datasets

To obtain the latest SODA reanalysis directly from UMD ftp to: ftp.meto.umd.edu, cd to /pub/outgoing/cao. Subdirectories contain the latest reanalyses (the reanalysis is improving with time). Here is a copy of the README file for beta7.

-- Jim (5-10-02)

Directory Contents:
-------------------

This directory contains version #7 of a retrospective analysis of the
global ocean based on the Simple Ocean Data Assimilation (SODA) package of
Carton et al. (2000a,b), with some improvements (see "Last Changes" below).

The analysis is monthly and spans the period of 1/1950-12/2001.

Data Format:
------------

    The data were written in unformatted SUN direct access binary files.
    Each file contains one year's = 12 monthly analyses (after unzipping
& untarring).
    The first file begins with January, 1950 which duplicates with
February, 1950; both are corresponding to February, 1950.
    Within each month the files include 4 (resolution: 360x128x20)
arrays containing global temperature (degC), salinity (ppt), and
horizontal velocity components (cm/s). In addition, the files contain
6 two dimensional arrays (360x128): wind stress components (Dyn./cm^2),
diagnosed sea level (cm), depth of the 20C contour (determined by linear
nterpolation, m), and heat content 0-125m and 0-500m (degCm).
    The data format was chosen so that the files can be read using Grads
software (see: http://www.grads.iges.org). It can also be read with a simple
FORTRAN program given in the Appendix.

Analysis Methodology:
---------------------

The analysis relies on subsurface temperature and salinity from NODC's
WOA-94, additional CTD and XBT data from the GTSPP and other sources,
thermistor temperature from the TAO array, in situ and satellite SSTs,
and finally, satellite altimeter sea level from GEOSAT, ERS/1, ERS/2 and
TOPEX/POSEIDON obtained from the T/P Pathfinder website. References for
these data sets are given in Carton et al. (1999a,b).
Since the data used in this analysis comes from data archives listed
above, they are first subject to the archive quality control. We conduct
additional quality control including checks for duplicate reports, errors
in the recorded position and time of observations, static stability,
deviation from climatology, and checks on the relationship between
temperature and salinity.

The analysis software uses a multivariate version of optimal
interpolation in which the temperature, salinity and sea level fields
are analyzed using statistical objective analysis. Analyses are produced
every 10 days using a predictor-corrector time-stepping routine (IUA)
that reduces spurious gravity waves as well as the effects of forecast bias.
Temperature, salinity and sea level error covariances are basin-, depth-
and latitude-dependent based on statistical studies of forecast errors.
The zonal and meridional correlation scales at the surface in the tropics,
for example, are 450km and 250km, smoothly approaching a uniform 375km
in midlatitudes. The scales decrease somewhat with depth. We anticipate
that these error covariance models will become more sophisticated in the
next year.

The analysis forecast model uses the GFDL MOM2.b numerics driven by
COADS and NCEP winds, with conventional choices for mixing, etc.. The
domain of this analysis is global with 1.0x1.0 lat-lon horizontal
resolution in midlatitude reducing to 0.45x1.0 resolution in the tropics
for a total 362x130 horizontal grid points. The model has 20 vertical
levels on a stretched grid with 15m resolution nearsurface.

Analysis Results: the accuracy of the results are determined by
examination of forecast error (Carton et al., 1999a), comparison to
independent observations (Carton et al., 1999b), and comparison to other
analyses (Chepurin and Carton, 1999). (NOTE: These papers are based on
our older version of lower resolution run which is global with 2.0x2.5
lat-lon horizontal resolution in midlatitude reducing to 0.5x2.5
resolution in the tropics for a total 96x146 horizontal grid points.
The vertical resolution is the same 20 levels.)

Last Changes:
-------------

1. Correlation functions depend on the velocity field;
2. Temperature and salinity are assimilated deeper to 1099.66 meter
     (used to be done to 443.79 meter deep);
3. Correlation radius is increased to 1000 km for the levels below 443.79
     meter;
4. Surface fresh water flux and river discharges are added;
5. 0-1000 meter dynamic height is used instead of the sea level height for
     altimetry assimilation;
6. Real ocean bottom topography is used;
7. Mixed layer determined from density instead of temperature;
8. updated NODC WOD-98 and GTSPP subsurface data are used.
9. updated ERS/1-G, ERS/2 and TOPEX satellite observasions are used.

References :
------------
see: http://www.meto.umd.edu/~carton/carton/ref.html

(several of the most relevant papers are available via anonymous ftp at nimbus04.umd.edu, cd to
/pub/outgoing/carton.)

Appendix:
---------

C
C --- FORTRAN PROGRAM TO READ DIRECT ACCESS DATA FILES
C --- (AFTER UNZIPPING AND UNTARRING).
c
C --- THE HORIZONTAL GRID STARTS AT 0.5 DEG EAST AND 59.99 DEG S.
C --- REMEMBER THE VERTICAL GRID BEGINS AT THE LOWEST LEVEL (3622.5 M)
C --- AND PROGRESSES TOWARDS THE SHALLOWEST LEVEL (7.5M). THE GRID IS
C --- DESCRIBED IN DETAIL IN THE "*.ctl" FILE. BE AWARE THE HORIZONTAL
C --- GRID FOR VELOCITY AND FOR TEMPERATURE, SALINITY, AND SEA LEVEL
C --- DIFFER.
C
c --- !!!!! note:
C --- IF USE DEC WORKSTATION, NEED TO ADD ONE LINE FOR OPENING UNIT 10.
C
      parameter (im=360, jm=128, km=20)
c
      DIMENSION T(im,jm),S(im,jm),U(im,jm),V(im,jm)
      DIMENSION TAUX(im,jm),TAUY(im,jm),SFH(im,jm)
      DIMENSION Z20(im,jm),HC125(im,jm),HC500(im,jm)
C
      OPEN (UNIT=10, FILE='asm16_****.dat',
c --- note: if use DEC workstation, add the following line
c    *     convert='big_endian',
     *     FORM='UNFORMATTED', ACCESS='DIRECT',
     *     RECL=im*jm, STATUS='OLD', READONLY)
C
      IREC=1
C
C --- FOR 12 MONTHS
C
      DO ICNT=1,12
C
C --- FOR 20 LEVELS, FROM BOTTOM TO TOP
C
        DO IT=1,20
          READ(10,REC=IREC)T
          IREC=IREC+1
        ENDDO
c
        DO IS=1,20
          READ(10,REC=IREC)S
          IREC=IREC+1
        ENDDO
C
        DO IU=1,20
          READ(10,REC=IREC)U
          IREC=IREC+1
        ENDDO
C
        DO IV=1,20
          READ(10,REC=IREC)V
          IREC=IREC+1
        ENDDO
C
C --- FOR 6 TWO DIMENTIONAL ARRAYS
C
        READ(10,REC=IREC)TAUX
        IREC=IREC+1
        READ(10,REC=IREC)TAUY
        IREC=IREC+1
        READ(10,REC=IREC)SFH
        IREC=IREC+1
        READ(10,REC=IREC)Z20
        IREC=IREC+1
        READ(10,REC=IREC)HC125
        IREC=IREC+1
        READ(10,REC=IREC)HC500
        IREC=IREC+1
C
      ENDDO
C
      STOP
      END

Grads Control File "***.ctl" :
-------------------------------

There are two grads control files available: beta7_ts.ctl
containing model t&s grid, and beta7_uv.ctl containing model
u&v grid. beta7_ts.ctl is usually used to display images. You
can use beta7_uv.ctl, if you want precise locations for velocities.

!!! NOTE : if you use DEC workstation, please add one line
"OPTIONS byteswapped" under the first line "DSET" in ***.ctl file.

Questions:
----------

If you have any questions, please e-mail to Xianhe Cao at
cao@atmos.umd.edu