fill_transport_y_impl Subroutine

   pure subroutine fill_transport_y_impl(h_layer, v_face, buf)
      ! assumed-shape-ok: diag fill — fires once per output frame (cadence-bounded).
      real(wp), intent(in)    :: h_layer(:, :, :), v_face(:, :, :)
      real(wp), intent(inout) :: buf(:, :, :)  ! assumed-shape-ok: diag fill — cadence-bounded
      integer :: i, j, k, nx, ny, nz
      real(wp) :: v_c, col
      nx = min(size(buf, 1), size(v_face, 1))
      ny = min(size(buf, 2), size(v_face, 2) - 1)
      nz = min(size(v_face, 3), size(h_layer, 3))
      do concurrent(j=1:ny, i=1:nx) &
         local(v_c, col, k)
         col = 0.0_wp
         do k = 1, nz
            v_c = 0.5_wp*(v_face(i, j, k) + v_face(i, j + 1, k))
            col = col + h_layer(i, j, k)*v_c
         end do
         buf(i, j, 1) = col
      end do
   end subroutine fill_transport_y_impl