Upload
brett-woods
View
213
Download
0
Embed Size (px)
Citation preview
www.cmar.csiro.au/staff/oke/
The end users perspective – who are we doing this for? (e.g., climate science and data assimilation efforts)
Peter Oke et alPeter Oke et al
CAWCR, CSIRO Marine and Atmospheric Research
June 2013
www.cmar.csiro.au/staff/oke/
Talk outline
What does the data assimilation community need from the observational community?
What can the data assimilation community do for the observational community?
GODAE OceanView
RTQC
Example
www.cmar.csiro.au/staff/oke/
What does the data assimilation community need from the observational community?
Observations delivered in NRT:
QC information would be used
Error estimates:
Known instrument errors the standard deviation of errors
Unbiased observations no systematic errors
www.cmar.csiro.au/staff/oke/
What can the data assimilation community do for the observational community?
Demonstrate impact of data:
Operational forecasts short-range (7-d), seasonal, …
Quantify the benefits to marine industry/applications:
Search & rescue; oil spill, fisheries, shipping, military support, …
Help identify quality control issues:
Feedback on what data are being excluded, and what data are developing systematic errors (bias)
Help identify emerging gaps in the GOOS
Help plan for future observation strategies
www.cmar.csiro.au/staff/oke/
GODAE OceanView (2009- )
Mission: Develop capabilities in operational ocean forecasting
Five Task Teams
Coastal Ocean and Shelf Seas
Inter-comparison and Validation
Marine ecosystem and prediction
Observing System Evaluation (OSEval)
Short- to medium-range coupled prediction
GODAE OceanView website: https://www.godae-oceanview.org/
www.cmar.csiro.au/staff/oke/
GOV OSEVal-TT organisation
Co-Chairs:
Peter Oke (CSIRO)
Gilles Larnicol (CLS)
Core Members:
Magdalena Balmaseda (ECMWF)
Laurent Bertino (NERSC)
Gary Brassington (BoM)
Jim Cummings (NRL)
Yosuke Fujii (JMA/MRI)
Pat Hogan (NRL)
Villy Kourafalou (Univ. Miami)
Daniel Lea (UKMet)
Matthew Martin (UKMet)
Avichal Mehra (NOAA)
Pavel Sakov (NERSC)
Anthony Weaver (CERFACS)
Associate members:
Mike Bell (UKMet)
Eric Dombrowsky (Mercator)
Fabrice Hernandez (Mercator)
Eric Lindstrom (NASA)
Andreas Schiller (CSIRO)
www.cmar.csiro.au/staff/oke/
Responses to “observing system events”
Continuation of Jason-1 data processing in inter-leaved orbit (June 2009)
UKMet and BoM provided a demonstration of the impact of Jason-1 data in inter-leaved orbit during recent outages
Contributed by G. Brassington, BoM
Mo
de
l-o
bs
mis
-fit
Many GODAE contributions to observing system evaluation
have been ad hoc
www.cmar.csiro.au/staff/oke/
Operational community needs a coordinated plan to respond to “observing system events”
www.cmar.csiro.au/staff/oke/
NRT OSEs
Routinely run parallel forecast at operational centers and with-hold a different data each month:
Quantify the impact of each data type on forecasts
Multi-system approach
Feb 2011Mar 2011Apr 2011May 2011Jun 2011Jul 2011
XBTTAOJason-2All altimsSSTArgo
www.cmar.csiro.au/staff/oke/
Provision of Observation Impact Statements (OISs)
10
www.cmar.csiro.au/staff/oke/
Provision of Observation Impact Statements (OISs)
www.cmar.csiro.au/staff/oke/
Inter-comparison participants: BoM, Coriolis, MyOcean, FNMOC, UKmet
www.cmar.csiro.au/staff/oke/
QC inter-comparison: Recall
Temperature
Salinity
Pressure
Recall:
A measure of success
Recall = 1: is perfect
RTQC could be useful to the obs community for identifying bad floats
www.cmar.csiro.au/staff/oke/
Inter-comparisons of intermediate-resolution reanalyses
Xue et al.
www.cmar.csiro.au/staff/oke/
In Situ Observations
Contributed by Y. Xue, NOAA/NCEP – from Saha et al. (2010)
www.cmar.csiro.au/staff/oke/
1993
Contributed by Y. Xue, NOAA/NCEP
Inter-comparison of CLIVAR systems:HC300 in Equatorial Pacific (2oS-2oN)
www.cmar.csiro.au/staff/oke/
Inter-comparison of CLIVAR systems:HC300 in Equatorial Atlantic (2oS-2oN)
2005
2005
Contributed by Y. Xue, NOAA/NCEP
www.cmar.csiro.au/staff/oke/
Conclusion
Active operational ocean forecasting community that it dependent on the obs community
Data availability in NRT is important
Each forecast center undertakes RTQC that could be useful for DMQC
GODAE OceanView is motivated to “support” the obs community by demonstrating impact
Historically been ad-hoc;
Plans to make it more organised (NRT OSEs and OIS).
Many DM OSE studies have been published that demonstrate impact
www.cmar.csiro.au/staff/oke/
Future opportunities
Help GODAE OceanView figure out how to disseminate/condense technical metrics into something that is meaningful
OSE study using DMQC-ed and RTQC-ed data – what is the benefit to a data assimilating model?
www.cmar.csiro.au/staff/oke/
• how we can help them (e.g., what is important for their data assimilation efforts) and in what they can help us?
• to cover the user needs from short term & longer term forecasts/predictions & climate hindcasts, and in what they also might be able to help us.
• need for gridded data to initialise models (e.g., decadal predictions)? which fields? what are the requirements?
• how observational fields/observations might be used to evaluate data assimilation efforts? what would be required?
• how poor data quality has impacted /might impact on data assimilation efforts? or lack of uncertainties?
• can assimilation efforts somehow contribute back to the QC system? (e.g., pointing to erroneous data).
Send to:
Jim Carton - [email protected]
Keith Haines - [email protected] GSOP Co-Chair
Detlef Stammer - [email protected]
Yosuke Fujii - [email protected] ocean/seasonal forecasting
Matt Martin - [email protected] Ocean Forecasting
Jim Cummings - [email protected] ocean forecasting
Gary Brassington - [email protected] ocean forecasting
Laurent Bertino - [email protected] (Norwegian ocean forecasting)
Tony Lee - [email protected]
www.cmar.csiro.au/staff/oke/
Example of an OSE (with-holding XBT)
OSEs using HYCOM after the DWH Oil spill to assess the impact of XBT data
Halliwell et al. (NRL & NOAA)
www.cmar.csiro.au/staff/oke/ Contributed by G. Halliwell, NOAA/AOML/PhOD
www.cmar.csiro.au/staff/oke/
Impact of P-3 Observations on Ocean Analyses
Collaboration between AOML and NRL-Stennis
NRL ran two experiments with the 1/25° regional HYCOM:
1. Assimilate all observations
2. Deny only the P3 observations
Critical issues affecting this evaluation:
Results depend on choices of model and DA scheme
Impact of update cycle
Impact of relative weighting of synthetic T,S profiles derived from altimetry vs. in-situ T,S profiles
Contributed by G. Halliwell, NOAA/AOML/PhOD
www.cmar.csiro.au/staff/oke/
RED: With P3 assimilation
BLACK: No P3 assimilation
Contributed by G. Halliwell, NOAA/AOML/PhOD
No assimilation ~ 4-5 degrees
No assimilation < 0.5
No assimilation +- 1 degree
www.cmar.csiro.au/staff/oke/
Experiment Bias (°C) RMS Diff. (°C)
Skill Score
P-3 Profiles Assimilated
-1.11 1.41 0.88
P-3 Profiles Denied
-1.18 1.79 0.84
No Data Assimilation
-0.40 4.5 0.31
Error Analysis, Nancy Foster T Profiles, 9 July
Experiment Bias (m) RMS Diff. (m)
Skill Score
P-3 Profiles Assimilated
-21.1 35.8 0.09
P-3 Profiles Denied
-24.3 44.3 < 0
No Data Assimilation
19.3 89.5 < 0
20°C isotherm depth
Temperature, 30 – 360 m
Contributed by G. Halliwell, NOAA/AOML/PhOD
8-10 July
www.cmar.csiro.au/staff/oke/
Inter-comparisons of intermediate-resolution reanalyses
Xue et al.
www.cmar.csiro.au/staff/oke/
In Situ Observations
Contributed by Y. Xue, NOAA/NCEP – from Saha et al. (2010)
www.cmar.csiro.au/staff/oke/
1993
Contributed by Y. Xue, NOAA/NCEP
Inter-comparison of CLIVAR systems:HC300 in Equatorial Pacific (2oS-2oN)
www.cmar.csiro.au/staff/oke/
Inter-comparison of CLIVAR systems:HC300 in Equatorial Atlantic (2oS-2oN)
2005
2005
Contributed by Y. Xue, NOAA/NCEP
www.cmar.csiro.au/staff/oke/
Evaluating options for altimeter constellationsLarnicol et al.
www.cmar.csiro.au/staff/oke/
Altimeter constellations
3 x Nadir 1x SWOT11 x Nadir
(Iridium 6 + Jason-CS +GFO2+ HYC+ S3A + S3B)
1 x SWOT + 11 x Nadir
2 x SWOT
Contributed by G. Larnicol, CLS
Reconstruction error (% of reality signal variance) for
geostrophic U and V
www.cmar.csiro.au/staff/oke/
OSEs using JMA/MRI seasonal prediction system
Fujii et al.
www.cmar.csiro.au/staff/oke/
Observing System Experiments (OSEs) using JMA/MRI system
Impact of TAO data decreases, and Argo data increases, as the number of Argo floats increases
Difference when TAO/TRITON data are with-
held
Difference when Argo data are with-held
Contributed by Y. Fujjii, JMA/MRI
www.cmar.csiro.au/staff/oke/
Impact of Argo and TAO data on JMA forecast skill
With-holding Argo data degrades the skill of forecasts over 8-13 months by almost 25% in the Pacific Ocean
With-holding TAO data degrades the skill of forecasts over 1-7 months by almost 15% in the Indian Ocean.
Contributed by Y. Fujjii, JMA/MRI
www.cmar.csiro.au/staff/oke/
Oke, P., and P. Sakov: Design and Assessment of the Australian Integrated Marine Observing System
Simple method to assess the potential impact of data from moorings
www.cmar.csiro.au/staff/oke/
Footprint of individual moorings
Cabbage Patch Mooring
Deep Slope Mooring