B081-0018
Understanding Spatial Variations in Maximum Stomatal Conductance

Monday, 14 December 2020
Poster
Olivia Flournoy, Stanford University, Stanford, CA, United States, Yanlan Liu, Duke University, Nicholas School of the Environment, Durham, NC, United States, Kimberly A Novick, Indiana University Bloomington, Bloomington, IN, United States, Quan Zhang Sr, Tsinghua University, Beijing, China and Alexandra G. Konings, Stanford University, Department of Earth System Science, Stanford, CA, United States
Abstract:
A large portion of evapotranspiration (ET) is contributed by transpiration, which is regulated by a range of plant traits. Among the most important plant traits is the maximum stomatal conductance (Gs,max), which controls the rate of transpiration under non-stressed conditions. In most land-surface models, Gs,max is prescribed as a constant for each plant functional type (PFT). However, previous research suggests that Gs,max can vary significantly within PFTs. Therefore, an improved representation of the spatial pattern of Gs,max may improve the accuracy of ET estimation in land surface models. This study aims to understand how Gs,max varies with local characteristics beyond PFT, including climate aridity and canopy height.

Using 114 sites in the FLUXNET2015 dataset, we first derive Gs,max at each site by inverting the Penman-Monteith equation in combination with a simple partitioning method for separating surface conductance into soil and canopy components. We then use regression analysis to evaluate how Gs,max varies with canopy height and climate dryness across a variety of PFTs and climate types. Initial results suggest that, when compared to the commonly used one-value-per-PFT approach, incorporating the canopy height and climate dryness improves both the accuracy (R2) and the Akaike information criterion (AIC) of Gs,max estimation. Our findings indicate that the spatial pattern of Gs,max can be better described beyond PFTs using readily available local properties of climate dryness and canopy height. Our approach may facilitate better parameterization of land-surface models for improved ET estimation.