Long-term carbon exchange in an old-growth temperate rain forest

Results from a six-year comparison of observed and simulated net ecosystem exchange of carbon dioxide (NEE) estimates are presented. Eddy-covariance estimates of NEE every half hour from January 1, 1999 through December 31, 2004 were gathered within an old-growth temperate rainforest in the Cascade Mountains in western North America. The Wind River Canopy Crane Research Facility (WRCCRF) is unique among AmeriFlux sites due to the size, age, and complexity of the canopy and provides a rare opportunity to investigate carbon cycling in such ecosystems.

The University of California, Davis Advanced Canopy-Atmosphere-Soil Algorithm (ACASA) was run for half-hour intervals using meteorological data measured at WRCCRF in a continuous fashion for the entire period. Results show excellent qualitative agreement between model and observed cumulative NEE, with annual totals falling well within observational uncertainty. Observed and simulated monthly cumulative NEE values were compared using linear regression, with an r2 of 0.89 for the period. Both model and observations reflect high seasonal and year-to-year variability in NEE, driven mainly by moisture and temperature variations, and phenology. Observed and modeled estimates show wide interannual variability, with 1999 being a strong sink year (Obs: -170 ±35 gC m-2) and 2003 a net source (Obs:+49 ± 29 gC m-2). Results indicate that NEE is sensitive to both temperature and moisture states of the canopy and soil. NEE for 2001 was near-equilibrium despite abundant soil moisture, due to frequent,abnormally high temperatures. ACASA results indicate that if the forcing air temperature had been 1.5 degrees C less than measured, all else being equal, 2001 would have been a moderate sink year. Source year 2003 involved both abnormal heat and drought.

Analyses of measured and modeled 30-minute mean vertical profiles of carbon dioxide concentrations for June 19-26, 2002 indicate model-observed agreement within 95% statistical confidence for most of the time period. An exception is a recurring pattern for a few hours after sunset, with observed profiles showing marked distortions which may be related to heterogeneous, isentropic drainage flows amid thermally stable and unstable regimes within the canopy. Previous work involving the diurnal evolution of the intra-canopy windfield, including directional wind shear, supports this hypothesis. ACASA, a single-point model, by definition cannot realize such phenomena, unless nested within a larger modeling framework which includes horizontal advection, such as fine-mesh mesoscale or large-eddy simulators.

Results indicate that on 30-minute to interannual time scales, ACASA is able to simulate net ecosystem carbon exchange within statistical and observational uncertainty. Such a tool is useful for assessing the nature of the terrestrial carbon cycle under current conditions as well as past, present, and future meteorological scenarios.