Sensitivity analysis of a process-based root water uptake model to predict drought stress in soybean and wheat in a tropical winter-dry climate

Drought stress is one of the main reasons for reduced yields in soybean and wheat crops in Brazil. Process-based root water uptake (RWU) models are valuable tools to assess soil-water-plant relations and improve crop water management. We aimed to perform a pioneer sensitivity analysis (SA) of a process-based RWU model using three methods and two sampling strategies. The SWAP agro-hydrological model with the recently implemented MFlux RWU function was used to predict drought stress in soybean and wheat crops simulated on five soils with different hydraulic properties sampled in southeast Brazil, characterized by a tropical winter-dry climate. Three SA methods were used: local, global Morris, and global Sobol. Seven parameters of the MFlux function were selected, together with their reference values and ranges of variability. The local sensitivities were predominately negative, indicating that the drought stress increased as the values for each RWU parameter decreased. The Morris method revealed parameter interactions not addressed in the local method. The Sobol method also evidenced parameter interactions calculated through robust variance-based indices. Although the three SA methods provided different results regarding parameter contributions to drought stress prediction, the root length density was the most sensitive parameter for all simulated scenarios. Hence, it should be a priority in future model calibration efforts.