GODAE OSE workshop: Towards Routine Monitoring

1
GODAE OSE workshopTowards Routine Monitoring

• Peter R. Oke1, Gilles Larnicol2, Kirsten
  Wilmer-Becker3, Corinne Guiose2
• 1CSIRO, 2CLS, 3UK Met Office

2
Workshop Objectives

• Key question How can we use GODAE systems for
  routine monitoring of the GOOS?
• Key objective Establish a plan for coordinated
  routine monitoring of the GOOS using GODAE
  systems
• - what will we do?
• - how will we do it?
• - when will we do it? how regularly?

3
Workshop Objectives

• Scope
• physical, biological, ocean-atmosphere ?
• global, regional, coastal ?
• what is required for each observing system from
  GODAE ?
• Relationships
• GOV OSE TT ?? ET-OOFS
• GOV OSE TT ?? GOV Inter-comparison
• Appropriate experiments
• Conventional OSEs/OSSEs
• analysis / forecast sensitivity
• assimilation diagnostics (obs-space)
• other
• Method of delivery
• Central host vs distributed web pages
• ET-OOFS JCOMM WMO ?

Perhaps for GOV meeting ??
4
We must learn from NWP activities

• The objectives of the THORPEX Data Assimilation
  and Observation Strategy
• Working Group (DAOS-WG) are two-fold
• to assess the impact of observations and various
  targeting methods to provide guidance for
  observation campaigns and for the configuration
  of the Global Observing System.
• to setup an optimal framework for data
  assimilation, including aspects such as targeted
  observations, satellite data, background error
  covariances and quality control.

5
The GOOS is ever-changing
6
Why routine monitoring?

• Politically
• GODAE systems critically depend on the
  availability of ocean observations but most
  observing systems are not maintained for GODAE
  activities they are primarily maintained for
  climate monitoring.
• To maintain/acquire relevance and influence,
  GODAE must proactively contribute to GOOS
  planning/discussions.
• GODAE must contribute positively and not
  contradict, or compete with our more
  climate-centric colleagues.

7
Why routine monitoring?

• Scientifically
• Contributing to the ongoing assessment and design
  of the GOOS
• where are the gaps developing?
• what instruments are looking suspect?
• where is there redundancy?
• what additional observations would be beneficial?
• Getting the most out of observations
• does my system over- or under-fit observations?
• how does my system compare to other GODAE
  systems?
• how can I get more information from observations?
• refined obs-error/length-scale estimates
• better assimilation procedures
• Helping each other deliver the best
  forecast/analysis products
• shared experiences, open communication, positive
  engagement

8
What does routine monitoring mean?

• Regular (eg, weekly/monthly/quarterly) evaluation
  of the impact/importance of each observation and
  each component of the GOOS on GODAE systems.
• Help to provide illustrations/arguments/requiremen
  ts for short term political/technical decision ?
  Could we anticipate some request ?

9
How could we perform routine monitoring?

• Conventional Observing System Experiments (OSEs)
• are probably too expensive to perform routinely
• both computationally and in their
  analysis/understanding
• are not always straightforward to interpret
• denial of obs A impacts influence of obs B
• Analysis sensitivity
• readily applied to all assimilation/mapping
  systems
• when we perturb obs A how much did the analysis
  change?
• Forecast sensitivity
• readily applied to adjoint-based assimilation
  systems
• when we perturb obs A how much did the forecast
  change?
• Red-flagged data (from QC systems)
• Single point assimilation
• Routine inter-comparisons of diagnostics
• quarterly/monthly/weekly

10
Application of methods in oceanography

• OSEs
• Balmaseda et al., Benkiran et al., Oke et al.,
  Pascual et al., Vidard et al., Shulman et al.,
• OSSEs
• Brassington et al., Guinehut et al., Hackert et
  al., Oke et al., Sakov et al., Schiller et al.,
  Tranchant et al., Vecchi et al.,
• Forecast sensitivity
• Fujii et al., Losch et al., Moore et al., Weaver
  et al.,
• Analysis sensitivity
• ???

11
Conventional methods OSEs
Assimilated and with-held observations
Assimilated observations
Evaluation/ Validation
Forecast or BGF
Analysis or Forecast
T/S

• Assimilate real observations
• Systematically with-hold observation types

12
Conventional methods OSSEs
Simulated observations
Assimilated observations
Evaluation/ Validation
Forecast or BGF
Analysis or Forecast

• Assimilate pretend observations
• from a model
• Systematically include different observation
  types
• including future observation types

13
NWP analysis sensitivity
40

• Information content, i.e. degrees of freedom for
  signal (DFS), for the main data types in the
  assimilation at the ECMWF centre, in percentage.
  blue globe cyan N Pole white S Pole. The
  sum of all percentages for the globe 100,
  representing the whole contribution from
  observations. (Courtesy of C. Cardinali)

14
NWP forecast sensitivity

• Adjoint-based sensitivity to observations over
  the Southern Hemisphere in the Canadian system
  for (a) 3-D-Var, (b) 4-D-Var. The computation
  corresponds to the sensitivity of the total
  energy of the 24-h forecast error with respect to
  the observations. Negative values indicate that
  the observations contribute to an improvement in
  the forecast. (Courtesy of S. Pellerin,
  Environment Canada)

15
NWP forecast sensitivity
Many moderately beneficial Radiosonde impacts in
CONUS and Europe best outcome
criteria Langland and Baker (2004)
16
Single point assimilation
17
Quality control metrics

• Comparison of innovation statistics between GODAE
  systems
• Obs minus Background
• look for consistently un-agreeable obs
• Obs minus Analysis
• look for consistently under-fitted obs

18
Method of delivery
Example page only
19
Agenda

• Session 1 Overview of observing system
  components and activities
• - status, plans, progress etc
• Rapporteurs Belbeoch Brassington
• Session 2 Assimilation diagnostics and metrics
• - what do we do already and how could we take
  advantage of that for OSE activities?
• Rapporteurs Drevillon Lea
• Session 3 OSE/OSSE activities
• - current research activities
• Rapporteurs Bertino Fukumori
• Session 4 How to move forward?
• - what, how and when?
• Rapporteurs Larnicol Oke

20
Agenda day 1, Thursday, 4 June

• 0900 Oke/Larnicol - Welcome and workshop
  objectives
• Session 1 Introduction and Overview of observing
  system components 0930 Schiller/Dombrowsky -
  Status of GODAE OceanView
• 0945 Brassington Status of JCOMM ET-OOFS
• 1015 Lambin/Dibarboure - Satellite altimetry
  status and products
• 1045 Belbeoch/Viola - JCOMM-OBS international
  coordination and Argo and other in situ programs
  status
• 1115 Coffee break
• 1130 Cummings - Forecast sensitivity towards
  routine monitoring of the GOOS at NRL
• 1145 Rabier - Lessons learned from THORPEX
• 1230 Lunch

• Session 2 Assimilation diagnostics and metrics
• 1330 Weaver - Ensemble/Var diagnostics
• 1350 Cummings - NRL
• 1410 Drevillon - Mercator-Ocean
• 1430 Lea - UK Met Office 
• 1450 Brassington/Oke - BLUElink
• 1510 Bertino - TOPAZ
• 1530 Coffee break
• 1600 Fujii - JMA
• 1620 Larnicol/Guinehut CLS
• 1640 Weaver/Oke - Discussion
• What diagnostics are could be fed back to
  observational groups to support design,
  maintenance and justification for the GOOS? Can
  we coordinate our routine activities to help
  support each other and help improve our systems?
• 1800 End of Day 1

21
Agenda day 2, Friday, 5 June

• Session 3 General Contributions
• 0900 Fujii - OSE experiments using the JMA ENSO
  forecasting system
• 0920 Bertino - Evaluating the assimilation of
  Ferrybox data between Norway and Denmark
• 0940 Brassington - Impact of SSS on a
  multivariate assimilation system
• 1000 Rio - Impact of GOCE for modelling centers
  status of GOCINO
• 1020 Dibarboure - Future altimetry design from
  impact studies to operational metrics or the
  reverse ?
• 1040 Coffee break
• 1100 Oke - Potential impact of HF radar and
  gliders on ocean forecast system
• 1120 Remy - Sensitivity studies within GLORYS
  project (title to be confirmed)

• 1140 Tranchant - Multivariate data assimilation
  (SAM2) of Simulated (SMOS and Aquarius) SSS in a
  1/3 Atlantic ocean model (MNATL)".
• 1200 Oke/Brassington - GODAE inter-comparisons
  around Australia
• 1230 Lunch
• Session 4 How to move forward?
• 1330 Discussion
• - Currently available assimilation and QC metrics
  
• - Proposal for routine observing system
  assessment
• - International coordination
• - Options for engaging with observational
  community
• 1530 Coffee break
• 1600 Action - Collating information and
  disseminating
• 1630 End of Day 2

22
(No Transcript)
23
Analysis sensitivity of Argo T over top 200 m
depth

• 5-different realisations of the self-sensitivity
  (HKii)
• Averaged over top 200 m depth
• Shows some dependence on the perturbations to
  observations

24
Analysis sensitivity of Argo T over top 200 m
depth
Average
Standard error
Influence

• If HKii std err gt0.5 ? High
• If HKii std err gt0 lt 0.5 ? Med
• If HKii std err lt0 ? Low (?)

25
Analysis sensitivity of Argo T over top 200 m
depth
26
Analysis sensitivity of Argo S over top 200 m
depth
27
Analysis sensitivity of Argo S over top 200 m
depth
Average
Standard error
Influence

• If HKii std err gt0.5 ? High
• If HKii std err gt0 lt 0.5 ? Med
• If HKii std err lt0 ? Low (?)

28
Analysis sensitivity of Argo S over top 200 m
depth
29
Adaptive sampling singular vectors and ETKF

• NWP example of leading singular vectors (fastest
  growing modes) from Meto-France, ECMWF, UK Met
  Office and NCEP

30
Conventional OSEs
Estimated SLA Errors

• 1/10o Bluelink system
• 6-month long OSEs starting December 2005

SST and Argo partially compensate for no ALTIM
but ALTIM is clearly necessary to represent the
mesoscale
Oke, P. R., and A. Schiller (2007) Impact of
Argo, SST and altimeter data on an eddy-resolving
ocean reanalysis. Geophys. Res. Lett., 34,
L19601, doi10.1029/2007GL031549.
0 8 16
24 RMS residuals (cm)
31
Conventional OSEs
Estimated SST Errors

• 1/10o Bluelink system
• 6-month long OSEs starting December 2005

Argo partially compensates for no SST but not
over wide shelfs and shallow seas
Oke, P. R., and A. Schiller (2007) Impact of
Argo, SST and altimeter data on an eddy-resolving
ocean reanalysis. Geophys. Res. Lett., 34,
L19601, doi10.1029/2007GL031549.