Just think of how you’d represent every game object on a piece of graph paper. In fact, get some graph paper and start drawing different scenes from your game.
If a single sheet of paper is enough, then the game’s data model is 2D. If you need more sheets of paper for different heights (and think hard if you need them or can fake it in the same sheet) then you have a 3D data model.
Afterwards, read up on 2d and 3d Matrices, which will be the first data structure you’ll probably use to represent your game world (Resulting in a tile-based map).
Finally you can jump to arbitrary coordinates within your gameworld, although that’ll probably only be for moving objects, keeping static scenery in a grid is rather useful.
Read through these forums, there’s plenty of information about all these topics, made by people better at explaining it all than me.