Seasonal Variation of Methane Flux From a California Rice Paddy

To allow increased understanding of the global budget of atmospheric methane, individual methane sources require investigation. We have measured methane emissions from a California rice paddy during the entire 1982 growing season. A very strong seasonal dependence was observed. Methane emissions were highest in the last 2-3 weeks before harvest; daily emissions reached 5 g CH,(cid:127)/m 2. Over the 100-day season, daily emissions averaged about 0.25 g CH,(cid:127)/m 2, higher than our previously reported values. Attempts to estimate global rice paddy emissions must recognize the possibility of seasonal variations. Soil temperature at 10-cm depth correlated poorly with our measured fluxes; soil redox potential was a more reliable indicator.


INTRODUCTION
Deducing the identities and sizes of the sources of atmospheric methane has been a subject of interest to atmospheric chemists and geochemists for about 20 years. Carbon 14 data reviewed by Ehhalt and Schmidt [1978] demonstrate that over 80% of atmospheric methane is biogenic. Of this biogenic methane, 33-49% was attributed to release from the world's rice paddies. In making these estimates, Ehhalt and Schmidt were forced to employ estimates of methane emission rates from laboratory incubations of rice paddy soil [Koyama, 1963[Koyama, , 1964; no field measurements were available. The first such measurements in rice paddies were reported by Cicerone and Shetter [1981]. They found lower methane emission rates than Koyama measured in the laboratory. Cicerone and Shetter also reported that the principal means of escape of methane is through the rice plants and not through diffusion or escape of bubbles across the air-water interface and that nitrogen fertilization rates affect the methane escape.
Questions about methane sources have acquired new importance now that it has been shown that atmospheric methane concentrations are increasing globally [Rasmussen and Khalil, 1981;Blake et al., 1982;Craig and Chou, 1982;Ehhalt et al., 1983]. Thus it has become important to quantify the individual methane sources and the temporal changes in these sources, including the most recent suggestions of potentially significant sources, for example, termites [Zimmerman et al., 1982;Rasmussen and Khalil, 1983] and biomass burning [Crutzen et al, 1979].
In this paper we present new field measurements of methane emissions from rice paddies. The data show strong dependence of the methane emission rate on time elapsed in the growing season and an apparently complicated dependence on nitrogen fertilization rate, factors that must be recognized in any attempt to extrapolate to global emission rates. Field sampling was performed with a saran bag collector with stainless steel tubing and flasks as described by Cicerone and Shetter [1981]. The collectors were placed over the rice plants with the lower rim into the water surrounding the rice, and gas samples were extracted from inside the collectors after 15 min. All other analytical procedures were as those in the work of Cicerone and Shetter [1981] except that methane concentrations were determined by peak area (HP3390 miniintegrator) rather than by peak height. During each such collection for flux measurements, we also measured temperatures of ambient air, air inside the collector, soil (at 10-cm depth), and rice height. Soil redox potential, Eh, was measured by means of a platinum electrode referred to a calomel half cell.
We attempted to place the collectors over equal numbers of rice plants, but because counting stalks was difficult physically and because we sought to avoid bending and jostling the rice stalks, this goal was not attained reliably.

RESULTS AND DISCUSSION
Dates and times of methane flux measurements are shown in Table 1   Soil temperatures, redox potentials, and rice heights are also shown. Subplot P1 was more heavily fertilized than subplot 2 (see text). Numbers in parentheses indicate powers of 10; for example, read  Table 1. Note logarithmic scale. effects of maturation of rice? Does fertilizer accelerate the growth of roots and allow the cortex of roots to conduct gases earlier in the season? This last question relates to our finding that nitrogen fertilizer appears to stimulate methane release rates during early and mid-season but does not influence the cumulative amount of methane released over the entire growing season. Clearly, more data are needed to determine the range of N fertilizer effects. Further questions about the mechanism of methane escape and how to extrapolate field data to global conditions are apparent when one notices the lack of any clear correlation between methane flux and soil temperature at 10-cm depth (Table 1). Note, for example, that methane fluxes rose 10-100 fold during July and August with soil temperatures always near 23øC. On the contrary, $eiler et al. [1983] have found a positive correlation between methane emissions and soil temperature at 0.5-cm depth. It is important to determine if methane emissions to the atmosphere, like methane production in soils, rise with soil temperature at some specified soil depth. Table 1

does show a positive correlation between soil E h and methane emissions, with largest fluxes when E h • -190 mV. The onset of methanogenesis for E h •-200
mV has been observed previously in flooded soils; methanogenesis appears to follow the sequential usage of oxygen, ferric ion, nitrate, and sulfate [Mah et al., 1977].
Thus the question of the fraction of atmospheric methane due to releases from worldwide rice agriculture is far from settled. Our 100-day averages for methane fluxes of 0.28 and 0.22 g CH,•/m 2 d are 56% higher and 21% higher than the values reported by Cicerone and $hetter [1981]. Accordingly, one might extrapolate our 1982 California fluxes to proportionately higher global annual emissions of CH, from the world's rice paddies than the Cicerone and Shetter [1981] estimate. Even though our 1982 data set is much more complete than that of Cicerone and $hetter [1981], we are reluctant to present an estimate of a global emission rate because of many remaining uncertainties such as the effects mentioned above and the effects of varying agricultural practices and because of variable organic contents of soils.
Finally, we note that fall maxima have been observed in atmospheric methane concentrations . As these authors have discussed, seasonal variations in methane sinks and sources are likely to be involved, but our observed methane emission pattern can partially explain the fall maximum in methane concentration. Davis, for assistance and for access to his experimental rice paddies.