A Pennsylvania State University study that utilized reams of data from producers, the state and the U.S. Geological Survey, found little evidence of natural gas contaminating groundwater in Bradford County, a hotbed for unconventional drilling.

An interdisciplinary team of geoscientists and computer scientists used new data-mining techniques and computer models to study 11,000 groundwater samples from the 2010s taken after natural gas drilling had occured in the rural area. They found “possible” methane contamination in groundwater near seven of the 1,385 Marcellus Shale wells studied.

The rest of the water chemistry data, researchers said in findings recently reported in the journal Environmental Science & Technology, shows that groundwater “might even be improving” or has remained at levels similar to samples taken prior to the 1990s, long before unconventional development began in the region, or anywhere else in the country.

Bradford County is among the state’s most heavily developed shale regions. Along with Susquehanna, Washington and Greene counties, they have consistently accounted for about two-thirds of all unconventional production in the state. Nearly 4,000 shale permits have been issued in Bradford, where producers reported natural gas volumes of 183.8 Bcf last year, according to Pennsylvania Department of Environmental Protection data.

Unconventional wells are drilled thousands of feet underground, far below drinking water. Dozens of studies across the country have examined water resources in regions of dense oil and gas development, with little evidence to suggest any natural gas or additive migration from deep underground. Risks of contamination, however, are more likely to stem from other sources, such as faulty steel or cement casing.

Methane is also naturally occuring underground and is present near geological features like fault lines and valleys. The researchers searched for patterns in methane concentrations, looking for any indications of contamination from shale wells that use high-volume hydraulic fracturing to extract gas. The data-mining techniques helped the researchers sift through large amounts of information to find elevated methane levels and possible sources.

“We look at methane in this groundwater data and we actually can see geology in the data,” said Susan Brantley, director of Penn State’s Earth and Environmental Systems Institute (EESI), in a statement after the study was released. “I can tell you where there are faults, where there are big folds in the rock based on the groundwater. Then what we look for is where the water chemistry does not look like the geology and looks like something we don’t understand. Our hypothesis is sometimes that shows you something about human activities, like shale development.”

There were also conventional wells in the area where methane was high near the seven shale wells. Tao Wen, a post-doctoral scholar at the EESI, said those sites should be sampled and studied further in the field.

Interestingly, the researchers said the study could help document how groundwater has improved since the implementation of the Clean Air Act in the 1970s and the reduction of atmospheric deposition caused by things like steel production or coal burning. The authors also analyzed iron, manganese, total dissolved solids, sulfate and pH levels, finding that they have either improved or remain the same since samples were taken in the 1980s.

Marcellus Shale Coalition President David Spigelmyer said the study reflects the industry’s commitment to reducing environmental impacts, and praised its objective and methods. The data and techniques used have been made available online. The research was supported by the National Science Foundation Interdisciplinary Research and Education program.