In the rock it is usually a transition over a height set by capillary pressure and by pore size, not a sharp plane. Tighter rock holds a thicker transition. The free-water level, where capillary pressure is zero, sits at the base of that transition and is the surface reservoir models usually hang the saturation curve from.
A contact picked on a log as 'the depth where we see water' can sit above the free-water level. Mixing the two in a volume calculation moves the oil in place. The contact is reported as true vertical depth, subsea or below a stated datum, never as measured depth along a deviated well.