CarbonTracker-CH4: an assimilation system for estimating emissions of atmospheric methane
- 1NOAA Earth System Research Laboratory, Global Monitoring Division, Boulder Colorado, USA
- 2National Institute for Environmental Studies, Center for Global Environmental Research, Tsukuba, Japan
- 3Environment Canada, Climate Research Division, Toronto, Ontario, Canada
- 4Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA
Abstract. We describe an assimilation system for atmospheric methane (CH4), CarbonTracker-CH4, and demonstrate the diagnostic value of global or zonally averaged CH4 abundances for evaluating the results. We show that CarbonTracker-CH4 is able to simulate the observed zonal average mole fractions and capture inter-annual variability in emissions quite well at high northern latitudes (53–90° N). In contrast, CarbonTracker-CH4 is less successful in the tropics where there are few observations and therefore misses significant variability and is more influenced by prior flux estimates. CarbonTracker-CH4 estimates of total fluxes at high northern latitudes are about 81 ± 7 Tg CH4 yr−1, about 12 Tg CH4 yr−1 (13%) lower than prior estimates, a result that is consistent with other atmospheric inversions. Emissions from European wetlands are decreased by 30%, a result consistent with previous work by Bergamaschi et al. (2005); however, unlike their results, emissions from wetlands in boreal Eurasia are increased relative to the prior estimate. Although CarbonTracker-CH4 does not estimate an increasing trend in emissions from high northern latitudes for 2000 through 2010, significant inter-annual variability in high northern latitude fluxes is recovered. Exceptionally warm growing season temperatures in the Arctic occurred in 2007, a year that was also anonymously wet. Estimated emissions from natural sources were greater than the decadal average by 4.4 ± 3.8 Tg CH4 yr−1 in 2007.
CarbonTracker-CH4 estimates for temperate latitudes are only slightly increased over prior estimates, but about 10 Tg CH4 yr−1 is redistributed from Asia to North America. This difference exceeds the estimated uncertainty for North America (±3.5 Tg CH4 yr−1). We used time invariant prior flux estimates, so for the period from 2000 to 2006, when the growth rate of global atmospheric CH4 was very small, the assimilation does not produce increases in natural or anthropogenic emissions in contrast to bottom-up emission data sets. After 2006, when atmospheric CH4 began its recent increases, CarbonTracker-CH4 allocates some of the increases to anthropogenic emissions at temperate latitudes, and some to tropical wetland emissions. For temperate North America the prior flux increases by about 4 Tg CH4 yr−1 during winter when biogenic emissions are small. Examination of the residuals at some North American observation sites suggests that increased gas and oil exploration may play a role since sites near fossil fuel production are particularly hard for the inversion to fit and the prior flux estimates at these sites are apparently lower and lower over time than what the atmospheric measurements imply.
The tropics are not currently well resolved by CarbonTracker-CH4 due to sparse observational coverage and a short assimilation window. However, there is a small uncertainty reduction and posterior emissions are about 18% higher than prior estimates. Most of this increase is allocated to tropical South America rather than being distributed among the global tropics. Our estimates for this source region are about 32 ± 4 Tg CH4 yr−1, in good agreement with the analysis of Melack et al. (2004) who obtained 29 Tg CH4 yr−1 for the most productive region, the Amazon Basin.