JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 95, NO. D10, PAGES 16,773-16,788, SEPTEMBER 20, 1990 Methane Flux From the Amazon River Floodplain' Emissions During Rising Water KAREN B. BARTLETT, 1'2PATRICK M. CRILL, 2'3 JOSEA. BONASSI, 4 JEFFREY E. RICHEY, 5 AND ROBERT C. HARRISS 2,6 During April and May of 1987, an extensive methane flux data set from Amazonian wetland habitats was collected during the wet seasonas river water levels were high and rising. This work extends measurements made in the dry seasonof 1985, when water levels were falling. Comparison of these two data setsallows estimation of the amount of seasonal variability in this globally significantsource of tropospheric methane. A total of 284 flux measurements were made in the three primary floodplain environments of open-water lakes and channels, floating grass mats, and flooded forests, along approximately 1500km of the central floodplain. Emissions(means and standarderrors) were 74 _ 14 mg CH4/m2/d (open water), 201 _+ 35 mg CH4/m2/d (grass mats), and 126 _+ 20mg CH4/m2/d (flooded forests). These values were not significantlydifferent from the majority of those from 1985, in part due to the high variability in flux seen at both times. Although ebullition was a significantcomponent of methane emissionsat both periods, the frequency of bubbling and its contribution to total flux was lower during the period of rising water than during falling water. A prominent diurnal pattern in atmospheric methane concentrations was observed, with minimumlevels of about 1.75 ppm at midday and a maximum of 2.12 ppm at about midnight. Given the relatively small seasonal changes observed in flux at the two stagesof the river's hydrographic curve, our earlier estimates of regional methane flux remain largely unchanged. Revision of global estimates of wetland methane sources based on these tropical data and recently published figures for northern peatlands indicated that tropical wetlands may be more important than previously suggested, but that wetland sources overall remain at approximately 110 Tg/yr. INTRODUCTION In recent years, research on global sources and sinks of methane gas (CH4) to the atmosphere has accelerated sharply. Impetus for much of this effort was the discovery in the late 1970s and early 1980s that concentrations of tropo- spheric methane were increasing rapidly on a worldwide basis [Graedel and McRae, 1980; Rasmussen and Khalil, 1981; Blake et al., 1982]. Increases in isotopically biogenic sources of the gas under anthropogenic control such as rice paddies and ruminant populations are now thought to be largely responsible for the rapid rise in concentration ob- served over about the last 200 years [Craig et al., 1988; Rasmussen and Khalil, 1984]. Recent estimates of the con- tribution of isotopically abiogenic anthropogenic sources range from 13% to 32% of the total source [Quay et al., 1988; Lowe et al., 1988; Cicerone and Oremland, 1988]. Global budget calculations indicate that natural wetlands make a substantial contribution to tropospheric methane. Current estimates assessing the global distribution of wetlands and using published wetlands emission rates calculate the wet- land methane source at approximately 110 Tg/yr, or about 1Biology Department, College of William and Mary,Williams- burg, Virginia. 2Nowat Institute for the Study of Earth,Oceans, andSpace, Science and Engineering Research Building, University of New Hampshire, Durham. 3Chemistry Department, College of William and Mary, Williams- burg, Virginia. 4Centro deEnergia Nuclear naAgricultura, Universidade deS•,o Paulo, Brazil. 5School of Oceanography, University of Washington, Seattle. 6Atmospheric Sciences Division, NASALangley Research Cen- ter, Hampton, Virginia. Copyright 1990by the American GeophysicalUnion. Paper number 90JD00564. 0148-0227/90/90JD-00564505.00 25% of the global biogenic source [Matthews and Fung, 1987; Khalil and Rasmussen, 1983; Seiler, 1984; Ehhalt, 1985; Cicerone and Oremland, 1988]. Calculations of the magnitude of the global wetlands contribution to atmospheric methane are difficult due to the temporal and spatial variability of sources. Emissions are influenced by a wide array of site-specific environmental factors and change on time scales varying from diurnal through seasonal [e.g., Crill et al., 1988a; Bartlett et al., 1987; Holzapfel-Pschorn and Seiler, 1986; Harriss et al., 1982; Svensson and Rosswall, 1984; Wilson et al., 1989]. Accurately capturing this variability in extrapolation to global-scale emissions requires extensive field measure- ments on the appropriate scales. Such data are available for only a limited number of environments. Of particular impor- tance are northern wetlands (50ø-70øN) and tropical to subtropical wetlands (20øN-30øS). These areas were calcu- lated to supply 57% and 29% of the total wetland methane figure of 110 Tg, respectively [Matthews and Fung, 1987]. The tropical emissions estimates used in these calculations were a large source of possible error as there were no data available when this calculation was made. The first extensive data sets on methane emissions from tropical wetlands were reported in 1988 in a series of papers on the Amazon River basin [Bartlett et al., 1988; Crill et al., 1988a;Devol et al., 1988]. Emission rates published in these papers suggest that rate estimates used by Matthews and Fung in their preliminary calculations for the tropics may have been too low. The reported measurements were con- ducted as part of the NASA Global Tropospheric Experi- ment's Amazon Boundary Layer Experiment/Amazon Ground Emissionsprogram (GTE ABLE 2A/AGE) and were made over roughly a 2-month period (July-August 1985) during the early part of the annual dry seasonin the region. Together these papers addressed variability in flux on a variety of spatial scales ranging from within a single habitat 16,773