Improved Understanding of Methane Emissions by Combination of Bottom-Up and Top-Down Methods
Maasakkers, Joannes Dyonisius
MetadataShow full item record
CitationMaasakkers, Joannes Dyonisius. 2018. Improved Understanding of Methane Emissions by Combination of Bottom-Up and Top-Down Methods. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
AbstractWe present a gridded inventory of US anthropogenic methane emissions with 0.1º x0.1º spatial resolution, monthly temporal resolution, and detailed scale-dependent error characterization. The inventory is designed to be consistent with the 2016 US Environmental Protection Agency (EPA) Inventory of US Greenhouse Gas Emissions and Sinks (GHGI) for 2012. The EPA inventory is available only as national totals for different source types. We use a wide range of databases at the state, county, local, and point source level to disaggregate the inventory and allocate the spatial and temporal distribution of emissions for individual source types. Results show large differences with the EDGAR v4.2 global gridded inventory commonly used as a priori estimate in inversions of atmospheric methane observations. We derive grid-dependent error statistics for individual source types from comparison with the Environmental Defense Fund (EDF) regional inventory for Northeast Texas. These error statistics are independently verified by comparison with the California Greenhouse Gas Emissions Measurement (CALGEM) grid-resolved emission inventory. Our gridded, time-resolved inventory provides an improved basis for inversion of atmospheric methane observations to estimate US methane emissions and interpret the results in terms of the underlying processes.
We use 2010-2015 observations of atmospheric methane columns from the GOSAT satellite instrument in a global inverse analysis to improve estimates of methane emissions and their trends over the period, as well as the global concentration of tropospheric OH (the main methane sink) and its trend. Our inversion solves the Bayesian optimization problem analytically including closed-form characterization of errors. This allows us to (1) quantify the information content from the inversion towards optimizing methane emissions and its trends, (2) diagnose error correlations between constraints on emissions and OH concentrations, and (3) generate a large ensemble of solutions testing different assumptions in the inversion. We show how the analytical approach can be used even when prior error distributions are log-normal. Inversion results show large overestimates of Chinese coal emissions and Middle East oil/gas emissions in the EDGAR v4.3.2 inventory, but little error in the US where we use a new gridded version of the EPA national greenhouse gas inventory as prior estimate. Oil/gas emissions in the EDGAR v4.3.2 inventory show large differences with national totals reported to the United Nations Framework Convention on Climate Change (UNFCCC) and our inversion is generally more consistent with the UNFCCC data. The observed 2010-2015 growth in atmospheric methane is attributed mostly to an increase in emissions from tropical wetlands, India, and China. The contribution from OH trends is small in comparison. We find that the inversion provides strong independent constraints on global methane emissions (552 Tg a-1) and global mean OH concentrations (atmospheric methane lifetime against oxidation by tropospheric OH of 10.7 ~ 0.4 years), indicating that satellite observations of atmospheric methane could provide a proxy for OH concentrations in the future.
Citable link to this pagehttp://nrs.harvard.edu/urn-3:HUL.InstRepos:41121226
- FAS Theses and Dissertations