Towards modeling hydrodynamic stress limitations on transpiration

Evapotranspiration is one of the major forcing functions of Earth's climate, providing the link for the soil-plant-water continuum. Current models for transpiration assume a coupling between stomatal conductance and soil moisture through empirical relationships that do not resolve the hydrodynamic process of water movement from the soil to the leaves. This approach does not take advantage of recent advances in our understanding of water flow and storage in the trees, or of tree and canopy structure. It has been suggested that stomata respond to water potential in the leaf and branch, and that this hydrodynamic response is a mechanism for hydraulic limitation of stomatal conductance. Hydraulic limitations in forest ecosystems are common and are known to control transpiration when the soil is drying or when vapor pressure deficit (VPD) is very large. Hydraulic limitation can also impact stomatal apertures under conditions of adequate soil moisture and lower evaporative demand. Hydrodynamic stresses at the tree level act at several time scales, including the fast, minute-hour scale. These dynamics are faster than the time scales of hours to days at which drying soil will affect stomata conductance. The lack of representation of the tree-hydrodynamic process should therefore lead to atypical intra-daily patterns of error in results of current models. We use a large-scale comparison between observations and land-surface models to characterize the patterns of intra-daily error in simulated water flux. Through the use of the North American Carbon Program (NACP) dataset, more than 10 years of water flux data for 35 Fluxnet sites in the US and Canada have been analyzed. The diurnal error for each of the 24 models represented in this dataset allows the models to be categorized and evaluated on their ability to accurately predict the fast temporal dynamics of transpiration in different ecosystems and atmospheric forcing. Among well calibrated models, two general error patterns prevail: (1) a daytime underestimation followed by nighttime overestimation of latent heat exchange; and (2) an underestimation during the morning and an afternoon overestimation. Although model error patterns are site specific, there are models that are more likely to favor one pattern above the other based on the model's sensitivity to VPD and soil moisture. The second error pattern occurs more frequently in sites where non-limiting soil moisture conditions exist. We hypothesize that the afternoon overestimation of transpiration in these scenarios can be explained by the models' lack of a mechanism to handle midday stomata closure due to hydrodynamic stresses. We introduce FETCH - a tree hydrodynamic model that can resolve the fast dynamics of stomatal conductance. We propose that coupling FETCH to other land-surface models would reduce intra-daily errors and improve the representation of canopy structure in atmospheric and hydrologic simulations.