Changes between Initial Version and Version 1 of Ticket #477


Ignore:
Timestamp:
12/30/10 21:37:10 (14 years ago)
Author:
arango
Comment:

Yes, indeed. This is a tricky one. The right-hand-side terms (ru, rv) for the 3D momentum equations are defined a vertical RHO-points. However, a k=0 index is added for the 2D/3D coupling in step2d. This k=0 index contains the vertically integrated forcing term. This term is needed in perfect restart. The solution here is to set the coordinates attribute to: 

                ru:coordinates = "lon_u lat_u s_w ocean_time" ;
                rv:coordinates = "lon_v lat_v s_w ocean_time" ;

This is achieved in def_rst.F by passing -u3dvar and -v3dvar variables (Aval(5)) to routine def_var.

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #477

    • Property Resolution → Fixed
    • Property Status new → closed
    • Property Type bug → defect

  • Ticket #477 – Description

    initial v1  
    1 For a perfect restart, the vars ru and rv are written to the rst file. There is a discrepancy between the dims and coordinates. For example :
     1For a perfect restart, the variables '''ru''' and '''rv''' are written to the restart file. There is a discrepancy between the dimensions and '''coordinates''' attribute. For example :
    22
     3{{{
    34 double rv(ocean_time, two, s_w, eta_v, xi_v) ;
    45                rv:long_name = "RHS of total v-momentum" ;
    … …  
    89                rv:field = "rv, scalar, series" ;
    910                rv:_FillValue = 1.e+37 ;
     11}}}
    1012
    11 The dim is s_w, but the coord is s_rho. this gives a small error when viewing in ncview.
     13
     14The dimension is '''s_w''', but the '''coordinates''' attribute is '''s_rho'''. This gives a small error when viewing in ''ncview''.
     15''''''