Net uptake of carbon dioxide (CO 2 ) measured by eddy covariance in a 60- to 80-year-old forest averaged 2.0 ± 0.4 megagrams of carbon per hectare per year during 1993 to 2000, with interannual variations exceeding 50%. Biometry indicated storage of 1.6 ± 0.4 megagrams of carbon per hectare per year over 8 years, 60% in live biomass and the balance in coarse woody debris and soils, confirming eddy-covariance results. Weather and seasonal climate (e.g., variations in growing-season length or cloudiness) regulated seasonal and interannual fluctuations of carbon uptake. Legacies of prior disturbance and management, especially stand age and composition, controlled carbon uptake on the decadal time scale, implying that eastern forests could be managed for sequestration of carbon.

We present 5 years of NOy and O3 eddy flux and concentration measurements and NOx concentration measurements at Harvard Forest (1990–1994), a mixed deciduous forest in central Massachusetts, and 2 months of data for a spruce woodland near Schefferville, Quebec, during the NASA ABLE3B/Northern Wetlands Study (1990). Mean midday values of net dry NOy flux from atmosphere to canopy were 3.4 and 3.2 μmole m−2 hr−1 at Harvard Forest in summer and winter, respectively, and 0.5 μmole m−2hr−1 at Schefferville during summer. Nighttime values were 1.3, 2.0, and 0.15 μmole m−2 hr−1, respectively. For 1990–1994, the net annual dry deposition of nitrogen oxides was 17.9 mmole m−2 yr−1 (2.49 kgN ha−1 y−1). Oxidized species such as HNO3 dominated N deposition, with minor contributions from direct deposition of NO2. Emissions of NO from the forest soil were negligible compared to deposition. Comparison of NOy deposition at Harvard Forest and Schefferville and analysis of the dependence on meteorological parameters show that anthropogenic sources dominate the nitrogen oxide inputs over much of North America. Heterogeneous reactions account for >90% of the conversion of NO2 to HNO3 in winter, leading to rates for dry deposition of NOy similar to fluxes in summer despite 10‐fold decrease in OH concentrations. In summer, formation of HNO3 by heterogeneous reactions (mainly at night) could provide 25–45% of the NO2 oxidation.

An analysis of boundary layer cumulus clouds and their impact on land surface-atmosphere exchange is presented. Seasonal trends indicate that in response to increasing insolation and sensible heat flux, both the mixed-layer height (z i) and the lifting condensation level (LCL) peak (ϳ1250 and 1700 m) just before the growing season commences. With the commencement of transpiration, the Bowen ratio falls abruptly in response to the infusion of additional moisture into the boundary layer, and z i and LCL decrease. By late spring, boundary layer cumulus cloud frequency increases sharply, as the mixed layer approaches a new equilibrium, with z i and LCL remaining relatively constant (ϳ1100 and 1500 m) through the summer. Boundary layer cloud time fraction peaks during the growing season, reaching values greater than 40% over most of the eastern United States by June. At an Automated Surface Observing System (ASOS) station in central Massachusetts, a growing season peak is apparent during 1995-98 but reveals large variations in monthly frequency due to periods of drought or excessive wetness. Light-cloud cover regression relationships developed from ASOS ceilometer reports for Orange, Massachusetts, and Harvard Forest insolation data show a good linear fit (r 2 ϭ 0.83) for overall cloud cover versus insolation, and a reasonable quadratic fit (r 2 ϭ 0.48) for cloud cover versus the standard deviation of insolation, which is an indicator of sky type. Diffuse fraction (the ratio of diffuse to global insolation) shows a very good correlation (r 2 ϭ 0.79) with cloud cover. The sky type-insolation relationships are then used to analyze the impact that boundary layer clouds have on the forest ecosystem, specifically net carbon uptake (), evapotranspiration (ET), and water use efficiency (WUE). During 1995, afternoon was 52% greater F F CO CO 2 2 on days with boundary layer cumulus clouds than on clear days, although ET was the same, indicating greater light use efficiency and WUE on partly cloudy days. For 1996-98, afternoon was also enhanced, especially F CO2 during dry periods. Further analysis indicates that the vapor pressure deficit (VPD) was significantly greater (Ͼ8 hPa) during 1995 and parts of 1996-98 on clear days as compared with partly cloudy days. A long-term drought combined with abnormally warm weather likely contributed to the high VPDs, reduced , ET, and F CO2 the dearth of clouds observed during 1995. In general, the presence of boundary layer cumulus clouds enhances net carbon uptake, as compared with clear days.

Eddy covariance flux observations at a deciduous temperate forest site (83 days) and at a boreal forest site (21 days) are analyzed for midday periods (1100-1400 LT). Approximate stationarity of the time series is demonstrated, and the ensemble-averaged roughness sublayer cospectra are presented. Spectral and cospectral forms in the roughness sublayer are more peaked than those found in an inertial sublayer. They exhibit similar forms dependent on (z Ϫ d)/(h Ϫ d), where d is the displacement height and h is the canopy height. The inertiallayer spectral forms are recovered when observations are made where this scaled height is approximately 4. For a sample summer at the midlatitude deciduous forest, large eddies with periods from 4 to 30 min contribute about 17% to surface eddy fluxes of heat, water vapor, and carbon dioxide (CO 2). Much larger contributions can occur in light-wind conditions. This effect, likely caused by the passage of convective boundary layer eddies, is not observed when using many currently popular averaging procedures. Several running-mean periods have been used to assess the effect of the mean removal procedure on flux estimates. Given the assumption that large eddies would have been sampled at the towers had an ensemble measurement been possible, a correction is proposed based primarily on the mean wind speed to adjust fluxes obtained using short averaging intervals. This correction is successful in achieving observational energy-balance closure at two dissimilar forested sites. Cospectral similarity is found for all scalars studied. Daytime fluxes of CO 2 , for example, can be underestimated at standard flux towers by 10%-40%, depending on wind speed.