RELATING CHAMBER MEASUREMENTS TO EDDY-CORRELATION MEASUREMENTS OF METHANE FLUX

Methane fluxes were measured using eddy correlation and chamber techniques during 1991 and 1992 at a peatland in north central Minnesota. Comparisons of the two techniques were made using averages of methane flux data available during I-week periods. The seasonal patterns of fluxes measured by the two techniques compared well. Chamber flux, in 1991, was about 1.8 mg m(-2) hr(-1) greater than the flux obtained by the eddy correlation technique. In 1992, the chamber flux was about 1.5 mg m(-2) hr(-1) higher than the eddy correlation flux prior to midseason and 1.0 mg m(-2) hr lower than the eddy correlation flux after midseason. Chamber data from individual hummock and hollow pairs were used to calculate the averaged dF/dZ (rate of change of methane flux with surface height). During midseason in 1991, the magnitude of dF/dZ ranged between 10 and 100 (mg m(-2) hr(-1)) m(-1). We speculate that high water table conditions caused a decrease in the magnitude of dF/dZ after midseason of 1992. As compared to 1991, greater variability of dF/dZ in 1992 probably resulted from less frequent sampling. To obtain a more valid comparison of the results from the two measurement techniques, chamber data were adjusted to account for the spatial variation in methane flux. Accordingly, the chamber flux values were ''scaled up'' using the dF/dZ values and distributions of surface heights representative of the footprint of the eddy correlation sensors. The scaling procedure reduced the chamber fluxes by an average of 1.8 mg m(-2) hr(-1) in 1991 and 1.0 mg m(-2) hr(-1) in 1992. The comparison of eddy correlation and chamber fluxes was improved both before and after midseason in 1991. The slope of the linear regression between eddy correlation and chamber fluxes decreased from 1.49 to 1.14 (r(2) increased from 0.53 to 0.75). During 1992, the scaling of chamber fluxes slightly improved their comparison with eddy correlation fluxes only prior to midseason. The lack of improvement after midseason in 1992 is likely the result of scaling assumptions when the water table was above the hollow surface. Results suggest that the adjustment of chamber flux data for spatial variations on microtopographical scales does provide fluxes more representative of a larger area. However, more information is needed on factors controlling spatial variation of methane flux to help refine the assumptions involved in the scaling procedure.