I opted for something completely different in the end (the tower).

Just as a quick layers example in the case of this map, if I was doing it on layers, I would have the photo image of the cliff on the bottom layer. I wouldn't edit this "base image" at all, I would just leave it as is.

Then on the next layer I would have the solid black background bit you have there covering up much of that photo image.. This is so I can add or remove more of the photo image just by adding more black or deleting it as needed. Because this black is on it's own layer and doesn't affect anything else, it's a super quick adjustment.

Then on a third layer, I'd begin adding the caves themselves. So you'd add that clouds bit and begin colouring it.