Rutgers University
  • Data assimilation vs rapid process evolution

    Posted on August 24th, 2020 Scott Glenn No comments

    Adding this post so that we do not loose track of this event revealed by NG645 as it is passed by Marco.

    First plot below is the temperature profile comparisons from this morning. The data assimilative GOFS is closest to the glider temperature profile.  There is some cooling in the upper ocean in RTOFS.  Recall that RTOFS backs up 2 days in GOFS, pulls the initial condition, and moves forward without data assimilation with the NOAA winds.  GOFS is doing data assimilation up to the present.  RTOFS relies on model processes to carry the evolution forward for the final 2 days. Hard to say which approach is better based on temperature.


    But now lets check the salinity profiles, and the evolution of the barrier layer.  Here GOFS (red) kept the barrier layer salinity low consistent with the pre-storm conditions.  RTOFS (green) mixed the barrier layer fully consistent with the Navy glider profile (blue) based on physical processes over the last 2 days.

    We saw similar behaviors in the data assimilation during Hurricane Michael in the Gulf a couple years ago.  Perhaps this is another case where the choice of which CTD profile is assimilated during a period of rapid evolution makes a difference.  Its an important time to investigate what we have noticed here, since RTOFS-DA is currently being run in experimental mode.

    Last 5 posts by Scott Glenn

    Comments are closed.