bottom_heights from the immersed boundary method don't look right with a coarse grid

[francispoulin]

Below is a simple script that defines an immersed boundary method for the case of a linearly sloping bottom. Nothing fancy but the resolution is very coarse.

using Oceananigans
Lx, Lz = 1000, 200
bottom(x) = -Lz +  2e-2*x
underlying_grid = RectilinearGrid(size = (5, 5), x = (0, Lx), z = (-Lz, 0),
                                  topology = (Bounded, Flat, Bounded))
grid = ImmersedBoundaryGrid(underlying_grid, GridFittedBottom(bottom))

The z levels are correct.

julia> underlying_grid.zᵃᵃᶠ[1:6, 1, 1]
6-element Vector{Float64}:
 -200.0
 -160.0
 -120.0
  -80.0
  -40.0
    0.0

However, the bottom heights can't be right as they do not align with the level heights. Can someone help me find exactly where the bottom_heights are computed?

julia> grid.immersed_boundary.bottom_height.data[1:6,1,1]
6-element Vector{Float64}:
 -198.0
 -194.0
 -190.0
 -186.0
 -182.0
 -182.0
```
  

[francispoulin]

Sorry! I see that these are the heights of the topography, which are correct, but not the heights used by the GridFittedBottom, as I thought.