It depends on the actual physical problem you are trying to solve. If you are trying to validate an experimental model that is exposed to air on the top surface you will likely need to model the air with something like your current approach, unless the experimental model is very deep in the water and the air interface is guaranteed to have no effect on your recirculation zone. If you are instead working on a pure cfd case then you will need to think very carefully about the implications of the BCs you choose (eg a pressure outlet represents an "infinitely" tall domain such that pressure doesn't change at the top surface) Keep in mind that air and water are both fluids. If your non-dimensional flow properties match (things like Reynolds number, Mach number, etc) then it doesn't actually matter what "material" your flow is. If the physical problem allows, you can simply pretend that there is no water-air interface, for example if this was on the sea bed you could safely ignore water surface-air interactions, or if you wanted to simply focus on the Reynolds number effects instead and neglect the air boundary