Effects of elevated CO2 on forest hydrological budget

Background/Question/Methods Elevated atmospheric CO₂ (eCO₂) is expected to alter the hydrological budget of forests. Some species, especially broadleaved deciduous, will reduce mean stomatal conductance (g), slowing the rate of soil drying between rain events, reducing the length or severity of drying cycles. This may lead to increased leaf area index (L) reestablishing canopy conductance (G = g * L), and transpiration (T). Direct eCO₂-induced stomatal response of other species, especially evergreen conifers, will be muted, but L can increase with carbohydrate availability if water is not limiting. Regardless of the mechanism, rainfall interception (I) will increase with L, but lower radiation at the forest floor and thicker litter could reduce below-canopy evaporation (Esc). The net effect on total evapotranspiration (ET=T+I+Esc) is therefore difficult to predict. Among models, eCO₂ effect on ET will vary depending on how the processes are captured in each. We compared the results from three models to T (from scaled-sapflux measurements) and ET (I measured; Esc modeled) from the Duke Forest Free Air CO₂ Enrichment (FACE) experiment at a loblolly pine-dominated forest over 12 years. We assessed each model as to whether it generated reasonable estimates of the quantities of T and ET and, regardless, whether it reproduced correctly the relative response to eCO₂. We also searched for the reasons models failed in either task.

Results/Conclusions Predicted daily quantities of ET and T were poorly correlated with measured quantities. Models varied in their ability to estimate annual ET and T, and the effect of eCO₂.  ED and LPJ-GUESS (with CENTURY-based N limitation) correctly reproduce essentially no eCO₂ effect on both fluxes. However, LPJ overestimated ET by ~75% and T by ~50% respectively. ED reversed the degree of overestimation, generating 17% higher ET but 33% higher T. OCN generated fluxes closest to the measured, +10% under ambient and -5% under eCO₂, thus incorrectly estimating lower fluxes under eCO₂. Large departure of modeled from actual fluxes will affect the ability of models to accurately reproduce the energy budget and, to the degree that T and photosynthesis are coupled, also CO₂ uptake and net primary production. Similarly, reproducing higher or lower fluxes under elevated CO₂ in this stand would introduce like errors in the energy and C budgets. We are currently obtaining input from additional models and assessing the sources of deviation of model-based estimates from measured fluxes and, where applicable, the reasons a model incorrectly show effects of eCO₂.