Start by seeing that the tension in the position is focused on c4: Black has two pieces attacking the bishop there; White has two pieces defending it. This means Black can’t win anything immediately with captures on that square, but if you stop there you're thinking about the position the wrong way. The important question when you have a chance to force a series of exchanges is how the board would look afterwards—what checks you then would have, and whether any of them would be (or could be made into) forks or other tactical devices.
Okay, so imagine liquidating the pieces trained on c4. Black plays BxB, and White replies RxB; Black plays RxR, and White replies RxR. The two sides have traded bishops and rooks. More importantly, one of the rooks in White’s battery is off the board and the other ends up on c4, leaving White’s back rank weak (bereft of defenders). The natural thought for Black, then, is to drop one of his own rooks there with check: Rd1+. This forces White’s king to f2—on the same file as the b2 bishop, which would have been left loose by the exchanges that started the sequence. The rook fork Rd2+ then wins the bishop next move.
The prospect of a rook fork is nowhere in sight at the outset of this position. You would see it only by patiently imagining the exhaustion of exchanges on c4, then your next check, then the resulting pattern with the king and loose bishop on the same line. If the latter position were set in front of you its solution would be clear. It is worth studying this position until its solution is equally clear because you are able to visualize the consequences of those initial moves. Notice that it involves several of our major themes in constructing forks: creating a loose target, loosening the forking square, and moving the enemy king onto a square where it can be forked.