The downregulation of plant transpiration and photosynthesis due to water stress have long been represented in most land surface models (LSMs) using an empirical ratio of the ambient to maximum soil moisture, called a β factor. Recent studies have shown that the specification of the β factor as a function of soil moisture is a major source of uncertainty in modeling global carbon and water cycles. An alternative is to use plant hydraulic models (PHMs) to describe physiological responses to water stress. Here, we present results showing that although β factors can, in many cases, adequately approximate canopy-scale predictions of PHMs, certain combinations of plant traits, soil types, and environmental forcings cause β factor and PHM-based predictions to diverge. In the latter cases, LSM predictions based on β factors can incur substantial errors.
We compared β factor and PHM predictions within a calibrated LSM we developed based on the Community Land Model v5, which can be run in either a standard PHM or modified β factor “mode.” In β mode, the model uses effective β factors derived from the outputs of the PHM mode under dynamic environmental conditions. The model was applied across FLUXNET sites selected to span a representative combination of vegetation, soil, and climate types.
Results/Conclusions
Using a supply versus demand framework, we demonstrate that: i) plant water use lies on a spectrum between supply-limited and demand-limited regimes, ii) β factors are an end-member scenario where the soil-plant system is infinitely conductive and entirely regulated by demand-limitation, and iii) the errors incurred by using β instead of PHM depends on where the soil-plant system lies along the supply-demand spectrum and are especially exaggerated under high atmospheric demand and low soil water supply conditions. This research provides a framework for understanding the role of plant hydraulics in differentiating atmospheric versus soil water limitation on transpiration, quantifies the errors incurred by using the empirical β factors, and provides guidance for when to include the added complexity and computational costs of PHMs.