Inferring the contribution of advection to total ecosystem scalar fluxes over a tall forest in complex terrain
Multiple data streams from a new flux tower located in complex and heterogeneous terrain at theCoweeta Hydrologic Laboratory (North Carolina, USA) were integrated to identify periods of advectiveflow regimes. Drainage flows were expected a priori, due to the location of the measurement site at thebase of a long, gently-sloping valley. Drainage flow was confirmed by examining vertical profile mea-surements of wind direction and by estimating vertical advection fluxes. The vertical advection flux ofCO2was most significant in early morning (000–0600 h) during the growing season, when it averaged∼5 _mol m−2s−1. Horizontal advection flux of CO2was not directly measured in this study; however, anexpected exponential relationship between nocturnal ecosystem respiration (RE) and air temperaturewas recovered when horizontal advection of CO2was assumed to be negatively correlated to verticaladvection, or when data were limited to periods when measured vertical advection fluxes were small.Taken together, these data imply the presence of a negative horizontal advection CO2flux during noc-turnal periods characterized by positive vertical advection of CO2. Daytime periods were characterizedby consistent anabatic (up-valley) flows in mid- to late-morning (0500–1200 h) and consistent katabatic(down-valley) flows in the afternoon. A combination of above-canopy flux profile measurements, energybalance closure estimates, and flux footprint estimates suggest that during periods of up-valley windflow, the flux footprint frequently exceeds the ecosystem dimensions, and horizontal advection fluxesrelated to landscape heterogeneity were a significant component of the total ecosystem flux of CO2. Weused sap flux from individual trees beneath the tower to explore diurnal patterns in stomatal conductancein order to evaluate gapfilling approaches for the unreliable morning data. The relationship between sto-matal conductance and vapor pressure deficit was similar in morning and afternoon periods, and weconclude that gapfilling morning data with models driven by afternoon data is a reasonable approachat this site. In general, results were consistent with other studies showing that the advection and windflow regimes in complex terrain are highly site specific; nonetheless, the site characterization strategydeveloped here, when used together with independent estimates of components of the ecosystem carbonflux, could be generally applied in other sites to better understand the contribution of advection to thetotal ecosystem flux.