Spatial and temporal variability in forest-atmosphere CO2 exchange

Seven years of carbon dioxide flux measurements indicate that a similar to90-year-old spruce dominated forest in Maine, USA, has been sequestering 174+/-46 g C m(-2) yr(-1) (mean+/-1 standard deviation, nocturnal friction velocity (u(*)) threshold >0.25 m s(-1)). An analysis of monthly flux anomalies showed that above-average spring and fall temperatures were significantly correlated with greater monthly C uptake while above-average summer temperatures were correlated with decreased net C uptake. Summer months with significantly drier or wetter soils than normal were also characterized by lower rates of C uptake. Years with above-average C storage were thus typically characterized by warmer than average spring and fall temperatures and adequate summer soil moisture. Environmental and forest-atmosphere flux data recorded from a second tower surrounded by similar forest, but sufficiently distant that flux source regions ('footprints'), did not overlap significantly showed almost identical temperature and solar radiation conditions, but some differences in energy partitioning could be seen. Half-hourly as well as integrated (annual) C exchange values recorded at the separate towers were very similar, with average annual net C uptake differing between the two towers by <6%. Interannual variability in net C exchange was found to be much greater than between tower variability. Simultaneous measurements from two towers were used to estimate flux data uncertainty from a single tower. Carbon-flux model parameters derived independently from each flux tower data set were not significantly different, demonstrating that flux towers can provide a robust method for establishing C exchange model parameters.