A gridded version of the US EPA inventory of methane emissions for use as a priori and reference in methane source inversions

Abstract:

The US EPA produces annual estimates of national anthropogenic methane emissions in the Inventory of US Greenhouse Gas Emissions and Sinks (EPA inventory). These are reported to the UN and inform national climate policy. The EPA inventory uses best available information on emitting processes (IPCC Tier 2/3 approaches). However, inversions of atmospheric observations suggest that the inventory could be too low. These inversions rely on crude bottom-up estimates as a priori because the EPA inventory is only available as national totals for most sources. Reliance on an incorrect a priori greatly limits the value of inversions for testing and improving the EPA inventory as allocation of methane emissions by source types and regions can vary greatly between different bottom-up inventories.

Here we present a 0.1° × 0.1° monthly version of the EPA inventory to serve as a priori for inversions of atmospheric data and to interpret inversion results. We use a wide range of process-specific information to allocate emissions, incorporating facility-level data reported through the EPA Greenhouse Gas Reporting Program where possible. As an illustration of used gridding strategies, gridded livestock emissions are based on EPA emission data per state, USDA livestock inventories per county, and USDA weighted land cover maps for sub-county localization. Allocation of emissions from natural gas systems incorporates monthly well-level production data, EIA compressor station and processing plant databases, and information on pipelines. Our gridded EPA inventory shows large differences in spatial emission patterns compared to the EDGAR v4.2 global inventory used as a priori in previous inverse studies. Our work greatly enhances the potential of future inversions to test and improve the EPA inventory and more broadly to improve understanding of the factors controlling methane concentrations and their trends. Preliminary inversion results using GOSAT satellite data will be presented.