Bridging the gap between empirical and mechanistic approaches in modeling plant transpiration under drought conditions

In hydrological modeling, both empirical and mechanistic approaches have their own advantages, the former favoring simple relations between observed state variables, and the latter being rather process-based and allowing a deeper understanding of the system. In cases both approaches match, a chance is given to interpret the simple empirical parameters in terms of bio-physical system properties. We recently developed a simple mechanistic plant water stress model, respecting the physics of water flow inside the plant architecture, from the soil-root interfaces to the leaves. The shape of the three-dimensional mechanistic model matches empirical observations of transpiration rate reduction under water stress. Yet, two points are questioned: (i) Do physical parameters still make sense in increasingly simplifying conditions, like 1-D soil domains? (ii) How would additional biological processes, like cavitation or dynamic aquaporin activity, affect the macroscopic behavior of the system? Scenarios of water dynamics in the soil-plant system were simulated with the detailed modeling platform R-SWMS. The parameters of the simple mechanistic model were characterized and used to predict water stress on 3-, 2- and 1-D soil domains. In these reduced spatial dimensions, the applicability of the simple mechanistic model was discussed and further evaluated on different soil types and under different climatic demands. Eventually, the shape of the water stress function was compared with empirical observations, and the impact of the system bio-physical properties on water stress was emphasized. These properties available in R-SWMS include dynamic adaptation of radial root hydraulic conductivity, xylem resistance to cavitation, and loss of contact between soil and roots due to shrinkage