Implementing a process-based representation of soil water movement in a second-generation dynamic vegetation model: application to dryland ecosystems (LPJ-GUESS-RE v1.0)

Abstract. Dryland ecosystems are globally important, yet state-of-the-art dynamic vegetation models often lack specific processes or parameterizations that are critical for accurately simulating dryland dynamics. These missing processes include a realistic calculation of soil water movement, detailed plant-water relations, or a representation of deep water uptake. In this study we show how including a process-based soil hydrology scheme in the LPJ-GUESS (Lund-Potsdam-Jena General Ecosystem Simulator) model can improve its usefulness for simulating the functioning of dryland ecosystems. By replacing the default 15-layer bucket representation of soil hydrology in LPJ-GUESS v4.1 with a mechanistic description of soil water movement based on the 1D Richards Equation, we show that the model is better able to capture seasonal patterns of water cycling through dryland ecosystems at both the site level and the regional level. In addition, the inclusion of a new set of bottom boundary conditions, such as a permanent groundwater layer, further expands the range of ecosystems the LPJ-GUESS model can simulate. We show that soil bottom boundary conditions, in particular varying levels of groundwater depth, can have a large influence on vegetation composition and water cycling. Our new model developments open new avenues to simulate dryland ecohydrology more realistically.