Methane flux in a boreal fen: Season-long measurement by eddy correlation

Eddy correlation measurements of methane flux were made at a fen in central Saskatchewan, as part of the Boreal Ecosystem Atmosphere Study (BOREAS) in 1994. Data were collected from mid-May to early October. The water table was above the average peat surface throughout the measurement period. Detailed (near continuous) measurements allowed examination of temporal variability at hourly and daily timescales. As compared with the average nighttime flux, the average daytime methane flux was 25-45% higher in July and in August and 5-15% higher earlier and later in the season. Distribution of midday (1130-1430 LT) methane emission showed varying trends in different parts of the season. From mid-May to early July, methane flux gradually increased from near zero to 4.1 mg m(-2) h(-1). The water table height (above an average hollow surface) varied from 0.09 to 0.18 m, but the trend in methane flux followed peat temperature (at 0.1-m depth) more closely. The peat warmed from 3.4 degrees to 16.3 degrees C during this time period. Methane flux was negligible until peat temperature (at 0.1-m depth) was above 12 degrees C. From early July to early August there was a fivefold increase in methane flux from 4.1 to its seasonal peak of 19.5 mg m(-2) h(-1) on August 1. The water table ranged from 0.23 m to a brief seasonal plateau of 0.30 m on July 20-23. Sharp increases in the water table were followed by increasing trends in methane flux by approximately 12 days. Feat temperature reached its seasonal maximum (19.3 degrees C) the same time when the methane flux peaked. After early August the methane flux declined steadily and reached a value of 2.4 mg m(-2) h(-1) in early October. The water table and peat temperature showed comparable trends and decreased steadily to 0.06 m and 5.7 degrees C, respectively. The seasonally integrated methane emission (mid-May to early October) was estimated at 16.3 g C m(-2). Nonlinear regression analysis of methane flux against water table (lagged by 12 days) and peat temperature was performed for different periods of the season. Except for a brief period of very high water table (when many hummocks were inundated) the regression using water table and peat temperature explained between 68 and 94% of the variability in methane flux. The sensitivity of methane flux to water table (or the slope of the log CH4 flux/water table relationship) obtained from our daily flux values ranged from 3.4 x 10(-4) to 5.0 x 10(-4) m(-1).