Bayesian-inversion adjusted methane fluxes in Colombia and Panama

Methane is the second most important long lived greenhouse gas (GHG) in the Earth's atmosphere accounting for ~20% of the positive radiative forcing. The first step towards developing GHG mitigation strategies is to obtain sufficiently accurate and detailed source and sinks estimations. While ~2/3 of the global methane emissions are anthropogenic, the wetlands are the single largest source. Therefore, in many cases, wetland emissions must be included in inverse modeling calculations aimed at validating anthropogenic emission inventories from ambient air concentration measurements. High accuracy and precision methane measurements carried out in 2007 during NASA's TC4 mission revealed elevated enhancements over Colombia and Panama (up to ~500 ppbv CH4 over Uraba, Colombia). Aiming at identifying the origin of these enhancements and at validating the anthropogenic emission inventory, we used STILT to estimate methane mixing ratios based on surface fluxes at regional level over four regions of both Colombia and Panama. STILT was applied along with assimilated (GDAS and ECMWF) meteorological fields and a priori methane inventories for anthropogenic (EDGAR) and wetland emissions (Kaplan's and Matthews and Fung's). The modeled mixing ratios were compared to the TC4 mission measurements. A Bayesian inversion analysis allowed us to scale prior fluxes taking into account the uncertainty on modeled mixing ratios due to transport errors, which were calculated by comparison with meteorological observations. We obtained flux scaling factors for the whole domain of study and for each one of the four regions. Overall, the Bayesian inversion indicates that the prior anthropogenic inventory is reasonably accurate and the a priori wetland methane fluxes are overestimated almost by a factor 2. Although the posterior enhancements show a better agreement with measurements, the discrepancies cannot be reduced for 4 regions simultaneously, which points to the calculated meteorological fields as the main source of uncertainty. This is not surprising bearing in mind the complexity of horizontal and vertical transport in the Intertropical Convergence Zone, particularly over the strong moist convection period experienced during the TC4 mission. Our calculations show a significant anthropogenic influence in the Savanna and Uraba regions. This is unexpected for Uraba. Our posterior wetland flux estimate for Colombia is ~4 Tg/year. This is relatively large when compared for instance with the Hudson Bay Lowlands (~ 2 Tg/year).