Evaluating year-to-year anomalies in tropical wetland methane emissions using satellite CH^sub 4^ observations

Natural wetlands are the largest source of methane emissions, contributing 20–40% of global emissions and dominating the inter-annual variability. Large uncertainties remain on their variability and response to climate change. This study uses atmospheric methane observations from the GOSAT satellite to evaluate methane wetland emission estimates. We assess how well simulations reproduce the observed methane inter-annual variability by evaluating the detrended seasonal cycle. The latitudinal means agree well but maximum differences in the tropics of 28.1–34.8 ppb suggest that all simulations fail to capture the extent of the tropical wetland seasonal cycle. We focus further analysis on the major natural wetlands in South America: the seasonally flooded savannah of the Pantanal (Brazil) and Llanos de Moxos (Bolivia) regions; and the riverine wetlands formed by the Paraná River (Argentina). We see large discrepancies between simulation and observation over the Pantanal and Llanos de Moxos region in 2010, 2011 and 2014 and over the Paraná River region in 2010 and 2014. We find highly consistent behaviour between the time and location of these methane anomalies and the change in wetland extent, driven by precipitation related to El Niño Southern Oscillation activity. We conclude that the inability of land surface models to increase wetland extent through overbank inundation is the primary cause of these observed discrepancies and can lead to under-estimation of methane fluxes by as much as 50% (5.3–11.8 Tg yr−1) of the observed emissions for the combined Pantanal and Paraná regions. As the hydrology of these regions is heavily linked to ENSO variability, being able to reproduce changes in wetland behaviour is important for successfully predicting their methane emissions.