Fig. 4 Example decomposition of the equipment-level emission factor for gas wellheads. This study ’ s equipment-level emission factor ( d ) for Western natural gas system wellheads is decomposed into constituent parts and compared with the Greenhouse Gas Inventory (GHGI). Error bars re fl ect the 95% con fi dence interval based on the 2.5 and 97.5 percentile values extracted from the empirical distributions and fi lled squares and triangles represent the mean. Constituent parts include component-level emission factors ( a ), fraction of components emitting ( b ), and component counts ( c ). When multiplied together, these factors have counteracting biases, with component-level emission factors and component counts contributing to higher emissions in our study versus the GHGI, and fraction of components emitting contributing to lower emissions in our study (Note that units differ for each panel, and also the logarithmic scale meaning that visible differences between points often span orders of magnitude). For illustrative purposes, there are several limitations to what is included in our decomposition plots. First, here we only show constituent data for Western natural gas systems; results for Eastern natural gas system are reported in Supplementary Methods 6 (Note that in actual usage in the GHGI, equipment-level emission factors for natural gas systems are a weighted average of both Western systems (API 4589 35 ) and Eastern systems (Star Environmental, 47 )). Second, we also limit this fi gure to connectors, valves, and open-ended lines (which account for the majority of components although our inventory and the GHGI also account for pressure relief valves, compressor seals, and other components in smaller numbers). Finally, decomposition plots are limited to component-level emission factors and fraction