The traditional argument appeals to common sense. Whatever is on the bottom had to have been laid before that which comes after it. See:

Whether you think the deposits came fast or slow, this logic is more or less sound, although perhaps problematic in actual application.
However, we are talking about water, here. Flooding in particular. You either have river sedimentation, or lake flooding, or whatever. The point is, that it is not enough to simply say that the lower rocks are older, because
water moves in a direction Its hard to imagine how we might conceptualize a sudden flood (they've almost all got to be sudden in order to trap the poor creatures that would certainly have fled to higher ground if it were static or slow) that didn't have any
lateral movement.
So, we actually have TWO directions to look at when measuring time from a geologic POV:

There are four points labeled here: A, B, C, and D.
Surely we can agree that B is 'older' than A- at least, it was deposited first. And C likewise is older than D. However, if the water is moving in a particular direction, carrying sediment along and laying it down, surely we can't be so certain that C is older than A?
In fact, point 'C' may be several miles down the road, while point A has already been deposited, but the part of the layer that C comes to reside in may not yet have been deposited at all. A in this scenario is older than C, being deposited first, but geologists would assume that C is older than A.
It is a simple empirical fact that sediment moves, accumulates, stratifies, because the water moves in some particular direction.
Doesn't it therefore follow that the only 'age' comparison we can legitimately make would be when we have two points aligned vertically, while points elsewhere in the same column may or may not be older than these points? So, you could only 'absolutely' date A/B and C/D- unless you can determine the direction of the water and the consequential flow of the sediment, you cannot make any relative dating pronouncements comparing A/B to C/D.
Do geologists account for this reality? Not that I know of. It is a purely vertical game, with few exceptions of very limited size and scale (like varves, perhaps).
So, finding fossil X in fossil layer Z in one spot but finding fossil Y in layer Z on the other side of the planet does not assure us actually that they were laid down in the same era of events. Similarly, finding fossil A in layer Z in China but finding fossil B in higher layer W in North America does not assure us that fossil A is actually 'older' than fossil B. Such a determination would require horizontal scrutiny as well as vertical.
I don't know how this conclusion can be escaped, but it certainly calls into question the confidence we can ascribe to using the geologic column to provide any REAL knowledge about times and dates and sequences.