Analysis: Ground water contamination

The BBC has just reported an analysis of two recent studies from the University of Texas and Duke University into possible contamination of groundwater by chemicals from hydraulic fracturing in the USA. Both studies are a great improvement on previous studies in terms of methodology, and both show concentrations of chemicals in ground water.

Moreover, both studies indicate that water tested from water wells away from current gas wells show low levels of chemicals compared with elevated levels at some of the locations that were closer to the shale gas wells. The chemicals concerned were methane, ethane, propane, methanol, ethanol, and the heavy metals arsenic, selenium, strontium and barium.

Despite both studies suffering from three major scientific problems, the data is sufficiently strong to indicate that casing leakage may exist in some wells. However, the problems in methodology limit the data to framing an hypothesis for further more rigorous testing, rather than representing robust scientific evidence of itself.

Problem 1

The first problem is that neither study has carried out, reported or used back-ground measurements that were co-located with the recent ground water measurements, i.e., measurements of ground water before hydraulic fracturing started. Hence we cannot say whether the chemical concentrations in the ground water have happened after drilling and hydraulic fracturing, or were there all along.

The University of Texas study does compare their results with historical data (1989-1999), but historical data from different water wells at unstated distances from the test wells, and moreover their study does not take into account the presence of other water wells and legacy hydrocarbon wells that may complicate the picture. Neither is there any analysis of how the aquifer may have changed in the intervening years.

Instead of back-ground measurements, the studies analyse the data with respect to distance from the nearest usually active shale gas well in the hope that the measurements farthest from a shale gas well might provide a rough regional back-ground reference.

Problem 2

The second problem is that samples were taken from areas where the presence of chemicals in the ground water was already known or suspected. This skews data and makes it difficult to interpret properly. In fact, the University of Texas study used water sampled from land owners who selected themselves to be part of the study, leaving the data open to claims that it is statistically invalid, while the Duke University researchers used a number of sources for their data amongst which were samples from homeowner associations.

Problem 3

The third problem concerns the analysis of data with respect to shale gas wells. It is incredibly difficult to associate a spatially distributed effect (chemicals in groundwater) with a cause (hydraulic fracturing in a well), as we have seen for decades in the attempted correlation of cancers in the hinterland of Sellafield, Windscale and Calder Hall in the UK with those nuclear facilities.

Both the University of Texas and the Duke University study chose to present the data qualitatively on graphs of chemical concentration versus distance from the nearest active shale gas well. At first sight it is clear that at large distances from shale gas wells the concentrations are very small, while at small distances, less than 1 km say, there is a cloud of elevated concentration values. However, this diffuse cloud of several dozen elevated values draws the eye away from the fact that there are also very many more values which are not elevated and are in fact almost zero. Proximity to a shale gas well does not, therefore, automatically imply that the ground water is contaminated.

Detectable limits and back-ground levels

The text associated with the Duke University study is equally misleading. It says

“Dissolved methane was detected in the drinking water of 82% of the houses sampled (115 of 141).”

This is true, and seems relevant, but actually it isn’t.

The detection threshold for methane is about one ten thousandth of a milligram per litre, which is way below the methane levels normally found in ground waters: the median value for the southern UK is twenty times the detection limit. Hence, almost all ground waters will contain detectable methane - The question should be ‘is it significant?’.

To put it all in perspective, the pre-drilling methane concentration in the Balcombe ground waters was recently measured at between 6.84 and 12.4 milligrams per litre from three samples taken on three days spread over a 10 day period! This shows five things:

(i)                 the value is naturally high in the area,

(ii)               the value is much higher than the detection limit,

(iii)             the value would be considered as a problem measurement by the two US studies as it is high and close to a shale gas well, even though it is completely natural,

(iv)             the value varies significantly from day to day, and

(v)               naturally high methane levels are found in areas close to shale gas wells precisely because of the shale gas resources underneath.

Hence it should not be a surprise that the distance from shale gas well graphs show more shale gas near shale gas wells. The question is how much is natural and how much is leakage, and to discover that you must have back-ground measurements at the location.

A fait accompli

When the three problems are combined, you know you are going to find raised chemical levels because the methodology guarantees it. It is a bit like testing for whether a farmer planted potatoes by searching for potatoes in a known potato field. One will, of course, find potatoes. However, that does not mean that they were planted by the farmer this year; they may have been growing there quite naturally for years. Nor does it mean that you can say that the farmer planted potatoes in other fields or not.

A better approach

As the BBC article points out, when studies are carried out randomly, the results are very different; in one USGS study 7 out of 20 randomly selected domestic water wells showed measurable dissolved methane, of which two contained sufficient methane to confirm its natural origin, and one in which methane was present in potentially explosive amounts. Despite this, none of the test water wells were close to gas wells in which hydraulic fracturing had been carried out.

Important points from the studies

However, despite the problems, the studies are very suggestive. Their main points are:

1. There is a tendency for groundwater to contain excess methane, ethane and propane close to shale gas wells. Currently, it is not known what proportion of this (if any) is due to leakage from shale gas wells and what is natural
2. The chemical signature of the gas in ground water is commonly biogenic, but sometimes is thermogenic, and a few samples share the same signature as the deep shale gas. However, the gas may have found its way naturally to the surface having taken millennia to do so.
3. There is an excess of four heavy metals (arsenic, strontium, barium and selenium) in the water wells. The cause is unknown, but is likely to be natural.
4. There is an excess of methanol and ethanol in the water wells of the University of Texas study. The cause is unknown, but is unlikely to be natural and may come from the fracking fluid.

But what of the UK?

The lack of background data in the USA is not the fault of the scientists. They were not forewarned of the development of the shale gas industry, and their studies are in a sense retrospective.

In the UK we have been warned, and there is no excuse for not having the background data to hand. In this respect Cuadrilla has acted very responsibly in commissioning an independent environmental monitoring company to carry out these measurements at Balcombe. The first report from Ground Gas Solutions has already been published and shows that

“The air and surface water samples generally show ambient conditions indicative of good environmental quality. Groundwater results have displayed some elevated heavy metals and sodium and has been confirmed to contain elevated dissolved methane and ethane gas.”

It seems unlikely, now, that hydraulic fracturing will be carried out at Balcombe, but if it were we would have the necessary data to test its effects.

Scientists are striving to provide data that will help us judge the impact of shale gas. The problem is complex, but we are slowly creeping towards a fuller understanding.