Fig. 3. Low production well site CH₄ emissions data as reported in previous studies.

a CH₄ emissions data for six basins with at least n > 5 observations shown as box plots (centerline, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range) and individual points (gray circles). Sample sizes are shown at the bottom of the plot. Only site-level measurements above method detection limits of 0.01–0.036 kg/h are shown. Appal.—Appalachian (Pennsylvania, Ohio, West Virginia); Delaware (Texas/New Mexico); Barnett (Texas); Uinta (Utah); UGR—Upper Green River (Wyoming); DJB—Denver-Julesburg Basin (Colorado). Low production well site data were a subset of site-level measurements reported by: Robertson et al¹³., Robertson et al²¹., Caulton et al¹²., Omara et al¹⁰., Omara et al¹⁷. and Brantley et al²⁰. b Relationship between measured site-level CH₄ emissions and O&G production in barrels of oil equivalent per day (boed). The plot shows the top 5% of high-emitting sites (n = 12, green symbols), the bottom 95% of sites (n = 192, blue symbols), and below-detection-limit (BDL) sites (n = 36, gray symbols). Each site’s CH₄ loss rate is indicated by the size of the circles. Oil-only sites or sites with reported CH₄ loss rates >100% are assigned values of 100%. The orange and pink symbols represent the mean wellhead-only CH₄ and O&G production for low production sites sampled in Ohio²² and West Virginia³¹. Data from these two studies were not used in emission models because they exclude other sources such as tanks and separators, but are shown here to illustrate that wellhead-only CH₄ emissions can be significant even at low production well sites. The solid-dark red line shows the nonparametric Bayesian regression model for the bottom 95% of sites (see Methods).