Mike Bell (Met Office, UK)

1
The Global Ocean Data Assimilation
Experiment (GODAE)
Mike Bell (Met Office, UK) Pierre-Yves le Traon
(Ifremer, France) Kirsten Wilmer Becker (GODAE
Project Office) and the International GODAE
Steering Team www.godae.org Neville Smith and
Stan Wilson
2
Content

• GODAE mission rationale
• Achievements and successes
• Overview
• National progress Australia, USA, Japan, France,
  UK, Canada
• Recent workshop activities
• The GODAE Final Symposium
• Discussion of coordination post GODAE

3
The Mission for GODAE
A practical demonstration of the feasibility
utility of high-resolution, global analyses
short-range forecasts of 3D temperatures,
salinities and currents
Timetable
1997 1999 Conceptual development 2000
2003 Prototype development 2004 2008
Operational demonstration consolidation

4
Specific Objectives

• To apply state-of-the-art models and assimilation
  methods to produce short-range ocean forecasts,
  boundary conditions to extend predictability of
  coastal and regional subsystems, and initial
  conditions for climate forecast models
• To provide global ocean analyses for developing
  improved understanding of the oceans, improved
  assessments of the predictability of ocean
  variability, and as a basis for improving the
  design and effectiveness of a global ocean
  observing system

5
The Rationale and Vision for GODAE

• Rationale
• Tools ready to be used (observations,
  computers, models, assimilation methods)
• Limited window of opportunity to transition
  observations from research to operations
• Approach
• Rapid international exchange of scientific
  information for public good
• Collaboration between research operational
  groups and internationally

6
Achievements Successes

• Support for observing and data processing systems
• Argo (pilot project of Clivar GODAE)
• GHRSST (pilot project of GODAE)
• Altimetry missions
• Implementation of high resolution regional and
  global modelling and data assimilation
  capabilities
• Making data and products freely available through
  web servers
• Intercomparisons using standardised metrics
• Sharing of information and approaches e.g. on
  demonstrations of feasibility utility

7
Systems Developed Australia - BlueLink
OFAM1.0
BLUElink OceanMAPS - Operational ocean prediction
system POAMA - Seasonal forecasting system ROAM -
Relocatable Ocean Atmosphere Model and graphical
user interface OFAM/AusCOM - Global ocean general
circulation models mesoscale/climate BODAS -
BLUElink Ocean Data Assimilation
System RAMSSA/GAMSSA - Operational high
resolution regional/global SST analysis
0.1??0.1?
OceanMAPSv1.0b
ROAM
8
Impact of observations
Real-time system
Impact to ocean reanalysis of different
observations types, Oke and Schiller, 2007, GRL,
34
9
Applications - government
Organisations
Marine Park Authorities Marine park management,
coral bleaching monitoring, eco-tourism
management Australian Fisheries Management
Authority Bio-catch planning and
management Environmental Protection
Agencies Salinity discharge, dredging, industrial
waste management
Heat content and heat content anomaly for
tropical cyclone forecasting, coral bleaching,
seasonal rainfall outlooks
Real-time system
Observed and OceanMAPS analysed Bonney coast sea
surface temperature and surface currents. Showing
an extreme upwelling event.
10
U.S. Navy Present and Planned Global Ocean
Prediction Systems
Global Product Mid-Lat Resolution Vert. Coord. Inputs Run By Actual or Target Date
1/16? NLOM 7 km Layered SSH, SST, hydro, FNMOC NOGAPS Atmospheric Forcing NAVO OP 9/01-3/06
1/8? NCOM 1 15 km ?/z SSH, SST, hydro, FNMOC NOGAPS Atmospheric Forcing NAVO OP 2/06
1/32? NLOM 2 3.5 km Layered SSH, SST, hydro, FNMOC NOGAPS Atmospheric Forcing NAVO OP 3/06
1/12? HYCOM 3,5 7 km ?/?/z SSH, SST, hydro, FNMOC NOGAPS Atmospheric Forcing NAVO 2007
1/25? HYCOM 3.5 km ?/?/z SSH, SST, hydro, FNMOC NOGAPS Atmospheric Forcing NAVO 2012
Near Real-time demonstration Near Real-time demonstration Near Real-time demonstration SSH, SST, hydro, FNMOC NOGAPS Atmospheric Forcing
1/12? Atl. HYCOM 4,5 7 km ?/?/z SSH, SST, hydro, FNMOC NOGAPS Atmospheric Forcing NRL 2002
OP operational 1 High vertical resolution for
mixed layer prediction. Assimilates SSH from NLOM
via T and S synthetic profiles. Web page
http//www.ocean.nrlssc.navy.mil/global_ncom 2
Web page http//www.ocean.nrlssc.navy.mil/global_n
lom 3 Running in real time at NAVO. 4 Under the
National Ocean Partnership Program (NOPP), 1/12?
Atlantic HYCOM demo is running in near
real-time. Includes the Mediterranean Sea. 5
Results at http//www.hycom.org (100Tb LAS server
at FSU)
11
Forecast Verification Statistics from 1/12?
Global HYCOM
1.0
0.9
World Ocean
Gulf Stream
Persian Gulf
0.8
0.7
Median SSH anomaly correlation
1.0
Equatorial Pacific
0.9
Kuroshio
Taiwan
0.8
0.7
0
10
20
30
0.6
0
10
20
30
0
10
20
30
Forecast length (days)
Atmospheric analysis forcing
operational forcing persistence
16 Forecasts included in statistics
(operational forcing reverts toward climatology
at the end of the atmospheric forecast).
12
SST Response in 1/12 Global HYCOM to Hurricanes
Katrina and Rita
HYCOM reproduces the deterministic SST response
to the wind forcing. Implies realistic upwelling
and mixing of subsurface waters as well as
realistic atmospheric wind and heat flux forcing
in HYCOM.
13
ECCO-GODAE

• ECCO-GODAE products have been applied to a wide
  range of research such as ocean circulation and
  relation to climate variability, biogeochemistry,
  geodesy, initialization of climate prediction,
  and providing boundary conditions for regional
  analysis systems.
• The ECCO adjoint tool readily facilitates the
  inclusion of additional data types (e.g.,
  Aquarius SSS), offers an effective mean to
  examine model sensitivity and related physical
  processes and evaluate impact of observations.
• ECCO products satisfy property conservation
  principles, which make it possible to investigate
  the budgets of ocean variables (e.g., mechanisms
  of heat balance).

14
DJF hindcast for March initial conditions
Example of applications Improvement of seasonal
climate forecast by using ECCO-JPL product as
initial state in a coupled model (UCLA atmos.
coupled to MIT-OGCM)
ECCO
baseline
persistence
Anomaly correlation increases with ECCO
ECCO
baseline
Cazes-Boezio, Menemenlis, and Mechoso, 2008 J.
Climate, 21, 1929-1947 .
persistence
Standard error reduces with ECCO
15
Japanese capabilities developed during GODAE

Group System-name
Kyoto Univ. Japan Marine Science Foundation KU-JMSF
Frontier (FRCGC) Kyoto Univ. K-7
Frontier (FRCGC) Tokyo Univ. J-COPE2
Kyushu Univ. RIAMOM
Japan Fisheries Agency Fisheries Research Agency JADE FRA-JCOPE FRA-ROMS
JMA MRI (JMA operation MRI - research.) COMPASS-K MOVE/MRI.COM-WNP
JMA MRI MOVE/MRI.COM-G
16
Group Frontier (FRCGC) Kyoto Univ. JMA MRI
System-name K-7 COMPASS-K MOVE/MRI.COM-WNP
Method Coupled 4DVAR 4DOI, 3DVAR
Aims Climate Pacific-reanalysis (1993-2004) Model improvement. 90s ElNino Ocean Weather Kuroshio, Oyashio, Western North Pacific Variability Predictability Nowcasting-Forecasting of ocean state, oil spill, sea-ice Reanalysis (1985-2007) Japan GODAE server
17
Best Demonstration COMPASS-K (Former
Operational Ocean Assimilation/Prediction System
in Japan Meteorological Agency) Success of
60-day Prediction of the 2004 Kuroshio Large
Meander
Assim/initial state (2004/05/09)
Forecast (2004/06/30)
JMA Japan-GODAE SERVER http//godae.kishou.go.jp/

18
Coupled Data Assimilation improved state
estimation and prediction of ENSO
K-7 results
The long-term ocean reanalysis dataset reproduces
the climate variations, e.g., 1986/1987 and
1997/1998 El Nino including subsurface processes.
The RMSE of the estimated field is reduced by 1/2
comparing with that of simulated one.
Sea surface temperature in Jan1997. (top left)
simulation, (top right) observation, (bottom)
assimilation
NINO3.4 anomaly correlation scores
betweenobservation and forecast as function of
lead time
time
Initialized by coupled data assimilation
Initialized by Oceanic data assimilation
imply capability of 1.5-year lead prediction of
ENSO
Persistence
western equatorial
eastern equatorial
The ensemble forecast from the optimized initial
condition demonstrates much better prediction
almost up to 1.5-year forecast than earlier
results.
Time evolution of heat content anomaly during the
1997/1998 El Nino period with realistic eastward
propagation of Kelvin wave
19
French capability developed during GODAE
Mercator system

• Global Ocean eddy permitting (1/4) forecasting
• 15-day forecast every week
• Including prognostic ice (Artic and Antarctic)
• Multivariate assimilation (using SEEK filter) of
• T/S profiles (Incl ARGO),
• along track altimetry from Jason-1, Envisat and
  GFO
• NCEP RTG SST
• Global ecosystem at 1 (N5P2Z2D2 Pisces model,
  demo mode)
• Enhanced capacity in the North and Tropical
  Atlantic (lat gt 20S) and the Mediterranean sea
• Eddy resolving horiz. res. (1/12)
• Daily update of 1-week forecast
• Global eddy permitting (1/4) reanalysis (20-year
  period, under dev.)

20
Impact of altimetry on analysis and forecast
accuracy

• Results from an Observing System Experiment (OSE)
  using
• North Atlantic Med eddy resolving system
  (1/15)
• Multivariate assimilation of SLASSTT/S profiles
• 1 year integration (2004)

Benkiran, 2008
SLA RMSDIFF No altimetry Jason only Jason Envisat JasonEnvisatGFO JasonEnvisatGFOT/P
7-day Forecast (cm) 10,27 9,67 8,95 8,62
Nowcast (cm) 9,15 8,36 7,50 7,08
Hindcast (cm) 12,94 8,38 7,07 6,18 5,63
Accuracy of hindcast with Jason only requires 1
additional satellite in Nowcast Accuracy of
hindcast with JasonEnvisat requires 2 additional
satellites in Nowcast Accuracy of hindcast with
Jason only can be reached in 7-day forecast with
4 satellites Without altimetry, no mesoscale
prediction skill
21
Tropical Cyclone MonitoringExample RAMASSUN
(May 2008)
Sea surface temperature
Prediction of cyclone intensity from Ocean near
surface heat content
Bulletin significant intensification in the next
48 hr significant weakening north of 20N
Depth of 26oC isotherm
Sea surface height anomaly
Buarque, 2008
22
My Ocean

• will deliver regular and systematic reference
  information (processed observations, hindcasts
  and forecasts) on the state of the oceans and
  regional seas
• at the resolution required by intermediate users
  downstream service providers, of known quality
  and accuracy,
• for the global and European regional seas.

• Physical state of the ocean, and primary
  ecosystem
• For global ocean, and main European basins and
  seas
• Large and basin scale mesoscale physics
• Hindcast, Nowcast, Forecast
• Data, Assimilation and Models

23
UK capability developed during GODAE

• Existing system
• global (1o) with 1/3o and 1/9o models in
  Atlantic, Indian
• daily 5-day forecasts made by operational suite
• multivariate assimilation of profile, SST,
  sea-ice altimeter data
• Two-scale error covariances and correction of
  biases
• Separate shelf sea forecasting system (POLCOMS
  ERSEM)
• Revised system (implementation by Oct 2008)
• based on NEMO in collaboration with NCOF and My
  Ocean
• 1/4o global and 1/12o for N Atlantic Indian

24
Control of ocean mesoscale by altimeter data
impact of number of satellites
Jason only
None
Jason, Envisat GFO
Jason Envisat
25
Daily mean RMS Errors in the North Atlantic near
surface
Results from 3-D Twin ExperimentsHemmings,
Barciela, Bell JMS (2008)

Phytoplankton (mmolN/m3)
Zooplankton (mmolN/m3)
Control - truth
Assimilation - truth
Detritus (mmolN/m3)
Nutrients (mmolN/m3)
26
Canadian Coast Guard systemfor Search and rescue

• 4 person life raft

OLD
Observation Model
Improvement in search and rescue performance by using operational oceanography products (Mercator) Err m/hr improvement
4 pers life-raft No Drogue 1 512 4
7 pers life raft No Drogue 1 509 9
4 pers life raft DROGUE 1 201 67
7 pers life raft DROGUE 2 313 51
NEW
72 hrs
Davidson et al.
27
Satellite measurements of SST
Satellite surface temperature data from
GHRSST Routine production of scientifically
qualified products
N-17/18 AVHRR GAC (9km)
GOES-E/W (5km)
AMSRE (25/12km)
N-17/18 AVHRR LAC (1km)
AATSR (1km)
MSG (5/10km)
28
Intercomparisons

• A standard set of diagnostics for
  inter-comparison of model outputs has been
  developed (originally led by Christian le
  Provost)
• Procedure for calculation of comparable root mean
  square differences between model outputs and
  observations recently established

Initial FOAM-NEMO
Mercator v2 NAtl
World Ocean Atlas
29
The GODAE IMBER initiative (Ecosystem
modelling)
At the first GODAE-IMBER Meeting at CNRS
(12-13 June 2007) a new working group has been
formed to facilitate dialogue between those
developing new ecosystem models and the
developers of the operational systems.

• Particular emphasis is given to
• Improvement of present GODAE systems for IMBER
• - ocean analysis / reanalysis products
• - intercomparisons (global regional)
• - forcing fields
• Improvement of the observing system
• - new sensors/measurements
• - joint physical/biological data
  assimilation
• Use of GODAE products by IMBER
• - bio-ecosystem reanalysis
• - Multi model ensembles

• Areas of importance
• Ecosystem modelling data assimilation
• - Schemes for assimilation of
• biogeochemical data are under
• development
• - Current assimilation schemes degrade
• the biogeochemistry
• - Overall high horizontal and vertical
• resolution models for the upper ocean
• are needed.
• - Advanced schemes for a finer vertical
• structure are a key issue for nutrient
• transport.
• Interaction with coastal and shelf seas
• systems
• Support for B-Argo (see friends of Oxygen
• on Argo)
• Reanalysis

Meeting outcomes available on GODAE WWW site
30
The GODAE Coastal and Shelf Seas Working Group
(CSSWG) mission and activities

• The usefulness of GODAE systems to coastal and
  shelf seas forecasting is one of the measures of
  the success of the GODAE project.
• The GODAE CSSWG conducts actions to promote the
  assessment and demonstration of the value of
  GODAE results for regional, coastal and shelf
  seas models and forecasting systems

• Continuously updated census of coastal projects
  worldwide using GODAE products
• CSSWG White Paper (79 pages, 3 editions so far)
• Web site www.godae.org/CSSWG.html
• 2007 Liverpool workshop, 26 institutes from
  US/Can, Europe, Japan and Australia

31
2007 Liverpool Workshop

• Some overarching questions
• Sensitivity of coastal ocean systems to
  i.c.s/b.c.s/remote forcing
• Best type of GODAE product for one particular
  type of application
• Does downscaling add value to GODAE products?
  Does using GODAE b.c.s improve forecasts in the
  coastal products?
• Does assimilation of local observations add
  further value to coastal products? What
  observations would add value?
• Types of metrics
• Scientific value impact assessment
• Final value utility

Three examples from workshop
32
Objectives
GODAE-OOPC OSSE/OSE Workshop Paris, November 5-7
2007

• Review work done on impact studies, OSEs and
  OSSEs over the past years
• Identify robust and common features
• Understand differences in assimilation systems
• Provide good examples of the contribution of
  observing systems
• Provide preliminary recommendations on observing
  system design
• Define and agree on activities to be carried out
  before the GODAE final symposium
• Develop new ideas on the way to assess and design
  the ocean observing systems and propose new
  experiments to be carried out in the coming years
  

33
GODAE OSE/OSSE workshop Outcomes

• Very good talks and discussion. New ideas.
• OSE/OSSE in the ocean are still at the early
  stage. Science technical issues
• High resolution/mesoscale versus large
  scale/climate
• What needs to be optimized (SST, surface
  currents, full ocean state, analysis or forecast)
  ?
• - Good demonstrations of impact of Argo and
  altimetry (high resolution) exist. To be
  developed and extended to other elements of the
  global ocean observing system.
• Efforts need to be organized.
• Experiences need to be shared
• International collaboration required Long term
  activity
• Setting up of a Working Group in 2008 strong
  links with the intercomparison WG. Synthesis
  paper / final symposium

34
GODAE Final Symposium 10-15 November 2008, Nice,
France

• The revolution in global ocean forecasting Ten
  years of progress in operational oceanography
• Invited review papers on main themes
• Operational oceanography infrastructure current
  status
• Demonstrations of feasibility
• Key scientific technological advances
• Applications
• Coordination post GODAE
• Highlights to appear in a special issue of
  Oceanography

35
Coordination post-GODAE

• An Expert Team for Operational Ocean Forecasting
  Systems is being set up within the JCOMM Services
  Programme area for technical coordination of
  standards
• The most pressing societal issues to which ocean
  forecasting systems should contribute have become
  very diverse (coastal, ecosystem, coupled,
  decadal)
• Duplication of international coordination (e.g.
  by Clivar) should be avoided
• Funding and stamina for international
  coordination are limited

36
Coordination post-GODAE (2)

• What is the central activity which a post-GODAE
  group should coordinate ?
• Development of operational systems for ocean or
  coupled model data assimilation
• Demonstrations of impact and value of observing
  systems on forecast systems
• How should links to other groups be pursued ?
• Formal representatives to or from other groups
• Limited life working groups