The USGS study examined water quality in Sullivan County, Pennsylvania. Crucially, this study focussed on areas that have yet to be drilled for shale gas. So far in the USA such baseline studies have been rare, which means that it can be difficult to establish whether water has been contaminated by shale gas drilling, or whether it was already of poor quality to begin with.
As I have discussed in a prior post, it is wrong to assume that all groundwater was of Evian-quality prior to drilling: there are many potential sources of groundwater contamination, both natural and man made, that could have caused problems well before shale gas drilling began.
Therefore, although this USGS study hasn't received the press attention given to the Duke study, it could come to be seen as really important, because it provides a baseline against which changes caused by drilling can be assessed.
So what does the Sullivan County baseline look like? Well 85% of the 20 water samples taken contained at least some radon-222 above the US maximum level of 300 picocuries per liter (pCi/L), while 10% exceeded the alternate maximum level (I have no idea why the USEPA has a maximum level and then an alternate maximum level - what does that mean?) of 4,000 pCi/L. 35% of the wells had some methane in them, although only 2 samples (10%) had levels to get excited about: 4.1 and 51.1mg/L (the US maximum level is 10-28mg/L).
Importantly, the USGS carried out isotopic fingerprinting of the methane in these wells. It is possible to tell the difference between methane that has been created in the shallow surface by bacterial activity (so-called 'biogenic' methane) and that created at depth by heat and pressure (so-called 'thermogenic' methane), and it has been suggested that, where thermogenic gas is found, shale gas drilling is the likely culprit.
Unfortunately, the isotopic fingerprinting of the Sullivan County gas showed that is WAS thermogenic, in an area WTHOUT any gas drilling. Does this mean that the thermogenic/biogenic testing can no longer be used as the determining factor in the debate over whether methane is caused by shale gas extraction? I think at the least more caution might be required.
The main take-home point from the USGS study is that water quality in the region is highly variable. This means that, whether you believe that shale gas drilling has or has not caused contamination, proving your case either way will be very difficult, especially without baseline studies.
Moving on to the Duke study, which examined the water quality in gas-producing areas of Pennsylvania. Rather than bore you with statistical analyses about P<0.0007 for this and P=0.0001 for that (necessary for science, less interesting for blogs), I think the story would be better told with a couple of pictures. Firstly, where the groundwater was sampled (along with the locations of shale gas drilling):
and then the results - methane concentration plotted versus distance from shale gas wells:
The Duke researchers also looked at the isotopic composition of the methane gas, finding it to be thermogenic, and they also found ethane (a slightly heavier form of natural gas, not produced biogenically), implying that the gas must have originated at depth, and therefore be caused by shale gas drilling. I think that this part of their conclusions is slightly shakier - we've seen from the USGS study above that it is possible to have thermogenic methane, and smaller amounts of ethane, in areas that haven't seen drilling.
I also have some doubts about how the wells were chosen for analysis. You can see from the first plot that the wells sampled occur in clusters - they're very non-random. Above I pretended to chose not to discuss the statistics in the Duke paper to save the poor reader from a load of boring numbers. In fact, my concern is that such statistical analyses usually assume a certain degree of uniformity in a sampling process, and that doesn't seem to be the case here, so I'm not sure as to their validity.
In the methods section of their paper, the Duke team mention that they got their samples via Homeowner Associations. Although which homeowner associations these might be are not described, it seems likely that these are homeowner associations would be those opposed to natural gas drilling. In turn, such associations are presumably likely to have (a) elevated methane concentrations in the water and (b) gas drilling in the near vicinity. So, if your sampling is biased by choosing to select water from sites that match both descriptions, it's not particularly surprising to find homes near gas wells with elevated methane - because that's how your samples were chosen in the first place.
You can see this by looking at the distribution of sampling points in the above figure, which are clustered in certain places on the map, rather than spread evenly. To really robustly establish a link between drilling and methane contamination, a much more uniform sampling and testing program would be required. It's a shame that the Duke researchers did not do this, because they've left what could have been a really fundamental study open to criticism.
To summarise my thoughts - I think that it is likely that at least some of the incidences of methane contamination in the Duke study HAVE been caused by drilling. The example of Dimock shows a pretty clear case where of shoddy drilling practices leading to methane contamination. However, based on the sampling methods used, I'm not sure that this study, despite the media fanfare, actually adds anything to this. The real question we want to know is - are methane issues ubiquitous and inevitable, or do they represent a handful of 'bad apple' cases of poor drilling practice, the number of which can be minimised with good practice, strong regulation and good oversight? A more rigorous attempt to sample PA wells more uniformly is required (for example, many of the Duke samples do not have much/any methane in them).
Finally, keep in mind that methane, while a nuisance, is not toxic, and most private drinking wells should have filters to remove methane prior to the water entering the house. The Duke researchers also looked for other chemicals that might be associated with drilling. Opponents of shale gas often point to a smorgasbord of nasty-sounding chemicals associated with fracking fluids that they claim are polluting groundwater.
Much like their previous studies both in PA and elsewhere, the Duke researchers found no evidence for any of these chemicals. Nor would we expect them to: methane, being buoyant, has the potential to move up through the annular gaps and/or cracks left in a poorly-cemented well. Drilling and/or fracking fluids are not buoyant (i.e., they have similar densities to the brines that saturate the rocks at the depths of the shale reservoirs), so there is no force available to push them back to the surface - you may as well try to suggest that dumping sewage/pollution in London is going to contaminate the headwaters of the Thames in the Cotswolds. The water is simply flowing in the wrong direction.
So, because it's been another long and rambling post, a summary:
- lots of the water in PA is of poor quality to begin with,
- it is possible (likely) that there have been some instances of methane contamination,
- whether these cases represent a few outliers, or are more ubiquitous, is not established by the Duke study,
- and there is absolutely no evidence that any of the various chemicals associated with fracking fluids are getting into groundwater.