Operational oceanography A GODAE perspective

Document Sample
Operational oceanography A GODAE perspective Powered By Docstoc
					                    Satellite Altimetry
             Improvements expected for ENACT

                 P.Y. Le Traon, F. Hernandez P. Schaeffer, M.H. Rio
                        CLS Space Oceanography Division


 CLS tasks for ENACT
 Absolute dynamic topography from satellite altimetry
 Comparing and combining altimeter data with in-situ data
 Planning/points to be discussed
                     CLS tasks in ENACT
Produce accurate/intercalibrated multi-mission altimeter data sets (SLA
and MSLA) for the period 1987/1988 (GEOSAT) and 1992-2004 (T/P
and ERS-1/2 then Jason-1 and ENVISAT) (delayed mode + real time)

         Improved processing algorithms will be incorporated : better
         ocean tide model, improved methods for orbit error
         reduction and mapping methods (e.g. new signal/noise
         covariance models), SLA relative to a 7-year mean (1993-

Estimate a global mean dynamic topography (7-year mean - 1993-
1999) to reference altimeter data (absolute dynamic topography)
        by combining the CHAMP geoid data with available in situ
        data and altimeter data using a synthetic geoid methodology

 Compare altimeter data with in-situ data (T and S) (importance of
 contribution of barotropic signals)
Spatial scales, zonal and
meridional propagation
 velocities, time scales
derived from 7 years of
    TP+ERS maps
       Long wavelength errors
         and large scale high
     frequency signals from T/P
        SLA (Schaeffer et al.,

      These signals are taken
        into account in the
       mapping procedure

     They are also corrected for
      in the SLA along-track
  EKE TOPEX/POSEIDON+ERS-1/2 and Los Alamos Model (Ducet, 2000)
  Use of Los Alamos model to analyze the sampling characteristics of multiple
altimeter missions and improve the T/P+ERS processing (Le Traon et al., 2001;
                            Le Traon et al., 2002)
                  SSALTO/DUACS : Real time processing of
            TOPEX/POSEIDON, ERS-1/2 (GFO, Jason-1 and ENVISAT)
       Real time processing
            (2-3 days)

 Global crossover minimizations,
inverse techniques to remove long
wavelength errors=> high accuracy
            SSH data

    Consistent mean profiles to
 reference multiple altimeter data
      => consistent SLA data

   Products directly useable for
    scientific and operational
    applications (climate and

CLS and CNES agreement for long
  term operation. Real time data
available for GODAE and scientific
 (climate) research (need to sign a
  data use agreement with CNES)
                                      http://www.cls.fr/duacs will be soon
                                             moved to the Aviso site
SSALTO/DUACS : High resolution from T/P, ERS-2 and GFO
          + synthetic mean dynamic topography
                  SSALTO/DUACS - Real time system

Short term improvements (March 2002)

    Include GFO (orbit error, mean profile)
    SLA relative to a new 7-year error corrections
    New covariance for ocean signal (propagation mean (1993-1999)
    Improved algorithms for orbit and long wavelength velocities) and noise
    Maps on MERCATOR grid (1/3°) - NetCdf format

Improvements for end of 2002

    Include Jason-1 and ENVISAT when available (after verification phase)
    Provide a mean dynamic topography derived from geoid (and/or in situ) to add
    to SLA data to yield absolute dynamic topography (solution for the Atlantic
    already available and used/tested by MERCATOR)
    Twice weekly processing and distribution (envisioned)
    Differences between the TP and ERS 7-year means (1993-1999) and the
GEOSAT 2-year mean (1987-1988) (= interannual variability). This correction
is applied to GEOSAT data get a GFO/GEOSAT mean profile consistent (i.e.
              1993-1999 mean) with TP and ERS mean profiles.
      Contribution of GFO

(a)                         (b)

Contribution of GFO
                           Mean dynamic topography

       Needed to reference altimeter data and get absolute dynamic topography

A precise mean dynamic topography will have a major impact on scientific and
operational applications of satellite altimetry (e.g. Woodworth et al., 1998; Le Provost et
al.,1999) :

•better interpretation of altimeter signals (absolute dynamic topography)

•comparison/combination with in-situ data

•general circulation and heat transport

•data assimilation and ocean forecasting

Estimation of a mean dynamic topography from gravity missions
                        and altimetry

  The computation of a mean dynamic topography will require :

  1. To estimate a precise mean geoid from CHAMP/GRACE/GOCE.

  2. To subtract this geoid from a very precise altimeter mean sea surface. It will
     be then necessary to filter the resulting mean dynamic topography taking into
     account the geoid and mean sea surface errors.

  Will be mainly useful when GRACE and GOCE geoid are available. CHAMP should
           improve the today situation (EGM96) but not much (to be investigated)

     => use of in-situ data and combine with CHAMP mean dynamic topography
Mean dynamic topographies from
                                    EGM96 - MSS
POCM, Levitus and EGM96-MSS

  rms differences are about 12 cm
     (wavelengths > 2000 km)

         Levitus (700 m)

             Mean dynamic topography from in-situ data
                with the “synthetic geoid” approach

One of the disadvantages of mean dynamic topographies derived from in-situ data
is that to reduce the ocean variability noise, they have to use in-situ data over
several decades. Thus they do not correspond to mean dynamic topographies
compatible with the altimeter sea level anomaly (SLA) data (i.e. a mean over
several years).

Another approach is to combine the altimeter SLAs (h’) with simultaneous in-situ
data. In-situ data can provide estimates of h (although the barotropic part maybe
more difficult to estimate). Space/time interpolation of satellite altimetry gives h’
at the time and location of the in-situ data; the combination of the two estimates
can thus yield <h> over the needed time period (<h> = h - h’).

This is a powerful methodology but it requires a large number of simultaneous
data. As part of ENACT, we will apply this technique to the global ocean using
T/P and ERS data and XBT and surface drifters. This estimation will then be
combined to the one derived from CHAMP geoid and altimetry.

Preliminary results for XBT data (global) and surface drifters (Atlantic)
Synthetic geoid - Number of XBT measurements used
Synthetic geoid from XBT data
Synthetic climatology - Levitus (in cm)
Synthetic climatology : application for Gulf Stream path monitoring

        Climatologie Levitus         Synthetic climatology
                                      Climatologie Synthétique
 Synthetic geoid - Impact on
   the comparison between
  altimeter and in-situ data.
(mean dynamic topography error is
 the dominant error when altimeter
data are compared with in-situ data)
      Synthetic geoid
from WOCE surface drifters
                Comparison altimetry and in-situ

Vertical structure of sea level variability (barotropic/baroclinic)

Sampling errors, salinity effects, inconsistencies between means

Crucial for a joint assimilation (combination) of altimetry with in-situ data

  Systematic comparison of in-situ and altimetry will be carried out as
                           part of ENACT
     Large scale variability from in-situ (XBT) and altimetry
                                    (Guinehut, 2002)

Methods :
Space/time interpolation of
altimetry at the XBT/CTD
Use the synthetic climatology to
estimate Hdyn anomalies
Use T/S relationship for XBT

      __ SLA altimétriques
            SLA hydrographiques

Very good agreement (Pacific
   ocean). What about the
        differences ?        10
Mean seasonal Differences altimetry/in-situ            (1993-1999)
              about 120 000 XBT/CTD - TP + ERS-1/2
Hiver                              Printemps

Eté                               Automne

      -4 -2   0   2   4               -4 -2    0   2   4       cm
  Comparison with Anomaly Sverdrup transport (1993-
           1999) (ERS-1/2 winds, CERSAT)

        40° - 50° N

         20° - 30° N

                        6                                     30
         20° - 30° S    0                                      0
                       -6                                    -30

 At large scale, the seasonal cycle of the anomaly of the
  700 m topography = barotropic response to wind forcing
Regression SSH/Hdyn (0/700) in-situ/altimetry     Clipper model

                                                Clipper : SSH/H700
D1.4 Mean sea level from altimetric, gravimetric, and in situ ocean
observations (Month +18)
D1.5: Complete uniform set of altimeter sea level anomalies, along track and
gridded (final version) (Month +33)

M1.3: Start distribution of altimetric sea level anomaly data (month +3)
M1.4: Distribution of absolute mean sea level estimates (month +18),   50%)
Points to be discussed
SLA relative to a 7-year mean. When do you need the data ?
GEOSAT data will not be ready before the end of 2002. How GEOSAT data
will be used ?
Climate oriented product (e.g. SLA/MSLA with mesoscale signal filtered out ?)
Mean dynamic topography over the North Atlantic is available for testing.
Altimeter missions
    Vertical structure of sea level variability
from a low resolution model (Fukumori et al., 98)
         Combining altimetry (+SST) with in-situ data (Argo) to
             estimate 3D temperature and salinity fields

     T at 200 m (model)       Altimetry        Altimetry+Argo
T/200m   Champ modèle        AO (altimétrie)    AO (altimétrie + Argo)

     model (wind


Mean dynamic topography from EGM96 and CLS MSS
   (wavelengths < 2000 km are filtered with a 2D Loess filter)
Mean dynamic topography derived from Levitus climatology
Mean dynamic topography from POCM
       Mean dynamic topography from in-situ data
Mean dynamic topographies have been derived in the past from climatological
data (e.g. Levitus). These estimations generally assume a level of no motion and
are thus missing part of the signal, in particular, at high latitudes (e.g. ACC).

In-situ data are now providing the “best” global mean dynamic topographies.
Typical accuracy is about 10 cm rms for scales larger than 300-500 km but these
figures should improve in the coming years :
     - by using new global data sets such as Argo (and deep floats for the
       barotropic component)
     - by using more elaborated processing techniques (inverse modelling, data
       assimilation, synthetic geoid)

These estimations should be compared and then combined with estimations
derived from gravity missions
3000 floats providing T and
S profiles (0-2000 m) every
  10 days and measuring
    velocity at 2000 m =
  improved estimation of
  absolute mean dynamic
topography (a few cm rms)
                       CLS Mean Sea Surface
A 7 year mean (1993-1999) surface estimated from T/P, ERS and GEOSAT data
CLS MSS formal error

Shared By: