This file describes some additional details of the OI figure.

Top Panel

   Real-time forecast from June 10, 2001
      Temperature at 3m overlain with vectors of total velocity at 3m.

Middle Panel

   Left:  Nondimensional temperature error field.
          Dark blue indicates area of maximum impact of data.
          Dark red indicates area of minimum impact of data.

   Right:  Temperature at 3m overlain with vectors of total derived
           (geostrophic) velocity at 3m.

Bottom Panel

   Data is assimilated into the restarted forecast at 4 distinct times
   with increasing relative weights ("ramped in")
            Time (days)       Weight
             0.125             0.2
             0.375             0.667
             0.625             0.875
             0.875             0.99

   Barotropic velocity is NOT assimilated.  The issue here is that the
   error field is nonconstant along the coast but the transport streamfunction
   must remain a constant along the coast.

   The areas of the forecast replace by the data are obvious.

   Areas where forecast has been maintained:
     - Inside "elbow" of Cape Cod (Brewster).  High forecast temperatures
       maintained by large error field.
     - Strong flows off Plymouth.  Partially maintained by barotropic
       velocity (not assimlated anywhere).