Simulating plant water stress dynamics in a wide range of bi-specific agrosystems in a region using the BISWAT model

The ability to simulate soil and crop processes in many bi-specific systems (vineyards, orchards, silvo-arable agroforestry, strip-intercropping of arable crops...) is one of the major challenge for crop modelling in order to contribute to the design of agro-ecological cropping systems. A typical question is how soil, climate and management would influence the soil water deficit experienced by a plant grown alone or intercropped with a cover crop, with another crop or a tree, in order to improve the resilience of a cropping system to climate change and limit the use of chemical input. This study introduces BISWAT - Bispecific Intercrop System WATer Stress dynamic model - a new water balance model designed to simulate the dynamic of Soil Water deficit Experienced (SWEP) by two Plants when grown together or separated. BISWAT has been built to simulate a large range of agrosystems (annual and perennial crops, mono- or bi-specific) cultivated in various conditions. The model is primarily based on three modelling concepts: i) a 2D generic pattern for the system's spatial representation, ii) the use of the Radiation Interception Efficiency (RIE) to drive potential plant transpiration and soil evaporation, iii) the use of the Total Transpirable Soil Water (TTSW) concept coupled with a simple root dynamics representation. These concepts are not new but they allowed us to define a model able to simulate many crops and trees (including vineyards) using a limited number of inputs and without an explicit need for parameter calibration. The model was evaluated on five reference agrosystems (mono-specific salads, mono-specific vineyards, bispecific vineyards, mono-specific peach orchards and bi-specific peach orchards). The RMSE of the SWEP variable ranged from 0.049 to 0.123. A combined sensitivity and uncertainty analysis performed on typical farmer's fields situations stressed the particular importance of model inputs related to the TTSW of the soil-crop system. We conclude that the genericity of the BISWAT model, its method of parameterization and its performance open the perspective to use the model in a wide range of conditions where the dynamic of water stress between two species grown together is a key variable to be accessed with limited data for parameterization and a large number of fields to simulate.