Hydraulic fracturing (hydrofracking) for natural gas has increased rapidly in the area of the Marcellus Shale in the last thirty years and estimates of CH4 emissions from hydrofracking operations are still uncertain. Previous studies on CH4 emissions at hydrofracking operations have used bottom-up approaches collected at discrete timepoints or discrete aerial surveys covering a wide spatial area, constraining the temporal scale of inference regarding these emissions. This project monitored atmospheric CH4 concentrations and stable carbon isotopes at a half-hourly temporal resolution from a 20-m tower downwind of a hydrofracking well pad in West Virginia for eighteen months. We collected four months of baseline observations prior to onsite well development to construct an empirical artificial neural-network model of baseline CH4 concentrations. We compared measured CH4 concentrations against the ANN-modeled CH4 baseline to identify CH4 concentration spikes that coincided with different stages of onsite well development, from the baseline period through fracking. CH4 concentration spikes were significantly more frequent than baseline conditions during the vertical drilling and fracking phases of operations. We found that the median magnitude of CH4 concentration spikes during the vertical drilling phase was 316% larger than that of the baseline phase, and the median magnitude of CH4 concentration spikes was 509% larger in the hydraulic stimulation (fracking) stage compared to the baseline phase. We also partitioned the sources of measured CH4 concentrations to biogenic ruminant and geologic shale gas isotopic signatures by measuring 13CH4 gas at high temporal resolution and using a source-partitioning 13CH4 model. The measured median value of half-hourly CH4 concentration spikes attributed to a geologic shale gas isotopic origin was 27% larger than the median CH4 concentration spikes attributed to ruminants, and the maximum half-hourly CH4 concentration spike attributed to shale gas was up to 179% higher than maximum CH4 concentration spike for ruminant-dominated half-hours. This study developed a framework for off-site, single tower measurements to identify CH4 concentration spikes associated with the phases of unconventional natural gas well development in a complex CH4 emissions airshed.
