I would suggest to first look at all available logs (if available) and see weather or not there's disconnect in the OGWC (if you do then you know right away you may have some type of heterogenity as already mentioned in previous post). If in fact you do have some type of heterogenity, you could confirm it through interference test (depending on quality and characteristics of the reservoir this may take too long to be practical). You may also want to look for structural features, does the well that started producing water happen to be the deepest / nearest to potential aquifer.

The other thing could do is to compare well performance / productivity index (if well test data is available), because the poorer the well the more likely to cone early (requires more drawdown for the same cubic feet of gas). If you see a trend that way perhaps your worst well gives you indication what may come ahead in the rest of (better) well meaning its not localize but rather systemic. You may also want to compare cum gas production, are wells that shows increased WGR happen to be well that have been put on production longer, it is possible that few wells have been produced significantly harder (above the critical cone rate) and hence coning up much earlier then the rest.