We present the annual patterns of net ecosystem‐atmosphere exchange (NEE) of CO2 and H2O observed from a 447 m tall tower sited within a mixed forest in northern Wisconsin, USA. The methodology for determining NEE from eddy‐covariance flux measurements at 30, 122 and 396 m above the ground, and from CO2 mixing ratio measurements at 11, 30, 76, 122, 244 and 396 m is described. The annual cycle of CO2 mixing ratio in the atmospheric boundary layer (ABL) is also discussed, and the influences of local NEE and large‐scale advection are estimated. During 1997 gross ecosystem productivity (947−18 g C m−2 yr−1), approximately balanced total ecosystem respiration (963±19 g C m−2 yr−1), and NEE of CO2 was close to zero (16±19 g C m−2 yr−1 emitted into the atmosphere). The error bars represent the standard error of the cumulative daily NEE values. Systematic errors are also assessed. The identified systematic uncertainties in NEE of CO2 are less than 60 g C m−2 yr−1. The seasonal pattern of NEE of CO2 was highly correlated with leaf‐out and leaf‐fall, and soil thaw and freeze, and was similar to purely deciduous forest sites. The mean daily NEE of CO2 during the growing season (June through August) was −1.3 g C m−2 day−1, smaller than has been reported for other deciduous forest sites. NEE of water vapor largely followed the seasonal pattern of NEE of CO2, with a lag in the spring when water vapor fluxes increased before CO2 uptake. In general, the Bowen ratios were high during the dormant seasons and low during the growing season. Evapotranspiration normalized by potential evapotranspiration showed the opposite pattern. The seasonal course of the CO2 mixing ratio in the ABL at the tower led the seasonal pattern of NEE of CO2 in time: in spring, CO2 mixing ratios began to decrease prior to the onset of daily net uptake of CO2 by the forest, and in fall mixing ratios began to increase before the forest became a net source for CO2 to the atmosphere. Transport as well as local NEE of CO2 are shown to be important components of the ABL CO2 budget at all times of the year.

Time series of mixed layer depth, z i , and stable boundary layer height from March through October of 1998 are derived from a 915-MHz boundary layer profiling radar and CO 2 mixing ratio measured from a 447-m tower in northern Wisconsin. Mixed layer depths from the profiler are in good agreement with radiosonde measurements. Maximum z i occurs in May, coincident with the maximum daytime surface sensible heat flux. Incoming radiation is higher in June and July, but a greater proportion is converted to latent heat by photosynthesizing vegetation. An empirical relationship between z i and the square root of the cumulative surface virtual potential temperature flux is obtained (r 2 ϭ 0.98) allowing estimates of z i from measurements of virtual potential temperature flux under certain conditions. In fair-weather conditions the residual mixed layer top was observed by the profiler on several nights each month. The synoptic mean vertical velocity (subsidence rate) is estimated from the temporal evolution of the residual mixed layer height during the night. The influence of subsidence on the evolution of the mixed, stable, and residual layers is discussed. The CO 2 jump across the inversion at night is also estimated from the tower measurements.

No abstract

Methodology for determining fluxes of CO 2 and H 2 O vapor with the eddy-covariance method using data from instruments on a 447-m tower in the forest of northern Wisconsin is addressed. The primary goal of this study is the validation of the methods used to determine the net ecosystem exchange of CO 2. Two-day least squares fits coupled with 30-day running averages limit calibration error of infrared gas analyzers for CO 2 and H 2 O signals to ഠ2%-3%. Sonic anemometers are aligned with local streamlines by fitting a sine function to tilt and wind direction averages, and fitting a third-order polynomial to the residual. Lag times are determined by selecting the peak in lagged covariance with an error of ഠ1.5%-2% for CO 2 and ഠ1% for H 2 O vapor. Theory and a spectral fit method allow determination of the underestimation in CO 2 flux (Ͻ5% daytime, Ͻ12% nighttime) and H 2 O vapor flux (Ͻ21%), which is due to spectral degradation induced by long air-sampling tubes. Scale analysis finds 0.5-h flux averaging periods are sufficient to measure all flux scales at 30-m height, but 1 h is necessary at higher levels, and random errors in the flux measurements due to limited sampling of atmospheric turbulence are fairly large (ഠ15%-20% for CO 2 and ഠ20%-40% for H 2 O vapor at lower levels for a 1-h period).

Abstract.We