Purging and other sampling variables affecting dissolved methane concentration in water supply wells

Determining whether changes in groundwater methane concentration are naturally occurring or related to oil and gas operations can be complicated by numerous sources of variability. This study of 10 residential water supply wells in Northeastern Pennsylvania evaluates how i) sampling from different points within the water well system, ii) purging different water volumes prior to sampling, and ii) natural variation over time, affects concentrations of naturally occurring dissolved methane and other water quality parameters. Among the population of wells, all had dissolved methane concentrations >1mg/L. Regardless of the volume of water purged or the timing between events, the maximum change in methane concentration (ratio of maximum to minimum concentration) among samples from a single well was 3.2, with eight out of ten wells exhibiting a maximum change less than a factor of two (i.e., <±100%). Among water wells where methane concentration changed by ±50% or more, there was a strong correlation with changes in the concentrations of sodium, chloride, and other salinity indicators such as specific conductivity and TDS. This suggests that significant variability in methane concentration is predominantly related to changes in the relative volumes of sodium-rich fluids feeding the wellbore at any given time. Among study well locations with bladder and diaphragm pressure tanks, there was no significant difference in dissolved methane concentrations between samples collected either upstream or downstream of a pressure tank. There appears to be little benefit to purging multiple casing volumes of water from a well prior to sampling because such volumes tend to be much larger than those representative of normal residential use. We recommend purging a volume sufficient to remove standing water in the pressure tank and lines above the pump intake. This article culminates with additional recommendations for improving sample collection methods and interpreting sampling data.