In situ comparison of four approaches to estimating soil CO2 efflux in a northern larch (Larix kaempferi Sarg.) forest

Soil surface CO2 efflux (F-c) in a 45-year-old northern larch forest was continuously measured, using a multichannel automated chamber system, between June and October, 2001. The results were compared with periodic measurements obtained at the same site with an LI-6400 chamber system and an open-top chamber system, and also with continuous measurements obtained at the same site with a soil CO2 gradient system. The diurnal and seasonal changes in F-c measured by the automated chamber and Soil CO2 gradient approaches followed the soil temperature patterns. We found only a weak correlation between F-c and soil moisture, probably because the volumetric soil moisture (30-40% at 95% confidence interval) in this larch forest favors microbial and root activities. Among the four approaches, the LI-6400 chamber yielded a substantially higher F-c value than the other three approaches. The open-top chamber gave higher values of F-c than the automated chamber at efflux rates below 3.0 mumol m(-2) s(-1) and lower values above this level; however, the results of both these approaches were closely correlated (R-2 = 0.87, n = 105). Although F-c measured with the Soil CO2 gradient approach was, on average, 45% higher than the results of the automated chamber approach, the correlation between the two techniques was good (R-2 = 0.77, n = 2945). The annual soil CO2 efflux estimated by the automated chamber (665 g C m(-2)) and open-top chamber (720 g C m(-2)) matched the total annual ecosystem respiration, as estimated by the above-canopy eddy covariance method. However, the LI-6400 chamber and soil CO2 gradient approaches gave higher values for annual Soil CO2 efflux than the eddy covariance method. (C) 2003 Elsevier B.V. All rights reserved.