Comparison of GMAO and GFDL global ocean assimilation systems in the ODASI experiments

1
Comparison of GMAO and GFDL global ocean
assimilation systems in the ODASI experiments

• Chaojiao Sun
• NASA/GSFC Global Modeling and Assimilation Office
  (GMAO)
• UMBC/GEST

People who contributed to this work
Michele Rienecker, Christian Keppenne, Jossy
Jacob, Anna Borovikov, Nicole Kurkowski,
Robin Kovach (GMAO) Tony Rosati, Andrew
Wittenburg, Matthew Harrison (GFDL) NCEP EMC
Seminar December 14, 2004
2

• Outline
• ODASI
• GFDL and GMAO ocean data assimilation systems
  and experiments
• Comparison independent validation of analyses
• - Comparison of T(z) with TAO (not
  independent data)
• - Independent validation of U(z) using TAO
  ADCPs
• - Independent validation of T(z), S(z),
  U(z) using TAO servicing cruises (Greg Johnson,
  PMEL)
• Surface current seasonal climatology (as compared
  to OSCAR Reverdin)
• Sensitivity of salinity treatment
• Case study TIWE for validation of U(z) to
  deeper levels
• Constraining S(z)
• Independent validation of salinity with TAO

3
CDEP Consortium Ocean Data Assimilation
Consortium for Seasonal-to-Interannual Prediction
(ODASI)
COLA, GFDL, IRI, LDEO, NCEP, GMAO
GMAO Michele Rienecker Chaojiao Sun Jossy
Jacob Nicole Kurkowski Robin Kovach Anna Borovikov
GFDL Tony Rosati Matt Harrison Andrew Wittenberg
COLA Jim Kinter Ed Schneider Ben Kirtman Bohua
Huang
NCEP Dave Behringer
IRI Steve Zebiak Eli Galanti Michael Tippett
LDEO Alexey Kaplan Dake Chen
http//nsipp.gsfc.nasa.gov/ODASI
4

• ODASI Themes
• ODA product intercomparisons (models,
  assimilation methodologies, assimilation
  parameters) using a common forcing data set and
  common QCd in situ data streams
• Models MOM4, MOM3, Poseidon, Cane-Patton,
  LDEO4
• Methodologies 3DVAR, OI, EnKF, Reduced state KF
  and optimal smoother, bias correction strategies
• Coupled Forecast Sytems CGCMs, Hybrid models,
  Intermediate models
• Development of observational data streams
• Validation of assimilation products in forecast
  experiments
• 4. Observing system impacts - focused on TAO
• TAO array was established for S-I forecasting.
• Is it effective in its present configuration?
• Could it be modified to provide better support
  for S-I forecasts?
• what is its role c.f. other elements of the
  ocean observing system?

5

• ODASI observations
• XBT (QC by NCEP/Dave Behringer)
• TAO/TRIRON/PIRATA (delayed mode, QC by PMEL)
• Argo (delayed mode, QC by GODAE/Monterey server)
• ODASI forcing
• NCEP CDAS forcing with wind stress
  climatology replaced by Atlas/SSMI analyses
  (produced by GFDL), Reynolds SST
• GMAO forcing (referred to as XBT03 forcing in
  the following graphs)
• - Atlas/SSMI time varying wind stress
• - GPCP monthly mean precipitation
• - NCEP CDAS SW (for penetrating radiation)
  LH (for evaporation)
• - Reynolds SST

6
ODASI observations TAO/TRITON/PIRATA XBTs
ARGO
7

• GFDL GMAO models and assimilation schemes
• GFDL MOM3
• Derber-Rosati 3D-Var
• GMAO Poseidon
• OI, MvOI, EnKF (results shown here only uses
  OI)
• - Two GMAO schemes treat salinity
  differently
• T scheme T assimilation only, S updates
  based on model T-S relationship
  (Troccoli and Haines, 1999)
• TS scheme T and S assimilation (S comes from
  ARGO when available, otherwise from
  Levitus climatology)
• GFDL GMAO forcings
• GFDL forcing ODASI forcing
• GMAO forcing
• - ODASI forcing
• - GMAO forcing (XBT03)
• - Experiment duration 10 years (Jan 1993 to
  Dec 2002)
• - Assimilation interval 5 days data window
  10 days
• - IAU implemented

8
ODASI ?x
XBT03 ?x
XBT03 E-P
ODASI E-P
9

• Comparison and independent validation of
  analyses
• Comparison of T(z) with TAO temperature (not
  independent data)
• Comparison of U(z) with moored TAO ADCPs
• Comparison of T(z), S(z), U(z) with TAO servicing
  cruises

10
Blue line mean Red line STD
T(z) at 140W
11

• Comparison and independent validation of
  analyses
• Comparison of T(z) with TAO temperature (not
  independent data)
• Comparison of U(z) with moored TAO ADCPs
• Comparison of T(z), S(z), U(z) with TAO servicing
  cruises

12
U(z) at 140W
Blue line mean Red line STD
13
U(z) at 110W
Blue line mean Red line STD
14

• Comparison and independent validation of
  analyses
• Comparison of T(z) with TAO temperature (not
  independent data)
• Comparison of U(z) with moored TAO ADCPs
• Comparison of T(z), S(z), U(z) with TAO servicing
  cruises

15
Meridional sections of T, U, S (from TAO
servicing cruise)
T
S
U
165 E (7/1998)
125 W (10/1997)
T
T
125W
165E
U
U
16
Nino-4
Nino-3
Nino-3
Nino-4
T
S
0-5S
0-5N
U
RMS of T, S, U as compared TAO servicing cruises
data from Greg Johnson
(7/96-12/98)
17

• Outline
• ODASI
• GFDL and GMAO ocean data assimilation systems
  and experiments
• Comparison independent validation of analyses
• - Comparison of T(z) with TAO (not
  independent data)
• - Independent validation of U(z) using TAO
  ADCPs
• - Independent validation of T(z), S(z),
  U(z) using TAO servicing cruises (Greg Johnson,
  PMEL)
• Surface current seasonal climatology (as compared
  to OSCAR Reverdin)
• Sensitivity of salinity treatment
• Case study TIWE (deeper validation of U(z))
• Constraining S(z)
• Independent validation of salinity with TAO

18
Seasonal climatology of surface zonal current (U)
Summer
Fall
Winter
Spring
19

• Outline
• ODASI
• GFDL and GMAO ocean data assimilation systems
  and experiments
• Comparison independent validation of analyses
• - Comparison of T(z) with TAO (not
  independent data)
• - Independent validation of U(z) using TAO
  ADCPs
• - Independent validation of T(z), S(z),
  U(z) using TAO servicing cruises (Greg Johnson,
  PMEL)
• Surface current seasonal climatology (as compared
  to OSCAR Reverdin)
• Sensitivity of salinity treatment
• Case study TIWE (deeper validation of U(z))
• Constraining S(z)
• Independent validation of salinity with TAO

20
Treatment of salinity matters A case study from
the TIWE cruise - a detailed look at salinity
treatment constraining S(z) - validation of
U(z) time series to deeper levels T scheme
Troccoli Haines (1999) relies on model to
maintain ltT- Sgt, but this can drift TS
scheme When Argo is not available
Synthetic S(z) profile from T(z) using monthly
Levitus climatology assimilating
climatology R ratio of observed salinity
error over model salinity error (std) R 1, the
obs. S is given the same weight as the model S R
2, more weight given to model R 4, even more
weight given to model T scheme gives model S
total confidence ( R 8 )
21
Background conditions during TIWE
?x
SST
U from TAO mooring
July 1990
July 1991
Jan
Jan
(Lien et al., 1995)
22
XBT03-T
ODASI-T
XBT03-TS (R4)
ODASI-TS (R4)
XBT03-TS (R2)
ODASI-TS (R2)
Shipboard ADCPs
23

• Outline
• ODASI
• GFDL and GMAO ocean data assimilation systems
  and experiments
• Comparison independent validation of analyses
• - Comparison of T(z) with TAO (not
  independent data)
• - Independent validation of U(z) using TAO
  ADCPs
• - Independent validation of T(z), S(z),
  U(z) using TAO servicing cruises (Greg Johnson,
  PMEL)
• Surface current seasonal climatology (as compared
  to OSCAR Reverdin)
• Sensitivity of salinity treatment
• Case study TIWE (deeper validation of U(z))
• Constraining S(z)
• Independent validation of salinity with TAO

24
Salinity comparisons Salinity time series from
XBT03-T and TAO moorings at 156E
25
Salinity comparisons Salinity time series from
XBT03-T (red) and TAO moorings at 165E
26
Conclusions weve learned a lot about the
working of different assimilation systems 1.
GMAO and GFDL temperature analysis is consistent
with TAO mooring data (not independent data)
slight impact from salinity correction 2.
Subsurface current structure independent
validation - GMAO compared well with TAO ADCPs
(zonal structure on the EQ) GFDL compares well
with Johnsons data (meridional structure) -
sensitive to treatment of salinity TS appears
to be effective 3. TS significantly improves
subsurface salinity structure time series of
salinity analysis agrees with TAO mooring data,
but model bias is present in T scheme Ongoing
work treat surface salinity properly in TS
tune observation error for synthetic
salinity investigate the impact of ARGO
assimilation
27

• TIWE issues
• Vertical resolution does it make sense to
  compare the structure below 200m?
• What is the best measure to decide the
  effectiveness of the scheme how to tell which
  scheme is better?
• Daily comparison may not be the best comparison,
  since the assimilation uses a 10-day window is
  10-day running mean every 5 days the better
  scheme to compare

28
TAO servicing cruise tracks (meridional sections)
(Johnson et al., 2002)
29

• Comparison and independent validation of
  analyses
• Comparison of T(z) with TAO temperature (not
  independent data)
• Comparison of U(z) with moored TAO ADCPs
• Comparison of T(z), S(z), U(z) with TAO servicing
  cruises

30

• Comparison and independent validation of
  analyses
• Comparison of T(z) with TAO temperature (not
  independent data)
• Comparison of U(z) with moored TAO ADCPs
• Comparison of T(z), S(z), U(z) with TAO servicing
  cruises

x
(G. Johnson)
31

• Comparison and independent validation of
  analyses
• Comparison of T(z) with TAO temperature (not
  independent data)
• Comparison of U(z) with moored TAO ADCPs
• Comparison of T(z), S(z), U(z) with TAO servicing
  cruises

x
x
(G. Johnson)