship to the GHRSST-PP GDAC.

A GDAC (global data analysis centre) will be implemented as a joint system by the Jet propulsion
Laboratory Physical Oceanography Data Active Archive Center (PO.DAAC, http://podaac.jpl.nasa.gov)
and the US-GODAE data server system at Monterey (http://www.usgodae.org). The PO.DAAC will be

GODAE IP July 20-22, 2004                       Page 65                                   11/06/10 19:06
responsible for the data management of the GDAC including metadata (using an International Master
Metadata Repository, MMR), and the GHRSST-PP international match up database (MDB, collocated
satellite and in situ data that are used to establish error statistics for each satellite data stream) data
serving and user interactions. The US-GODAE system will be responsible for the archive of GHRSST-
PP data sets and the production of a global analysed SST fields initially using the US-Navy Coupled
Ocean Atmosphere Mesoscale Prediction System (COAMPS).

Data Processing and data products

Each RDAC will implement a data processing system called the GHRSST-PP Data Processing
Specification (GDS) that will generate and disseminate the GHRSST-PP data products. The GDS is
designed to produce SST data products that satisfy the requirements of operational ocean forecast
and prediction systems. Based on a user requirements survey (see the proceedings of the 3
GHRSST-PP workshop and the Medspiration User Requirements Document for example, both
available from the GHRSST-PO or http://www.ghrsst-pp.org), the main requirements are:
    1. SST data are required operationally in near real time and should ideally be available within 6
       hours of data acquisition at the satellite platform.
    2. An error estimate (bias and standard deviation) and confidence data (such as an indication of
       atmospheric aerosol presence at the time of measurement) for each SST measurement
       including a bias and SD. is required by users and by the GHRSST-PP as a necessary
       precursor to data analysis.
    3. A global ocean SST data product providing a best measure of the Foundation Temperature
       (SSTfnd, defined as the temperature of the upper layer free of diurnal variation) and SSTskin
       (including an estimate of diurnal magnitude and phase) is required at a minimum interval of 24
       hours having a maximum spatial grid size of 1/12° (latitude x longitude).
    4. Coastal modelling groups require ultra-high spatial resolution SSTfnd and SSTskin data
       products on a grid cell size of 1-2km and a timeliness of ~6 hourly.

The International GHRSST-PP Project Office (GHRSST-PO)

The GHRSST-PO is responsible for the international co-ordination of the logistical, political, scientific,
and the administrative aspects of the GHRSST-PP, the GHRSST-PP Data processing Specification
(GDS) and the Diagnostic Data Set (DDS) under the oversight of the GHRSST-PP Science Team. In
practice, the GHRSST-PO manages the GHRSST-PP in cooperation with international, national and
regional institutions, committees, and offices as well as related global programmes. It acts as a central
point of contact during the implementation of the GHRSST-PP and interacts with related international
scientific and intergovernmental bodies. It monitors and oversees the management of GHRSST-PP
data sets, and it will ensure good information flow among GHRSST-PP participants

The GHRSST-PO is located at the Hadley Centre for Climate Prediction and Research, part of the
Met Office, UK (http://www.metoffice.com) which is one of the worlds leading providers of
environmental information. The GHRSST-PO is jointly funded by the European Space Agency (2/3)
and the Met Office (1/3). Contact Craig Donlon using craig.donlon@metoffice.com for further

GHRSST-PP Schedule

The GHRSST-PP project implementation timeline is split into three main parts; a preparation phase, a
demonstration phase and, an intensive application phase as indicated in Figure 7.

The GHRSST-PP preparation phase (2003-2004) is concerned with engaging and consolidating the
GHRSST-PP user, scientific, and operational community as a networked team. This period of the
project focuses on the creation of a globally distributed system that can use satellite and in situ data in
real-time in an operational manner. A version 1.0 of the methods and algorithms used to generate
GHRSST-PP data products will initially configure the GHRSST-PP system during the preparation
phase. The emphasis during this phase of the project is to implement and test an internationally
distributed demonstration system that can be upgraded to provide better quality data products as the
project progresses.

GODAE IP July 20-22, 2004                        Page 66                                   11/06/10 19:06
        Figure 7 A summary of the GHRSST-PP schedule identifying major project events and phases.

A Prototype Phase will focus on the commissioning of the GHRSST-PP demonstration system in late
2004 in collaboration with the NASA Jet Propulsion Laboratory Physical Oceanography Data Active
Archive Centre, the US-Navy Naval Research Laboratory, MetOffice and the EU Medspiration project.

The preparation phase will be superseded by the GHRSST-PP demonstration phase in 2005 which
will continue until 2007. During the demonstration phase, GHRSST-PP data products and services will
be made available, in real time, to a broad GHRSST-PP user community. Throughout the
demonstration phase, a parallel and continual process of project development and refinement will take
place with particular emphasis on (a) the improvement of the scientific methods used to generate the
demonstration data products and (b), the timely delivery of these products to operational users.

During 2004-2006, the GHRSST-PP will also operate in an intensive application phase in parallel to
the demonstration phase. During this period, GHRSST-PP data products will be provided to a number
of operational users (E.g., ECMWF, UK Met Office, MERCATOR, FOAM, etc) who will work closely
with the GHRSST-PP Science Team to evaluate the impact of GHRSST-PP data products in their
operational applications. Dedicated ocean and atmosphere model assimilation runs and model output
inter-comparison exercises will all take place in real time. Two key workshops will be used to
consolidate the experiments that will be undertaken with the GHRSST-PP operational application
community. The first will be used to specify the exact period, area and, input-output definitions that will
be used together with establishing metrics and assessment criteria that will be used to asses the
performance of GHRSST-PP data products as used in each application. The second workshop will be
used as a forum in which the results of the GHRSST-PP experiments will be presented (2006/7).

11.2 On-going Actions and Work

11.2.1 Data set characterization for assimilation
The timely distribution of data and data products (through data assembly centres and data servers) to
GODAE assimilation centres is not sufficient to ensure that data are properly used in the assimilation
systems. Data sets must be characterized (error covariances, information content) and for some data
sets additional processing is needed to get data sets ready for assimilation.

Below is a list of tasks that will require collaboration among GODAE partners:

         Observation data error characterization. Different assumptions will be made by the
          GODAE partners depending on their model (e.g. resolution) and data assimilation systems.
          Sharing of this information will be highly beneficial. Specific studies should also be

GODAE IP July 20-22, 2004                        Page 67                                  11/06/10 19:06
            undertaken to intercompare input data sets to better characterize errors. GODAE should
            also coordinate efforts to ensure that there is continuing discussion of the appropriateness
            of these error assumptions.
         Impact of altimeter and Argo data error characterization on analyses and forecasts.
         Content of altimeter (sea level) data: upper/lower steric, barotropic component (high and
          low frequency), relative importance of salinity and temperature. This is needed to improve
          the way altimeter data are assimilated in ocean models. In particular, knowledge of the
          covariance between sea level and temperature and salinity fields is critically needed.
         Absolute dynamic topography from altimetry: There is a strong need to estimate a mean
          dynamic topography consistent with altimeter sea level anomaly data to get absolute
          dynamic topography measurements from altimetry. This will allow us to much better
          constraint the mean state of the ocean. Several mean dynamic topographies have been
          and will be estimated by GODAE partners. They make use of in-situ data (hydrography,
          drifters) and/or gravimetry (CHAMP and GRACE) and/or model outputs (with/without data
          assimilation). Characterization of their errors and analysis of their impact on the analyses
          and forecasts should be carried out.

11.2.2 Forcing fields
GODAE will depend heavily upon the operationally produced ocean surface forcing data sets (wind
stresses, heat fluxes) from the various national meteorological centres Simple intercomparison of
these forcing data sets has revealed that very considerable differences often exist between different

The timely distribution of NWP forcing fields is thus far from being sufficient for GODAE. Forcing field
errors need to be characterized. For this reason, GODAE seeks to establish links to all the operational
centre air-sea forcing data sets in real time or near real time, and to quantify the differences between
them. The SURFA project (see Section 7) will be important for this task. In the absence of widespread
ocean surface validation data, these differences will be used to prepare estimates of the uncertainty in
air-sea forcing. The amplitudes, covariances, and spatial scales of these differences are needed as
input to ocean data assimilation systems. Another complementary approach will be to compare the
NWP products with satellite products. Satellite products may offer, actually, an interesting alternative
for specific applications that do not require forecasting capabilities and high frequency resolution (e.g.

The following is a list of tasks that should be carried out as part of GODAE activities to characterize
the atmospheric forcing fields (analyses and forecasts):

         Gathering information on NWP products and their evolution (resolution, assimilated data,
         Intercomparison of the different NWP products
         Comparison of NWP products with satellite products (wind, radiative and turbulent heat
         Determination of error characteristics (covariances)
Impact studies should also carried out to analyse the impact of forcing field errors on ocean analyses
and forecasts but also to better assess the contribution of satellite products. In the longer run, one
should seek to improve the NWP products using satellite products and by developing stronger links
with the NWP community.

11.2.3 Quality control of observations
Data quality control is a fundamental component of any ocean data assimilation system. In NWP, it
has been found that decisions made at the quality control step can determine the success or failure of
forecasts. Quality controlled observations are also required in delayed-mode to assess the quality of
the products issued from the assimilation system itself. Quality control should involve the checking of
observations using scientifically based procedures. At present, the scientific basis for the quality
control of ocean observations is weak, and very few comparisons have been made of the procedures

GODAE IP July 20-22, 2004                       Page 68                                  11/06/10 19:06
in place at the operational centres or the results from them. A workshop has been organized
immediately prior to the Biarritz GODAE symposium to discuss the potential and priorities for the
exchange of information and collaboration on the quality control of ocean observations. This effort is
of enormous scope and the workshop at the GODAE symposium is only the initial step of a longer
process that will extend throughout GODAE. One aim of the GODAE workshop is to ensure that the
results of quality control procedures are recorded for independent analysis and later use.

11.2.4 Ocean Modelling
The ocean numerical model is one of the three essential components of an ocean forecasting system.
Observational data, via data assimilation, set the stage for the model forecast. The quality of the
forecast will however primarily depend on the ability of the ocean numerical model to faithfully
represent the ocean physics and dynamics. Even the use of an infinite amount of data to constrain the
initial conditions will not necessarily improve the forecast against persistence of a poorly performing
ocean numerical model. GODAE participants will therefore need to not only quantify the performance
of their model component when in free-running mode (i.e., no data assimilation), but also assess the
impact of model free-running simulation skills on data assimilation and forecast skills.

Most of the ocean models used within GODAE are widely used by the oceanographic community and
user groups have been in existence for several years (see section on national activities for a list of the
GODAE numerical models). As shown by various ocean modeling studies such as DYNAMO (Meincke
et al., 2001) and DAMÉE-NAB (Chassignet and Malanotte-Rizzoli, 2000), the choice of a vertical
coordinate system is the single most important aspect of an ocean model's design. Currently, there
are three main vertical coordinates in use, none of which provides universal utility. The models chosen
by the GODAE participants predominantly use z-coordinates (MOM, OPA, POP, MIT, …), isopycnic
coordinates (NLOM), or hybrid coordinates (HYCOM, NCOM, NEMO). The practical issues of
implementing subgrid-scale parameterizations are often directly linked to the vertical coordinate
choice. The GODAE assimilation groups will naturally take advantage of their respective model‘s user
groups as well as working groups that have been put in place to improve model numerics within other
programs such as WOCE and CLIVAR.

Specifically, GODAE will be able to provide an assessment of numerical choices made when
configuring the GODAE systems by evaluating the impact of these choices on data assimilation and
forecast skills. Several modeling issues are most relevant to the GODAE goal of operational
oceanography (Chassignet et al., 2002):

           systematic model error, data constraints mostly from the surface, prescribed atmospheric
           topography, ice modeling, representation of overflows, coastal transition zone, tides
           Subgrid scale parameterization: diapycnal mixing, viscosity, diffusivity

Chassignet, E.P., and P. Malanotte-Rizzoli (Eds.): Ocean Circulation Model Evaluation Experiments
   for the North Atlantic Basin. Elsevier Science Ltd., Special issue of Dyn. Atmos. Oceans., 32,
   (2000), 155-432.

Chassignet, E.P., M.J. Bell, P. Brasseur, G. Evensen, S.M. Griffies, H.E. Hurlburt, C. LeProvost, G.
   Madec, J. McClean, J. Verron, and A. Wallcraft (2002): The Modeling Component of Ocean
   Forecasting. In En Route to GODAE”, Symposium on the Global Ocean Data Assimilation
   Experiment, Biarritz, France, June 13-15, 2002.

11.2.5 Data assimilation
Some of the most important and pressing developments for GODAE are:

         Better multivariate error covariance models to make effective use of all data streams
         Determination of error covariances for surface stresses and air-sea fluxes,
         Assimilation of sparse in situ data into eddy-resolving models,

GODAE IP July 20-22, 2004                       Page 69                                  11/06/10 19:06
         Estimation of representativeness errors for data with respect to the hierarchy of models
          used for GODAE,
         Methods for assimilation of different types such as Lagrangian current measurements,
         Innovative applications of assimilation tools to investigate observing system design
         Utilization of error estimates to identify specifics of model deficiencies,
         Correction of model biases, and
         Identification of appropriate and tractable assimilation methodologies for routine
          operational implementation.
Partnerships are evolving between groups involved in assimilation of oceanographic data. For
example the ENACT (Enhanced ocean data Assimilation and ClimaTe prediction) project, funded by
the European Commission (EC), seeks to intercompare advanced methods for data assimilation for
seasonal forecasting. A partnership within the U.S. on data assimilation for seasonal forecasting has
just been established under NOAA‘s program on Climate Diagnostics and Experimental Prediction.

The IGST is encouraging the exchange of information and experience in ocean data assimilation
through workshops (e.g. GODAE/WOCE Assimilation Workshop, Baltimore, March 1998; International
GODAE Workshop, Hawaii, July 2001). These contributions to building the GODAE common will
become more concrete as intercomparison plans (section 6.4) come to fruition.

Within GODAE, there is scope for relatively simple but useful intercomparisons. Intercomparisons of
the time mean surface height fields used to assimilate altimeter data have proved to be useful.
Following the experience of NWP, intercomparisons of the model and observation error covariances
used by the assimilation centres may be worthwhile. The International GODAE workshop in Hawaii
enabled experience to be shared on a number of problem areas.

11.2.6 Product Characterization and Internal Metrics
The assimilation centres need to evaluate the performance and effectiveness of their systems. This
requires the definition of internal metrics, which can be systematically used and compared by the
assimilation centres. This also requires that some selected fields (e.g. SST, temperature and salinity at
certain depths) be systematically compared. The quantities that will be considered include
temperature, salinity, velocity, time-varying mixing tensors, sea surface height, and other passive
tracers. From these variables, other quantities are derived that are representative of major dynamical
and thermodynamical ocean characteristics, such volume transports of major currents, mixed layer
depth, heat fluxes and water mass characteristics.

The internal metrics will be based on three types of diagnostics:

1) Consistency analysis: diagnostics based on the experience in understanding and modelling of the
   ocean by the oceanographic community,

2) Quality analysis: assessments based on direct comparisons with available observations of some
   of the above listed variables, accessible either in real time, or delayed mode,

3) Performance analysis: assessments aiming at the evaluation of the technical effectiveness of the
   systems in terms of modelling and data assimilation (innovation vector, residual vector, forecasting

The scientific rationale and a more detailed description of the GODAE internal metrics are given in Le
Provost et al. (2002). This description will be complemented by several technical documents that will
specify the list of metrics to be implemented by the different modelling/assimilation centres. This
activity has started for the North Atlantic, North Pacific, and the Equatorial Pacific.

GODAE IP July 20-22, 2004                       Page 70                                 11/06/10 19:06
The internal metrics for the North Atlantic and the Mediterranean basin are described in the following
document: Metrics definition for the North Atlantic and Mediterranean sea, C. Le Provost et al., 2004,
Version 5, May 3 2004, MERSEA Strand-1 - WP4 intercomparison project (available at
http://www.mersea.eu.org). Metrics were sorted into 4 classes (Class 1, Class 2, Class 3 and Class 4):
     Class 1: 2D fields (wind (x, y), total net heat fluxes including relaxation terms, water flux (E-
      P-R) including relaxation terms, Barotropic Stream Function, Mixed layer depth MLD ( and
      ), sea surface height SSH), 3D fields (T,S,U,V), 3D fields (T,S) from climatology.
     Class 2: (T,S,U,V) fields along high resolution sections and at moorings locations.
     Class 3: Integrated quantities such as daily volume transport through given section, meridional
      heat transport, Overturning stream function.
     Class 4: Statistics in the model and observation space to assess data assimilation method
      performances (forecast skill).

A detailed description of the MERSEA Strand-1 North Atlantic and Mediterranean basin
intercomparison    experiment      and      the   metrics   document   can  be    found    at
http://www.mersea.eu.org/html/strand1/intercomparison.html (see also section 9.10.1). Metrics
document for the North Pacific can also be found at the MERSEA WWW site.

Diagnostics and metrics will be progressively extended to the global ocean through the
intercomparison exercises that are starting in other regions (North Pacific, Equatorial regions,
Southern hemisphere).

11.3 Running regional and global prototype systems and transition to
     operational systems
Prototype systems contain most of the functions of the final system. They are designed to test and
promote the end-to-end system (i.e. from the observing system through to the end users) at regional
scales (e.g. North Atlantic) or at global scales with preliminary versions of the modelling and
assimilation systems (e.g. lower model resolution, simplified assimilation scheme). The main objective
is to gain experience on real-time data flow aspects and to begin product design, product assessment,
and product distribution in response to user specifications. This experience will be directly applicable
for the final systems.

Several regional prototype systems have been running over the past years (see section 8) (North
Atlantic, North Pacific and Equatorial Pacific). During the main demonstration phase when global
systems are in place, prototype systems will generally continue to be used. They will provide more
flexibility (because of lighter configuration) for testing improvements in model and data assimilation
techniques, experiencing new data and product serving capabilities…

Over the period (2002/2005), the prototype systems will be also used for several intercomparison
experiments (see next section). The main objective will be to define a methodology to assess the
quality of GODAE products and to identify strengths and weaknesses of the data assimilation systems.
The intercomparison experiments will use a series of internal metrics and the systematic comparison
of selected fields (following the strategy defined in section 11.6).

To be completed : explain the transition towards operational systems.

GODAE IP July 20-22, 2004                       Page 71                                  11/06/10 19:06
11.4 Intercomparison activities

11.4.1 North Atlantic
A complete intercomparison exercise was carried out as part of the EC MERSEA Strand-1 project.
The intercomparison included the MERCATOR, FOAM, TOPAZ and HYCOM systems for the North
Atlantic and the MFS, FOAM and MERCATOR systems for the Mediterranean sea. The systems were
                                            st              st
intercompared over a one year period (June 1 , 2003 – June 1 , 2004).

To help the intercomparison exercise, it was first requested that products from the different
assimilation systems be harmonised and standardised as follows:
    •   Model documentation: Each center listed and documented their operational system and their
        distributed data (model overview, data status, history file).
    •   Convention and format issues: A high degree of product coherency / harmonisation was
        asked to greatly reduce the complexities of the intercomparison exercise. We chose the
        NetCDF format and COARDS/CF conventions.
    •   Grids: All Class 1 fields from the systems were interpolated on the same MERSEA grid with a
        horizontal resolution of 1/8° and vertical resolution with 8 vertical levels for the Mediterranean
        and 12 levels for the Atlantic.

Each model partner implemented an Opendap aggregation server to make data available on Internet.
The main purpose of this distribution server was to read NetCDF data files and allow the "aggregation"
of multiple datasets into one (virtual) dataset. Each ocean system partner also implemented the
internal metrics (Class 1, 2, 3 and 4) as defined in section 9.7.

A European GODAE WWW portal was developed (http://www.MERSEA.eu.org). It includes:
    •   A detailed description of the MERSEA Strand-1 intercomparison experiment.
    •   A central federating LAS connected to individual OpenDAP data servers
        (http://las.MERSEA.eu.org). This portal to European model products federates and unifies all
        individual Opendap catalogues.

The real-time experiment went as follows:
•   Every week the forecasting systems produced the required NetCDF files, exposed them via their
    Opendap aggregation server and upgraded their configuration with the previous week analyses
    and the sixth day forecasts.
•   At the same time, each forecasting system produced an Opendap/LAS synchronisation file called
    ‗notification‘. This file is a formatted ‗xml‘ file, which is put on an ftp site (identified by its release
    date) to communicate the entrance of new data to the manipulation/visualization software (LAS).
    There is an ftp site for each system.
•   LAS synchronisation was performed every day of the week. As a result of this daily interrogation,
    the new daily fields were automatically charged on the visualization software (LAS) to be ready for
    their manipulation and visualization on Internet.

Main outcomes and recommendations from the intercomparison experiment

The main outcomes of MERSEA Strand-1 North Atlantic intercomparison experiment are listed
o   Data and data service that have been developed over the past years are providing ―good‖ (i.e.
    directly useable and validated) data for basin-scale assimilation systems. This is particularly true
    from the SSALTO/DUACS and CORIOLIS systems for respectively altimetry and in-situ data.
o   Sound methodologies to assess the product quality (metrics) have been defined and fully tested.
    They provide an effective means for assimilation system validation.
o   Mersea Strand-1 implemented and successfully run an Information Management System
    prototype (Opendap/LAS/WWW) for an ocean gate on Internet. This is an efficient preliminary

GODAE IP July 20-22, 2004                         Page 72                                    11/06/10 19:06
    system that can be easily adapted to other systems. It uses a standardized approach for model
    product intercomparison and distribution:
        o   ocean metadata and data coherency (NetCdf format, convention, grid), class definitions
        o   decentralised architecture: each center in charge of its own products with a compatible
            model system architecture for distribution (notification, data status, history file).
o   A successful demonstration of real time operation and inter-comparison of basin-scale systems
    was carried out. This allowed us to test the whole chain (from observing systems to products).
    Production was almost nominal for the different systems and products were delivered on time.
    Access to input data was quite satisfactory.
o   Product evaluation. This is the first time that an inter-comparison exercise has been conducted on
    five forecasting systems over Atlantic & Mediterranean Basins. The metrics definition and
    standardisation efforts (grid, format, opendap and LAS) have strongly facilitated the
    intercomparison and internal assessment exercise.
o   These efforts had a very positive impact on product improvements. This exemplifies the ―virtuous‖
    chain: intercomparison and validation => identification of weaknesses => corrections and
    improvements. Several major upgrades of the different systems were thus made thanks to inter-
o   Differences between systems are, however, (still) too large. Product improvements are needed;
    there are, for example, inconsistencies between products for key parameters (e.g. velocity). This
    requires R&D works (modelling and mainly data assimilation) and better observing systems.

From this assessment based on existing pre-operational systems, the following improvements and
recommendations are proposed:
    o   There is a strong need for a precise mean dynamic topography in the open ocean and near
        the shelves. GOCE should allow us to partly solve the problem but a significant effort will have
        to be devoted to developing the use of GOCE product (validation, impact studies, merging
        GOCE products with in-situ data and models).
    o   We need multiple altimeter data sets to constrain the ocean mesoscale circulation and upper
        ocean currents. Our systems show a significant degradation when the number of altimeters
        flying decrease. A high resolution altimetry observing system is needed. This is likely to
        require a constellation of altimeters (up to 3 in addition to the Jason series) and, in the longer
        run, when the concept is demonstrated, wide swath altimeter systems.
    o   There is a strong need for in-situ data to constrain T and S fields. Argo array should be, in
        particular, completed and maintained.
    o   Operation of SSALTO/DUACS and CORIOLIS systems should continue on the longer run to
        provide real time/delayed mode intercalibrated/merged high quality altimeter data and in-situ
        data from different sources (Argo, XBT, drifters).
    o   There is a strong need to develop new high resolution SST products. This requirement is
        mainly addressed through GHRSST-PP and Medspiration projects.
    o   There is a need to develop surface current data product for model validation and for serving
    o   Forcing fields must be improved, in particular, by merging satellite (scatterometry, heat fluxes,
        precipitation) and ECMWF products.
    o   We must continue the routine internal assessment of product quality and extend it to the other
        systems and the global ocean (GODAE and MERSEA Integrated Project). This is crucial to
        ensure that by 2008 high quality products are delivered for GODAE applications. A specific
        effort must be made to develop metrics for ecosystem variables.
    o   A prototype system is now in place for product exchange, intercomparison and distribution
        (standards, NetCdf, conventions, Opendap, LAS, WWW, etc).         Some Mersea Strand-1
        constraints can/could be relaxed (e.g. common grids). Improvements are needed for product
        distribution service and work on new tools to manipulate and visualise the data should be
        carried out.

GODAE IP July 20-22, 2004                       Page 73                                  11/06/10 19:06
    o   There is a need for higher resolution and better physics for ocean models, in particular, close
        to shelf regions.
    o   Improvements in data assimilation techniques are needed for a better use of all observation
        data. This requires, in particular, fully multivariate assimilation schemes that are now tested in
        R&D mode and should be integrated in the operational systems.

11.4.2 North Pacific
Planning for product intercomparisons in the North Pacific is not as advanced as for the North Atlantic,
however, there was much interest and support expressed for such activities at the ―International
Workshop on GODAE with Focus on the Pacific‖, held at the IPRC, Hawaii, July 2001. At present, one
intercomparison experiment has been initiated, facilitated by a partnership between JMA and IPRC.
The experiments will not be controlled in the sense that participants will use initial conditions, surface
forcing, model resolution, etc. of their own prototype or operational systems. Participation in the
intercomparison will be open to interested partners conducting routine basin-wide analyses and
forecasts. Participants will prepare a common set of diagnostics, document the models and
assimilation systems, products and performance on the web, and make a standard set of products
available through the Internet for inclusion in web-based analysis tools. The IGST has recommended
that details of the intercomparison exercise be posted on the IPRC web page as well as on the
Monterey server. It is anticipated that in addition to JMA and IPRC that GODAE partners with global
prototype systems will participate in this intercomparison (e.g., ECCO/JPL, UMD, FRGSC).

Other intercomparisons experiments under development will also contribute to the GODAE Common.
One example is the controlled set of historical re-analyses to be conducted by the NOAA/CDEP
ODASI Consortium in the U.S. Some Consortium members (NCEP, GFDL, NSIPP) will undertake
global analyses with a common set of forcings and input data streams, but using different models and
assimilation systems focused towards seasonal-to-interannual prediction. The set of diagnostic
metrics is still under development.

11.4.3 Equatorial Pacific
The objective of an intercomparison experiment in the equatorial Pacific is focused specifically towards
evaluation and testing of systems for initialisation of seasonal-to-interannual predictions, with a goal of
developing a standard set of metrics for validation of equatorial models that would be maintained by
operational systems on a regular basis.

Discussions at both the ―International Workshop on GODAE with Focus on the Pacific‖, held at the
IPRC, Hawaii, July 2001, and at annual meetings of CLIVAR‘s Working Group on Seasonal-to-
Interannual Prediction (WGSIP) have evoked much interest and support for such intercomparison
experiments. The intercomparison would be primarily through common graphical output, and the
experiments would not be controlled – participants would use different models, resolution, forcing. For
such intercomparisons, the key is documentation of the model, and the experimental set up. Use of
standard graphics and software tools such as the LAS will facilitate the intercomparisons without the
need to collect the output into a central repository. The comparisons will be based on analyses of the
historical database as well as on real-time systems. Analyses will focus on equatorial Pacific sections
                                             o          o
and a few meridional sections between 30 S and 30 N. In addition to the real-time or time-varying
structure in the upper 300m, the climatologies of temperature, salinity, and zonal currents will be
analysed. Interest in participating in this intercomparison has been expressed by ECMWF, BMRC,
NSIPP, COLA, JMA, LODYC, but this participation will undoubtedly broaden as the plans for
intercomparison are more widely disseminated. The IGST will take the lead in developing and
coordinating this activity.

The ODASI Consortium in the U.S. will also undertake the same intercomparisons as agreed upon by
GODAE-WGSIP participants, but under a controlled experimental set up.

GODAE IP July 20-22, 2004                        Page 74                                  11/06/10 19:06
11.4.4 Global

11.4.5 Reanalysis activities
The reanalysis activities should be in section 7 (they are already for most of them but should be more
detailed) and we probably should focus here on the intercomparison activities.

The summary of the reanalysis products should then go into section 10 (serving the climate user

GODAE IP July 20-22, 2004                     Page 75                                 11/06/10 19:06
12 External

[To be completed]

12.1 Liaisons with International Organizations


12.2 Relation to other programs


GODAE IP July 20-22, 2004           Page 76      11/06/10 19:06
13 References

Bell, M. J., R. M. Forbes, and A. Hines, 2000: Assessment of the FOAM global data assimilation
system for realtime operational ocean forecasting. J. Mar. Sys, 25, 1-22.

Bell, M. J., M. J. Martin, and N. K. Nichols, 2001: Assimilation of data into an ocean model with
systematic errors near the equator. Ocean Applications Tech. Note 27, Met Office.

Cooper M., and K. Haines, 1996: Altimetric assimilation with water property conservation. J. Geophys.
Res., 101, C1, 1059-1077.

Davies, H. C. 1976 A lateral boundary formulation for multi-level prediction models. Quart. J. Roy.
Meteor. Soc., 102, 405-418.

Gordon, C., C. Cooper , C. A. Senior, H. Banks, J. M. Gregory, T. C. Johns, J. F. B. Mitchell and R. A.
Wood, 2000: The simulation of SST, sea ice extents and ocean heat transports in a version of the
Hadley Centre coupled model without flux adjustments. Climate Dyn., 16, 147-168.

Kamachi, M., T. Kuragano, N. Yoshioka, J. Zhu, and F. Uboldi, 2001: Assimilation of satellite altimetry
into a Western North Pacific operational model. Adv. in Atmos. Sci., 18, 767-786.

Kuragano, T. & M. Kamachi, 2000: The global statistical space-time scales of oceanic variability
estimated from the TOPEX/POSEIDON altimeter data. J. Geophys. Res., 105, 955-974.

Lorenc, A. C., R. S. Bell, B. MacPherson, 1991: The Meteorological Office analysis correction data
assimilaiton scheme. Quart. J. Roy. Meteor. Soc., 117, 59-89.

Barron, C.N., P.J. Martin, A.B. Kara, R.C. Rhodes, and L.F. Smedstad, 2004a: Description and
application of the Global Navy Coastal Ocean Model (NCOM) with examination of vertical coordinate
system choices. Ocean Modelling. (submitted)

Barron, C.N., A.B. Kara, H.E. Hurlburt, C. Rowley, and L.F. Smedstad, 2004b: Sea surface height
predictions from the global Navy Coastal Ocean Model (NCOM) during 1998-2001. J. Atmos. Oceanic
Technol. (submitted)

Barron, C.N., A.B. Kara, R.C. Rhodes, C. Rowley, and L.F. Smedstad, 2004c: Validation Test Report
for the 1/8° Global Navy Coastal Ocean Model Nowcast/Forecast System, NRL Formal Report (in

Bleck, R. 2002: An oceanic general circulation model framed in hybrid isopycnic-cartesian coordinates.
Ocean Modelling, 4, 55-58.

Cummings, J.A, 2003: Chapter 3, Ocean Data Assimilation, In, COAMPS Version 3 Model Description,
pp 21-28. NRL Publication NRL/PU/7500-03-448, May 2003.

Fox, D.N., C.N. Barron, M.R. Carnes, M. Booda, G. Peggion, and J.V. Gurley, 2002a: The modular
ocean data assimilation system. Oceanography, 15, 22-28.

Fox, D.N., W.J. Teague, C.N. Barron, M.R. Carnes, and C.M. Lee, 2002b: The Modular Ocean Data
Assimilation System (MODAS). J. Atmos. Ocean. Technol., 19, 240-252.

Jacobs, G.A., C.N. Barron, and R.C. Rhodes, 2001: Mesoscale characteristics. J. Geophys. Res. 106,

GODAE IP July 20-22, 2004                      Page 77                                11/06/10 19:06
Hurlburt, H.E., 1984: The potential for ocean prediction and the role of altimeter data. Mar. Geod., 8,

Hurlburt, H.E. and J.D. Thompson, 1980: A numerical study of Loop Current intrusions and eddy
shedding. J. Phys. Oceanogr., 10, 1611-1651.

Hurlburt, H.E., D.N. Fox, and E.J. Metzger, 1990: Statistical inference of weakly correlated
subthermocline fields from satellite altimeter data. J. Geophys. Res. - Oceans, 95, 11375-11409.

Hurlburt, H.E., M.J. Bell, G. Evensen, C.N. Barron, A. Hines, O.M. Smedstad, and D. Storkey, 2002:
Operational global ocean prediction systems. Proceedings of the "En route to GODAE" International
Symposium, 13-15 June 2002, Biarritz, France, pp. 97-105.

Kara, A.B., C.N. Barron, P.J. Martin, R.C. Rhodes, and L.F. Smedstad, 2004: Validation of interannual
simulations from the 1/8° Global Navy Coastal Ocean Model. Ocean Modelling. (submitted)

Rhodes, R.C., H.E. Hurlburt, A.J. Wallcraft, C.N. Barron, P.J. Martin, O.M. Smedstad, S.L. Cross, E.J.
Metzger, J.F. Shriver, A.B. Kara, and D.S. Ko, 2002: Navy real-time global modeling systems.
Oceanography, 15(1), 29-43.

Rhodes, R.C., H.E. Hurlburt, A.J. Wallcraft, E.J. Metzger, J.F. Shriver, A.J. Wallcraft, and A.B. Kara,
2003: Validation test report for the 1/16 global NRL layered ocean model nowcast/forecast system.
NRL/FR/7320--03-10,200, Naval Research Laboratory, Stennis Space Center, MS, 67 pp.

Shriver, J.F., H.E. Hurlburt, O.M. Smedstad, A.J. Wallcraft, and R.C. Rhodes, 2004: 1/32 real-time
global ocean prediction and value-added over 1/16 resolution. J. Mar. Sys. (in preparation for 2004
Liege Colloquium special issue)

GODAE IP July 20-22, 2004                      Page 78                                 11/06/10 19:06
Annex 1: URL List

MERCATOR Ocean Bulletin                                 http://www.mercator.com.fr/
Argo Home                                               http://www.argo.ucsd.edu/

Information on COFS and links. Provides a sample of
COFS products, including Nowcast and 24 hr weather
forecasts, maps and ocean cirulation models.

Bathymetry - BODC
TOPAZ                                                   http://topaz.nersc.no


Consortium for ocean circulation and climate http://www.ecco.ucsd.edu
estimation. Provides animations of models and trends
for sea surface temperature, heat and fresh water

ESODAE                                                  http://www.met-office.gov.uk/sec5/ESODAE/

Provides      meteorological    and      oceanographic
information and models in 3 versions: limited access, http://www.fnmoc.navy.mil/
full access and public access. It has a link to GODAE
on the public access version.

Gulf of Maine, Georges Bank and South Atlantic Bight
Provides real-time data on GLOBEC and NOPP, and http://www-nml.dartmouth.edu/circmods/gom.html
several circulation models and simulations.

Consortium for ocean data assimilation (NOPP
project).    Has regional, basin-scale, and global
assimilation examples.
Met Office FOAM
NAVOCEANO                                          http://www.navo.navy.mil/

New York Bight and Middle Atlantic Bight modeling and
Information on LEO-15 and links to several real time        http://marine.rutgers.edu/cool/LEO15.html
and archived satellite and oceanography data. (direct
link: http://marine.rutgers.edu/cool/data.html)

Observations of water level and ancillary data, and
listing of data that can be provided.

For Internet data- updated every 5 minutes

GODAE IP July 20-22, 2004                    Page 79                                  11/06/10 19:06
NRL                                          Modelling
(NRL) Layered Ocean Model (NLOM), including
snapshots, animations and forecast verification
statistics for many zoom regions, mainly SSH, SST and
surface currents. It also contains direct model-data

U.S. Coastal Observing Systems
Features coastal observing systems around the U.S.;
prepared by NOAA’s Coastal Services Centre

US                                              GODAE
This is a master site, with links to many others. Offers
general information on GODAE and has links to
This is a master site, with links to other In-Situ data
programs and providers such as ARGO, GOSUD, http://www.coriolis.eu.org
OceanSITES as well as MERSEA European project.

Satellite Data centre. Important site for air/sea fluxes,
SST and sea-ice datasets.
Near real time and delayed mode altimeter products http://www.jason.oceanobs.com/html/donnees/duacs
from multiple mission

GODAE IP July 20-22, 2004                     Page 80                                11/06/10 19:06

To top