Impact of Argo Salinity Observations on Ocean Analyses

1
Impact of Argo Salinity Observations on Ocean
Analyses
Chaojiao Sun, Michele Rienecker Christian
Keppenne, Jossy Jacob, Robin Kovach NASA/GSFC
Global Modeling and Assimilation Office (GMAO)
Acknowledgement Gregg Johnson (TAO servicing
cruise data) Willa Zhu (retrieving recent TAO
servicing cruise data) Dave Behringer
(quality-controlled XBT data)
2
Motivation
Previous studies have shown that assimilating
temperature and synthetic salinity has an impact
on salinity and current fields (Sun et al., 2006).
3
Purpose To assess the impact of Argo salinity
assimilation on ocean analyses using independent
salinity observations.

• Experiment details 2000-2004, focusing on the
  last year 2004
• Experiment 1 (ARGO) assimilates Argo
  temperature and salinity, in addition to the
  assimilation of subsurface temperature
  observations and synthetic salinity profiles
  (where no Argo salinity data is available).
• Experiment 2 (NARGO) no Argo data used, only
  subsurface temperature (XBT and moorings)
  observations and synthetic salinity profiles are
  assimilated.
• Experiment 3 (MODEL) model simulation without
  any assimilation.
• Observations
• XBT (QC by NCEP/Dave Behringer)
• TAO/TRITON/PIRATA (delayed mode, QC by PMEL)
• Argo (delayed mode, QC by GODAE/Monterey server)
• Forcing
• Atlas/SSMI time varying wind stress
• E-P forcing, with P from GPCP monthly
  mean precipitation
• NCEP CDAS1 SW (for penetrating
  radiation) LH (for evaporation)
• relaxation to Reynolds SST
• Salinity analysis validation independent
  observations
• CTD casts from TAO servicing cruises

4
Observation and model errors

• Observation error estimates are based on the
  vertical temperature and salinity gradient, with
  the maximum and minimum specified as the
  following
• Argo salinity error minimum 0.03 psu, maximum
  0.20 psu
• Synthetic salinity error minimum 0.05 psu,
  maximum 0.30 psu.
• Temperature observation error minimum 0.3oC,
  maximum 0.7oC.
• Model errors are assigned uniformly (based on
  ensembles model simulations)
• Temperature error 0.77oC
• Salinity error 0.20 psu

Profile distribution and volume in 2004 in the
tropical band of 10S-10N
5
180E Dec 3-12 2004
TAO CTD casts Jun 20-29,1004
ARGO analysis
NARGO analysis
MODEL simulation
Note color scale of model simulation and
analyses does not exactly match that of CTD casts
plotted at the EPIC website. Pink line denotes
the mixed layer depth.
6
155W, 8S-12N
TAO CTD casts Jun 20-29,1004
ARGO analysis
MODEL simulation
NARGO analysis
7
155W, 8S-12N
of Argo obs. during Jun 20-29 8
TAO CTD casts Jun 20-29,1004
of Argo obs. Apr-May-Jun,2004 67
of Argo obs. in Jun 2004 26
Number of ARGO profiles during and before the CTD
casts. Note that the same ARGO analysis is shown
in all three analysis plots.
8
Mean and STD of differences between analysis,
model simulation and independent CTD observations
9
Surface Salinity at Equatorial Pacific (2000-2004)
TAO
ARGO
NARGO
MODEL
10
Surface Salinity at 180W (2000-2004)
ARGO
NARGO
MODEL
11
150m Salinity at 180W (2000-2004)
ARGO
NARGO
MODEL
12
Conclusions

• Argo makes a difference
• Synthetic salinity profiles derived from Levitus
  T-S climatology are useful in reducing subsurface
  model salinity bias.
• Argo salinity observations reduce the impact of
  climatology of the synthetic salinity,
  introducing more subsurface variability than in
  the MODEL or NARGO cases
• Argo salinity observations improve the
  comparison with independent CTD observations in
  the overall salinity structure horizontal and
  vertical gradients
• Subsurface salinity assimilation impacts surface
  salinity distribution.
• Aquarius
• Aquarius data will help
• validation
• assimilation for the surface layers

13
Future work

• Validate salinity at depth and currents
• Assimilate available surface salinity data to
  evaluate the impact of surface salinity from
  Aquarius

14

• GMAO treatment of salinity via TS scheme T
  and S assimilation
• S comes from ARGO when available
• Synthetic S(z) - T(z) is used with T-S relation
  from Levitus climatology to generate a synthetic
  S(z) consistent with temperature variations
• No modification of salinity in the models
  surface mixed layer salinity varies according
  to estimated E-P

Climatology T(z)
Climatology T(z)
Climatology S(z)
Observed T(z)
Schematic of the derivation of synthetic salinity
profiles.