Abstract
Methane (CH4) plays important roles in atmospheric chemistry and short-term forcing of climate. A clear understanding of atmospheric CH4's budget of emissions and losses is required to aid sustainable management of Earth's future environment. We used an atmospheric chemistry-transport model (JAMSTEC's ACTM) for simulating atmospheric CH4. A global inverse modeling system has been developed for estimating CH4 emissions from 53 land regions for 2002-2012 using measurements at 39 sites. An ensemble of 7 inversions is performed by varying a priori emissions. Global net CH4 emissions varied between 505-509 and 524-545 Tg yr(-1) during 2002-2006 and 2008-2012, respectively (ranges based on 7 inversion cases), with a step like increase in 2007 in agreement with atmospheric measurements. The inversion system did not account for interannual variations in OH radicals reacting with CH4 in the atmosphere. Our results suggest that the recent update of the EDGAR inventory (version 4.2FT2010) overestimated the global total emissions by at least 25 Tg yr(-1) in 2010. The increase in CH4 emission since 2004 originated in the tropical and southern hemisphere regions, coinciding with an increase in non-dairy cattle stocks by similar to 10 % from 2002 (with 1056 million heads) to 2012, leading to similar to 10 Tg yr(-1) increase in emissions from enteric fermentation. All 7 ensemble cases robustly estimated the interannual variations in emissions, but poorly constrained the seasonal cycle amplitude or phase consistently for all regions due to the sparse observational network. Forward simulation results using both a priori and a posteriori emissions are compared with independent aircraft measurements for validation. Based on the results of the comparison, we reject the upper limit (545 Tg yr(-1)) of global total emissions as 14 Tg yr(-1) too high during 2008-2012, which allows us to further conclude that the increase in CH4 emissions over the East Asia (mainly China) region was 7-8 Tg yr(-1) between the 2002-2006 and 2008-2012 periods, contrary to 1-17 Tg yr(-1) in the a priori emissions.