Mapping of North American methane emissions with high spatial resolution by inversion of SCIAMACHY satellite data



We estimate methane emissions from North America with high spatial resolution by inversion of Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY) satellite observations using the Goddard Earth Observing System Chemistry (GEOS-Chem) chemical transport model and its adjoint. The inversion focuses on summer 2004 when data from the Intercontinental Chemical Transport Experiment-North America (INTEX-A) aircraft campaign over the eastern U.S. are available to validate the SCIAMACHY retrievals and evaluate the inversion. From the INTEX-A data we identify and correct a water vapor-dependent bias in the SCIAMACHY data. We conduct an initial inversion of emissions on the horizontal grid of GEOS-Chem (1/2° × 2/3°) to identify correction tendencies relative to the Emission Database for Global Atmospheric Research (EDGAR) v4.2 emission inventory used as a priori. We then cluster these grid cells with a hierarchical algorithm to extract the maximum information from the SCIAMACHY observations. A 1000 cluster ensemble can be adequately constrained, providing ~100 km resolution across North America. Analysis of results indicates that the Hudson Bay Lowland wetlands source is 2.1 Tg a−1, lower than the a priori but consistent with other recent estimates. Anthropogenic U.S. emissions are 30.1 ± 1.3 Tg a−1, compared to 25.8 Tg a−1 and 28.3 Tg a−1 in the EDGAR v4.2 and Environmental Protection Agency (EPA) inventories, respectively. We find that U.S. livestock emissions are 40% greater than in these two inventories. No such discrepancy is apparent for overall U.S. oil and gas emissions, although this may reflect some compensation between overestimate of emissions from storage/distribution and underestimate from production. We find that U.S. livestock emissions are 70% greater than the oil and gas emissions, in contrast to the EDGAR v4.2 and EPA inventories where these two sources are of comparable magnitude.