Yes, you can model first in 2D, but it will be simplified. You need to decide what parameters of the three layers are most appropriate to use in your 2D model.
If you decide that a simplified model isn't close enough to reality, then yes, you can add layers to make it 3D. Read this [url=http://www.feflow.info/html/help/default.htm?turl=HTMLDocuments%2Freference%2Fdialogs%2F3D-layer_configuration.htm]http://www.feflow.info/html/help/default.htm?turl=HTMLDocuments%2Freference%2Fdialogs%2F3D-layer_configuration.htm[/url]
You will apply the seawater variations as a varying head boundary referencing a power function. [url=http://www.feflow.info/html/help/default.htm?turl=HTMLDocuments%2Freference%2Fparameters%2Fboundaryconditions%2Fflow%2Fflow_boundary_conditions.htm]http://www.feflow.info/html/help/default.htm?turl=HTMLDocuments%2Freference%2Fparameters%2Fboundaryconditions%2Fflow%2Fflow_boundary_conditions.htm[/url]
Once you have that sorted out, you could think about adding mass transport and density dependent calculations to look at the seawater interface. Have a read of the relevant sections of the manual when you get up to that part of your modelling. But make sure you get the model geometry and hydraulics working first.