An intercomparison study of two methods for estimating forest soil CO2 effl
ux was conducted during a 3-week period in summer. An empirical model estab
lished from nearly 2 years of chamber measurements predicted that the soil
CO2 efflux ranged from 1.2 to 1.4 mu mol m(-2) s(-1). This small range was
due to the lack of variability in the model parameters (soil temperature an
d moisture) during the study period. Eddy covariance measurements showed mo
re variability and diurnal dependence. Turbulent fluxes of CO2 during the d
ay apparently were influenced by the presence of a moss layer on the forest
floor; the combined effects of moss photosynthesis and respiration reduced
the turbulent fluxes by an average of 0.6 mu mol m(-2) s(-1) relative to t
he modeled soil efflux. At night, the eddy fluxes of CO2 agreed well with t
he modeled soil efflux; however, the turbulent fluxes were highly variable
(standard deviation exceeding the mean), due to the imperfect sampling cond
itions associated with the nocturnal boundary layer. These results illustra
te the different processes measured by the two methods, and highlight some
of the limitations of the eddy covariance technique for estimating soil CO2
efflux. Finally, they demonstrate the need for more long-term intercompari
son studies, covering a broader range in soil temperature and moisture. (C)
2001 Elsevier Science B.V. All rights reserved.