There is an old-ish (like, from the 1950's I think) and fairly heuristic rule that the XANES edge energy shifts scale as 1/R^2 for the nearest neighbor.    I think that Grant Bunker has a nice set of slides showing this, somewhere on the web.  For example, I think this explains well slight energy shifts in the peak energy of the sulfate resonance.     My recollection is that it works pretty well for many well-constrained set of spectra, but isn't quite as quantitative and readily applicable to unknown systems as one might hope.

It's also sort of related to the idea of bond-valence that, for a given class of systems (say, oxides of a particular metal), R, N, and formal valence state are not independent quantities.