JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 98, NO. Dll, PAGES 20,649-20,655, NOVEMBER 20, 1993 Seasonal Distribution of Methane Flux in a Minnesota Peatland Measured by Eddy Correlation N.J. SHURPALI, S. B. VERMA, AND R. J. CLEMENT Department of Agricultural Meteorology, University of Nebraska, Lincoln D. P. BILLESBACH Department of Electrical Engineering, University of Nebraska, Lincoln Methane flux was measured using the micrometeorological eddy correlation technique during 62 days in mid-May through mid-October 1991 in a peatland ecosystem in north central Minnesota. Application of this technique allows measurement of spatially integrated fluxes. The distribution of methane flux consisted of a gradual pattern with several episodicemissions superimposed.The gradual -2 (nonepisodic) pattern of methane flux (daytime average) exhibited an increase from 30-120 mg m d-1 in late Ma to early July to 125-160 mg m-2 d-1 in mid-July to mid-August and then a decline to 100-35 mg m -[ d-1 in early September to mid-October. Peat temperature (at0.1 m depth) and water table depth accounted for about 70% of the variance in the methane flux data. The episodic emissions were associatedwith drops in atmospheric pressure and a declining water table. Generally, the hourly values of daytime methane flux were fairly stable, perhaps with a slight depressionduring the midday. However, on days with episodic emissions, the daytime methane flux had a peak during the midafternoon, and its pattern appeared to be similar to those of the standard deviation of atmospheric pressure fluctuations and mean horizontal wind velocity. The ratio of soil CO2 to methane fluxes (during nonepisodic emission periods) increased with increasing water table depth in a manner similar to that observed in a laboratory study of peat columns from different wetland types in Quebec. The annual methane emission for this ecosystem was estimated to be about 16-19.5 g m-2 yr -1. INTRODUCTION Northern wetlands are consideredto be a major biological source of atmospheric methane [e.g., Matthews and Fung, 1987; Cicerone and Oremland, 1988]. Methane fluxes have generally been measured using chambers. Chamber methods are limited to small areas (• 1 m2). Application of microme- teorological techniques for larger-scale, spatially integrated measurements has been highly recommended. Micrometeo- rological methods cause minimal disturbance to the microen- vironment and allow continuous flux measurements. During the summer of 1990, a pilot study was conductedat a peatland site in north central Minnesota. Methane flux was measured using the micrometeorological eddy correlation technique with a prototype tunable diode laser spectrometer (TDLS). The results of this pilot study [Verma et al., 1992] demonstrated the utility of this technique for measuring surface fluxes of methane. Here we report the results of similar eddy correlation measurements of methane flux made during a 5-month period (May-October, 1991) at the same site. The objective of this researchwas to study the seasonal variability of methane flux in this ecosystemand investigate the dependence of the methane flux on relevant environmen- tal variables. MATERIALS AND METHODS Site The study site, located in the Chippewa National Forest, adjacent to the Marcell Experimental Forest (47ø32'N, Copyright 1993 by the American Geophysical Union. Paper number 93JD02181. 0148-0227/93/93 JD-02181 $05.00 93ø28'W) in north central Minnesota, is characterized as a poor fen. The site is referred to as Bog Lake Peatland. The peatland is poorly minerotrophic to oligotrophic [e.g., In- gram, 1983], with its water having a specific conductanceof 36 /aS and a pH of 4.6. The Bog Lake Peatland receives water from the adjoining, higher peatlands and from the surrounding mineral soil. There is no surface water outlet, but water does seep underground from the peatland's edge near its southern end. The microtopography at the site consists of a pattern of hummocks and hollows. Mean sea level elevation is about 416 m. The average annual precipi- tation is about 770 mm, and the average annual temperature is 3øC. The moss surface is represented by Sphagnum papillo- sum, a major carpet-forming species. Emergent species are dominated by Rhynchospora alba (beak rush) and Scheuchz- eria palustris (a triseeded arrow grass). The other commonly occurring species are Carex spp. (sedges), Sararacenia purpurea (pitcher plant), Chamaedaphne calyculata (leather leaf) and the 1- to 3-m tall, widely scattered Larix laricina (Tamarack). Eddy Correlation Measurements Fluxes of methane, sensible heat, water vapor, and mo- mentum were measured using the eddy correlation system described by Verma et al. [1992]. The eddy correlation system consisted of one-dimensional sonic anemometers (Campbell Scientific, Logan, Utah), a three-dimensional sonic anemometer (Kaijo Denki Company, Tokyo, Japan), fine-wire thermocouples, krypton hygrometers (Campbell Scientific) and a tunable diode laser spectrometer (Unisearch Associates, Incorporated of Concord, Ontario, Canada). The TDLS was used to measure the fluctuations in methane 20,649