[Mitgcm-support] 1-year (1997) assimilation

[mitgcm-support at dev.mitgcm.org]

HyperNews notification of new message.  See: 
  http://escher.JPL.NASA.GOV:2000/HyperNews/get/forums/assim/110.html

Adjoint assimilation for the whole year of 1997, restarted from the
3-month (Jan-Mar, 1997) assimilation, is in progress. So far 6
iterations were completed. Some analyses were performed after iteration
5. This figure[1] shows the cost function and its various terms as a
function of iterations. The total cost is equal to the sum of SSH cost
(deviation from T/P data) and costs for flux components (deviation of
estimated tau-x, tau-y, heat flux, freshwater fluxes from the prior
values). The SSH cost is about 60% lower than the simulation. The costs
for tau-x and tau-y are picking up their magnitudes as the optimization
proceeds so as to account for uncertainties in prior wind. They will
eventually be comparable to the magnitude of SSH cost. This figure[2]
shows the comparisons of model temperature with TAO temperature data,
averaged over all TAO moorings in terms of rms variability (top), rms
deviation from (middle) and correction with (lower) the data for
simulation (red solid) and assimilation (of T/P data) experiements (blue
dashed). The assimilation is seen to improve the comparison with TAO
temperature data except for the top level (improvement is indicated by
larger smaller rms deviation from (middle) and larger correlation with
(lower) TAO data. This figure[3] shows the comparisons of model zonal
velocity with TAO mooring data (averaged over the 3 sites, 110w, 140w,
and 165e), again, in terms of rms variability, rms deviation from and
correlation with data. Overall improvement is seen pretty much at all
depths. The largest improvement in zonal velocity is near the
thermocline level (100 m) where the rms deviation is dropped almost from
30 to 20 cm/s, and correlation increases from about 0.48 to 0.73. For
both temperature and velocity, the magnitudes of rms variability are
larger and closer to those observed by TAO data. Some of these
statistics would change as the optimization proceeds further (most
likely for the better, based on experience with the 3 month assimilation
over Jan-Mar of 1997).

-----------------------------------------------------------------------
[1] http://escher.jpl.nasa.gov:2000/hosts/escher/escher4/medea/tlee/FIG_FORUM/cost_1yr_iter6.jpg
[2] http://escher.jpl.nasa.gov:2000/hosts/escher/escher4/medea/tlee/FIG_FORUM/plot_Tstat_ave_1yr_iter5.jpg
[3] http://escher.jpl.nasa.gov:2000/hosts/escher/escher4/medea/tlee/FIG_FORUM/plot_Ustat_1yr_iter5.jpg