How do we determine where something is?
We look at where there are effects, and then postulate a cause that brings about those effects.
That is, to say that a thing has a location in science is to say that it has effects on things that have locations, plus an inference that, therefore, the cause also has a location.
This is how causes in science come to be thought to be in some place, i.e., spatial.
What are exceptions?
The exception would be if something seemingly brute happens at a locale. One gets an effect, but can detect nothing separate from that effect at that locale. There are a few options: 1. It is a truly brute effect, i.e., there is no reason for it occurring, and so no local cause. (This is perhaps the idea behind ‘truly random’ quantum effects.) 2. There is a cause, but the only way we can interact with it is through the effect already detected, and so can investigate it no further. (This could be contingent and due to the technical apparatus we have for investigating the cause, or it could be simply how that cause works.) 3. There is a cause, which can be investigated further, but it is originating somewhere else, and if we could get there we could investigate it further. (So, there is a sort of limited one-way causal pattern in the place of the effect. Here, there is usually a medium through which the cause leads to the effect.) 4. There is a cause, but it is ‘outside’ space, and so the only way we can investigate it while ‘in’ space is indirectly. (For example, if the universe had a beginning, then does it have a cause, and if so, what caused it? It is reasonable to postulate that there is a cause which is ‘before’ or ‘outside’ the creation of space, and therefore non-spatial.)
Here is the question, however: how do we distinguish between 4., 3., 2., and 1.? How, methodologically speaking, would we distinguish?
Also see here.