Sensitivity analysis and calibration of an upscaled microscopic root water uptake model by inverse modeling

The assessment of limitations to water transfer through the soil-plant-atmosphere continuum allows a better prediction of evapotranspiration fluxes. Microscopic models of root water uptake (RWU) allow the process-based simulation of the involved processes. The objective was to perform a sensitivity analysis and calibration by inverse modeling (IM) of the SWAP model with a process-based microscopic RWU module (MFlux model) for a soybean crop in southeast Brazil. The sensitivity analysis was performed for five Brazilian soils and 32 years of weather data recorded in the municipality of Piracicaba, São Paulo, Brazil (22°43’ S, 47°38’ W, 524 m a.s.l.). The SWAP model was calibrated using experimental data (actual evapotranspiration and soil water content at two depths) from two seasons (2016/17 and 2017/18) of irrigated soybean in Piracicaba. Results show that the sensitivity of drought stress is high for low values of root length density and radial hydraulic conductivity of root tissue, and low for other parameters of the MFlux model in all considered value ranges. Seasonal and interannual meteorological conditions are more determinant to reductions in biomass production due to drought stress than root hydraulic properties and root geometry. A two-step calibration procedure of SWAP that combines manual calibration of the crop growth module and automatic calibration of the RWU and root distribution modules by IM leads to acceptable model performance in the simulation of evapotranspiration fluxes and soil water dynamics for soybean in southeast Brazil.