Abstract. Four in situ cavity ring-down spectrometers (G2132-i, Picarro, Inc.) measuring methane dry mole fraction (CH₄), carbon dioxide dry mole fraction (CO₂), and the isotopic ratio of methane (δ¹³CH₄) were deployed at four towers in the Marcellus Shale natural gas extraction region of Pennsylvania. In this paper, we describe laboratory and field calibration of the analyzers for tower-based applications and characterize their performance in the field for the period January–December 2016. Prior to deployment, each analyzer was tested using bottles with various isotopic ratios, from biogenic to thermogenic source values, which were diluted to varying degrees in zero air, and an initial calibration was performed. Furthermore, at each tower location, three field tanks were employed, from ambient to high mole fractions, with various isotopic ratios. Two of these tanks were used to adjust the calibration of the analyzers on a daily basis. We also corrected for the cross-interference from ethane on the isotopic ratio of methane. Using an independent field tank for evaluation, the standard deviation of 4 h means of the isotopic ratio of methane difference from the known value was found to be 0.26 ‰ δ¹³CH₄. Following improvements in the field tank testing scheme, the standard deviation of 4 h means was 0.11 ‰, well within the target compatibility of 0.2 ‰. Round-robin style testing using tanks with near-ambient isotopic ratios indicated mean errors of −0.14 to 0.03 ‰ for each of the analyzers. Flask to in situ comparisons showed mean differences over the year of 0.02 and 0.08 ‰, for the east and south towers, respectively.

Regional sources in this region were difficult to differentiate from strong perturbations in the background. During the afternoon hours, the median differences of the isotopic ratio measured at three of the towers, compared to the background tower, were &minus0.15 to 0.12 ‰ with standard deviations of the 10 min isotopic ratio differences of 0.8 ‰. In terms of source attribution, analyzer compatibility of 0.2 ‰ δ¹³CH₄ affords the ability to distinguish a 50 ppb CH₄ peak from a biogenic source (at −60 ‰, for example) from one originating from a thermogenic source (−35 ‰), with the exact value dependent upon the source isotopic ratios. Using a Keeling plot approach for the non-afternoon data at a tower in the center of the study region, we determined the source isotopic signature to be −31.2 ± 1.9 ‰, within the wide range of values consistent with a deep-layer Marcellus natural gas source.