Wednesday, 8 January 2014

Associated Press report on shale gas and pollution - what do the numbers really show?

The twittersphere has been alight in recent days with an Associated Press story examining records held by state regulatory agencies regarding complaints of water contamination related to drilling. This story has been widely reported across the media, generally with negative headlines, and extensively re-tweeted amongst anti-drilling campaign groups.

However, it pays to look beyond the headlines, to the actual numbers listed in the report, which is what this post will do. What does the AP report actually tell us about shale gas drilling and water contamination in 4 key US states?

The Question(s):

Firstly, however, it is important to state what we know, and do not know, and what we would like to find out. In scientifc terms, we must state our hypothesis. We already know that contamination can and has been caused at the surface by leakage of fluids from open waste storage pits (not allowed in the UK) and by illegal dumping of waste fluids into streams and rivers without treatment. We also know that contamination of groundwater by fugitive methane can and has been caused by faulty well cement and casing that allows deep sources of methane to move towards the surface.

The question we want to know is - are these types of incidents common or rare; and secondly, are they inevitable, or could they be prevented by better operating practice? If incidents are common and/or inevitable, shale development might be considered an inherently dangerous and therefore unacceptable process. If they are rare and can be mitigated by improved practices, shale gas development should be considered an acceptable technology.

The Data:

The AP report covers 4 states: Pennsylvania, Ohio, West Virginia and Texas. It does not specify in detail the nature of the complaints it discusses, which clouds the issue somewhat. However, from the report they appear to range from short term diminished water flow rates (not particularly serious), to contamination by stray methane migration (the most common complaint, apparently), to contamination by fracking fluid itself (the most serious allegation, I'd contend). Regardless, any incident or complaint is one too many, so for the purposes of this post they will all get lumped together.

If we are to determine whether these events are common or rare, we need to have data on the number of wells in the states considered by the report. The NRDC provides some figures for the number of oil and/or gas wells in each state, though these figures are from 2009, and don't say when the various wells would have been drilled, nor whether they are active or abandoned. However, the AP report doesn't specify whether complaints have originated from abandoned wells, old but still active conventional wells, or newly drilled shale wells, so perhaps the NRDC figures are the best to use. Regardless, I did a little more searching on various state regulator websites, finding the following:

Before comparing numbers of pollution incidents with the number of wells, I first want to mention one striking feature of the numbers in the AP report - the difference between the numbers of complaints received, and the number of incidents actually substantiated though tests carried out by the various agencies. For Pennsylvania, in 2012 the DEP received 499 complaints, but substantiated only 5, meaning only 1% were considered valid. In West Virginia, the DEP received 112 complaints, of which 4 were substantiated (just under 4%). For Ohio, 113 complaints in 2011-2012, with 4 substantiated (just under 4% again), while in Texas none of the 62 complaints relating to water quality have been substantiated (so 0%). 

Why is the the percentage of substantiated claims so low? I'm sure some readers might be tempted to drag out the litigious American stereotype, ready to complain and sue anything and anyone at the drop of a hat. However, perhaps the most relevant data comes from Pennsylvania, where a Penn State study revealed that 40% of private drinking water wells are failing at least one environmental standard anyway. Furthermore, there are over 1 million drinking water wells in Pennsylvania, and approximately 20,000 new ones are drilled every year

The scale of these numbers shows why it is not surprising that many people might have complaints about their water quality. If there happens to be a hydrocarbon well near by, then with all the media coverage of fracking, it is inevitable that drillers get the blame. However, the 40% figure shows that there are in fact there are many other potential sources of contamination, and rigorous testing is required to determine where the blame should properly be apportioned. The AP figures suggest that in over 95% of cases, gas drilling is not to blame.  

Are contamination incidents common?

Lets move on now and consider the numbers of substantiated complaints with the number of wells drilled. For Pennsylvania, 106 cases out of 70,000 existing (NRDC) and 32,000 new (PA DEP) wells = 0.1%. For West Virgina, 4 cases, out of (using the low end DEP figures) 2224 wells = 0.1%. For Ohio, 6 cases out of 50,000 wells = 0.01%. For Texas, 0 cases out of 22,000 shale wells = 0%. For what it's worth, these figures are in line with other reports that have looked into this, such are this report by the Groundwater Protection Council, which reported incident rates per well of 0.01 to 0.03%. Our initial question was: are incidents common or rare? The AP numbers show that incidents of drilling-induced contamination are rare. 

Can better regulations reduce the impacts?

The second question was: is contamination due to drilling inevitable or can it be mitigated by better practice? The rarity of these events alone suggest that they represent aberrations rather than an inherent problem with the drilling and hydraulic fracturing processes. We can go further than this, however. During the Marcellus drilling boom in Pennsylvania, a number of new regulations regarding drilling safety and safe disposal of waste fluids have been enforced from 2010 onwards. Fortunately, for Pennsylvania the AP report breaks down the number of complaints by year, allowing us to judge the effects of these regulations.

The numbers of wells spudded in 2010, 2011, 2012 and 2013, respectively, was 3,340, 3,238, 2,374, and 2,175. In addition to the new wells, remember that opponents of drilling like to remind us that "all wells fail through time", so with all these wells coming in you'd expect to see the number of issues increasing through time, as problems emerge from both new wells and old. 

In fact, the numbers of substantiated issues (and the % of new wells this represents) are, from 2010 to 2013: 29 (0.86%), 18 (0.55%), 5 (0.2%) and 2 (0.1%). This represents a clear decrease in the number of contamination incidents as new regulations have come in to force. The AP numbers show that better regulation can reduce the impacts of shale gas drilling.

In Conclusion
To conclude, just as you should never judge a book by its cover, so you should be careful about judging a newspaper story by its headline. The numbers themselves in the AP report tell a very different story from the headlines it generated.


  1. I would add third and fourth questions: how many people do the complaints represent?; and how are these people affected (in terms of lost income, illness, etc.)?

    In a densely populated area, relying on perhaps a single large aquifer, one contamination event could have much larger ramifications than the statistics you've given might suggest.

    But otherwise, I agree with you that with proper regulations, risks are measurably reduced. I suppose a fifth(!) questions is: what level of risk is acceptable?

    1. Good questions vitaminccs - does each complaint represent a single complainant, or a single well with a problem (i.e. if a well has a problem and affects 10 people, is that 10 incidents or one?). The reports don't make this clear, so I can't comment.

      However, where we have seen issues, they tend to be localised, rather than leading to wholesale contamination of a large aquifer. For example, at Dimock -the best (worst?!) example of poor casing leading to methane contamination, the issues were constrained to a few households along a single road, while the majority of the town didn't report any problems (hence the DimockProud group).

      As for what level of risk is acceptable? I suppose that is where science stops and sociology, politics and economics take over, and we have to decide for ourselves whether the risks outweigh the benefits.