[Mitgcm-support] Note: November 30, 2000

[mitgcm-support at dev.mitgcm.org]

HyperNews notification of new message.  See: 
  http://escher.JPL.NASA.GOV:2000/HyperNews/get/forums/wkmtg/74.html

     =======================================================================     
     
NOTES FROM THE ASSIMILATION MEETING : 30 November 2000    
     
1) Kalman filter  
---------------- 
   - We have a working barotropic (kf019) and baroclinic (kf017)
     Kalman filter.  That they are working is measured by the
     model-data difference of the prediction (innovation) being
     smaller than that of the simulation.  Both filters are being
     calibrated to optimize their performance.  The system matrices
     are rederived (more accurate dynamic modes for the BC filter,
     better mapping operator for the BT filter) and covariance
     matching is being redone using modally decomposed sea level.

 2) Adjoint  
----------- 
  - The 3-month adjoint solution shows encouraging improvements in
    equatorial zonal velocity changes from January to March 1997 (the
    beginning of ENSO) consistent with analysis by Garry Lagerloef
    (based on T/P and SSM/I).

  - Comparison with TAO shows degradation at half the moorings.
    Removing sea surface salinity relaxation (10 iterations over
    3-months) had little impact on improving the situation.  Perhaps
    the initial model error needs recalibration.  

  - From now till the January AMS meeting, we can probably do
    10-iterations of the 1-year adjoint over the weekends.  The run
    will take 18-hours per iteration (75 GB disk storage) and will be
    coordinated with the K-filter.   

  - The forward ECCO code still bombs with KPP (around day 70) in the
    control run configuration.  The adjoint with KPP looks OK as long
    as the forward code doesn't bomb.  

3) Others
----------
  - Alistair reports a bug in subroutine ImplDiff.

  - Some of TAO locations move due to broken mooring lines. Need
    statistics on how often this happens.  

4) Action Items 
----------------
  - Ichiro to ask Larry Eversole the status of our Ames account.  If
    available, we could move the 1-year adjoint to Ames. 
    
=======================================================================