Using Daytime Eddy Covariance Measurements to Infer Nighttime Respiration and the Annual Carbon Balance of Three Boreal Forests

Continuous measurements of annual carbon exchange in 2000 were made at three southern boreal forests using the eddy covariance (EC) technique as part of the Boreal Ecosystem and Monitoring Sites (BERMS) program. The BERMS program was initiated to ensure the long-term monitoring of climate and carbon exchange, and evolved from the Boreal Ecosystem Atmosphere Study (BOREAS). Here we compare estimates of annual ecosystem respiration (R) inferred from two respiration-air temperature algorithms - one based on nighttime high wind speed EC data and the other on daytime light-response characteristics. This approach is used to address the uncertainty in the annual carbon balance estimate due to problems associated with nighttime EC measurements. The annual carbon balance estimate for each forest ranged between 165 and 80 g C m^-2 at southern old aspen (SOA), 12 and -16 g C m^-2 at southern old black spruce (SOBS), and 47 and 17 g C m^-2 at southern old jack pine (SOJP). These annual estimates indicate that old southern boreal forests range from a moderate sink (SOA) to a possible small source (SOBS). Annual R ranged between 1104 and 1367 g C m^-2 at SOA, 1050 and 1062 g C m^-2 at SOBS, and 738 and 751 g C m^-2 at SOJP. This analysis indicates that R estimated from the daytime light-response function compared to the nighttime EC fluxes is significantly larger for SOA but is similar for SOBS and SOJP. Examination of the mean vertical wind speed, following Lee (1998), for each site suggests that R may be underestimated at SOA due to vertical advection and/or the development of thermal circulation such as cold air drainage. For example, Lee (1998) demonstrated that R at SOA could be underestimated by as much as 260 g C m^-2 y^-1, which compares favourably with our estimated difference of 263 g C m^-2 y^-1. Furthermore, an automated forest floor and tree bole chamber system operating at SOA confirms that carbon efflux is underestimated by nighttime EC measurements. It is also possible, however, that the observed differences in R are due to errors associated with the respiration parameter estimates. The difference in the carbon balance between these forests is complex. Despite having the shortest growing season carbon sequestration is greatest at SOA due to a relatively large photosynthetic capacity. At the conifer sites, photosynthetic capacity is marginally larger at SOBS; however, annual carbon sequestration is smaller due to large R. From the daytime light-response analysis we demonstrate that the ecosystem light compensation point (λ ) of these forests increases seasonally with air temperature. There is also some evidence to suggest that λ increases with forest age, suggesting that the potential for carbon sequestration may decrease for old growth forests.