Edge bathymetry values are wrong in the OVERFLOW test case (at U-points)

[jchanut]

Context

Trivial bug in OVERFLOW test case: bathymetry computation at U-points in usrdef_zgr.F90

Analysis

Bathymetry at U-points needs to be assigned non-zero values at eastern and western walls (so that later computation of scale factors does not give null thicknesses). This is done by the following lines:

      DO ji = mi0(1), mi1(1)              ! first row of global domain only
         zhu(ji,2) = zht(1,2)
      END DO
       DO ji = mi0(jpi), mi1(jpi)         ! last  row of global domain only
         zhu(ji,2) = zht(jpi,2)
      END DO

It should rather be:

      DO ji = mi0(1), mi1(1)              ! first row of global domain only
         zhu(ji,2) = zht(ji,2)
      END DO
       DO ji = mi0(jpiglo), mi1(jpiglo)   ! last  row of global domain only
         zhu(ji,2) = zht(ji,2)
      END DO

…

[jchanut]

In 11076:

Correct bathymetry at U-points, #2289

[jchanut]

In 11077:

Correct bathymetry at U-points, #2289