Subgrid-scale variability in the surface energy balance of arctic tundra

Surface fluxes of energy, water vapor, and CO, over homogeneous areas of the major tundra vegetation types in arctic Alaska were measured using a mobile eddy covariance tower for 5-day periods in the middle of the 1994 growing season. In order to account for differences in weather and time of season, data were analyzed in comparison to a nearby, fixed tower that operated throughout the summer. Among the different vegetation types, evaporation ranged from 1:3 to 2.7 mm d(-1). Net carbon uptake ranged from 0.5 to 2.4 g C m(-2) d(-1). Ground heat flux consumed 10-33% of midday net radiation. Typically, 38% of the net radiation was partitioned into latent heat flux, while the fraction of net radiation removed from the surface in sensible heat flux varied from 16 to 50% among vegetation types. The largest differences among vegetation types in surface energy partitioning were related to variations in soil moisture, with midday Bowen ratios ranging from 0.37 over wet sedge tundra to 2.25 over dry heath. Direct effects of vegetation on the driving gradients for energy and water vapor exchange were important in shrub tundra: shading of the moss layer by the canopy reduced ground heat flux and increased sensible heat flux, while latent heat flux was similar to other mesic vegetation types because the moss layer accounted for a larger portion of total evaporation than did evapotranspiration by shrubs. Scaling up from the vegetation types to the Alaskan arctic using an area-weighted average of the observed energy partitioning gave results similar to regional energy budgets measured over larger, more heterogeneous areas of tundra. An extrapolation based on the hypothesis that climate variability could cause a large fraction of the current tussock tundra vegetation to be converted to shrub tundra resulted in modest changes in the regional energy balance. However, nonlinear variations of surface evaporation with leaf area and uncertainties regarding changes in moss cover suggest that additional field experiments as well as modeling efforts will be required to predict the potential for changes in arctic tundra vegetation to feed back on regional climate.