KELVIN test case blows up

[m.hadfield]

In the Kelvin test case, a spike in sea surface height develops in the SW corner of the domain, reaching a value of 46 m at t = 2 hours. The model blows up at t = 3 hours.

[m.hadfield]

Found it! In the M2 boundary conditions at the western edge, this

          cff=SQRT(g*GRID(ng)%h(Istr-1,j))

should be this

          cff=SQRT(g/GRID(ng)%h(Istr-1,j))

Corrected file attached.

[arango]

Actually, we need here:

cff=SQRT(g*GRID(ng)%h(Istr-1,j))              ! gravity wave speed, c

the problem is that the boundary conditions for ubar are wrong. We need to have:

ubar_west = Uo * EXP(-f*y/c) * SIN(omega*t)
ubar_east = Uo * EXP(-f*y/c) * SIN(omega*x/c -omega*t)

where
        c = SQRT(g*h)
       Uo = c * Eo / h = Eo *SQRT(g/h)                 ! Eo: initial elevation

I guess that I didn't fixed this correctly when it was reported in the ​forum.