I'd be interested to learn more if there are any statistically significant results showing clustering - proving that there is spatiotemporal clustering that is different enough from what might be expected at random is pretty hard. I know that Sellafield has been studied in detail many times, with no definitive evidence yet found.
There are also lots of confounding variables to account for - for example, power plants (of all kinds) commonly have higher amounts of road freight than would otherwise be the case, thus greater air pollution etc. Inhabitants of areas around power plants move in and out of the area. Some people that live in the area work at the plant, whereas others do not. Prevailing weather and topography may affect how any emissions from a plant disperse. Spatial stats is hard!
