Eddy Covariance in a Tallgrass Prairie: energy balance closure, water and carbon budgets, and shrub expansion

Abstract

The exchange of water, carbon, and energy between grasslands and the atmosphere is an important biogeochemical pathway affecting ecosystem productivity and sustainability. The eddy covariance (EC) technique directly measures this mass and energy exchange. However, questions remain regarding the accuracy of EC-derived H[subscript]2O and CO[subscript]2 fluxes in landscapes with irregular topography and variable vegetation. These concerns stem from the "energy balance (EB) closure problem" (i.e., measured energy in does not equal measured energy out). My main objectives were to examine EB closure at two topographical positions within an annually burned tallgrass prairie watershed and to examine the effect of landscape position and woody encroachment on carbon and water exchanges. In tallgrass prairie, 14 km south of Manhattan, KS, USA, EC towers were deployed at three sites in 2007 and 2008. One upland and lowland tower were within an annually burned watershed dominated by C[subscript]4 grasses. Another lowland tower was deployed in a separate quadrennial-burned watershed where significant woody vegetation occupied the tower's sampling area. All towers measured EB components (net radiation, R[subscript]n; soil heat flux, G; sensible heat flux, H; and latent heat flux, [lambda]E). In the annually burned watershed, landscape position had little effect on G, H, and R[subscript]n with differences [less than] 2% between sites. However lowland [lambda]E was 8% higher, owing to larger plant biomass/leaf area and greater soil moisture. Energy balance closure (i.e., [[lambda]E + H] / [R[subscript]n - G]) was 0.87 and 0.90 at the upland and lowland sites, respectively. A nearby large-aperture scintillometer provided good validation of EC-derived H in 2007. Data suggested that underestimates of [lambda]E may have accounted for the closure problem; sample calculations showed that increasing [lambda]E by 17% would have resulted in near prefect closure. Data from this study suggests that EB closure does not strongly correlate with topographical position; however these data raise questions regarding accuracy of the [lambda]E term. Mass exchange analysis shows that the prairie carbon cycle is highly dependent on burning. The lowland and upland annually burned sites saw carbon gains of 281 to 444 g C m[superscript]-[superscript]2 yr[superscript]-[superscript]1 before burning with the shrub lowland showing the least (e.g. 159 and 172 g C m[superscript]-[superscript]2 yr[superscript]-[superscript]1). After the prescribed burn, the upland and lowland sites remained slight carbon sinks (68 to 191 g C m[superscript]-[superscript]2 yr[superscript]-[superscript]1), whereas the unburned shrub site was a carbon sink in 2007 (159 g C m[superscript]-[superscript]2 yr[superscript]-[superscript]1, because no carbon loss was incurred via burning) and a large carbon source in 2008 when it was burned the following year (336 g C m[superscript]-[superscript]2 yr[superscript]-[superscript]1 loss). Evapotranspiration (ET) was highest at the shrub lowland where greater soil moisture and abundance of deep-rooted C[subscript]3 shrub vegetation allowed greater uptake and loss of water.