Key points are not available for this paper at this time.
Projections of terrestrial climate feedbacks from Earth System Models (ESMs) depend on simulated carbon and water dynamics of vegetation, and their credibility depends on realistic representation of water stress responses. Here, we discuss how models simulate plant hydraulic processes and identify opportunities to improve confidence in model outputs. We focus on the vegetation sub-models that represent plant physiology and hydrodynamics, and also on groundwater flow models, which could influence water availability. Recently implemented plant hydrodynamic frameworks allow ESMs to represent water transport through plants but depend on crudely estimated fixed trait values. Few data are available for parameterizing, assessing, and validating these models across landscapes and at large scales. Furthermore, global models do not account for fine-scale soil water and groundwater flow patterns that influence growth and mortality. We encourage the development of model-compatible databases that include vetted data on plant hydraulic traits, states, and responses to drought. We also encourage efficient model representation of sub-grid variation in slopes, soil depths, and groundwater accessibility. Finally, we recommend using physically informed artificial intelligence approaches to integrate knowledge from process-based models and data from field observations, experiments, and remote-sensing products toward better prediction of ecosystem dynamics under water stress at large scales.
Dukes et al. (Tue,) studied this question.