Soil water content estimation in Central Brazil using HYDRUS-1D and a dual-porosity retention model

Abstract

Soil controls processes within the hydrological cycle, which are fundamental to environmental sustainability and agricultural production, such as infiltration, redistribution, and storage of water. Numerical models capable of estimating soil water content at low operational costs are potential tools for the management of water resources. This study evaluated the HYDRUS-1D model for simulating soil water content in five tropical soil classes representative of the Brazilian Federal District, at two soil depths, using statistical indicators. Model parameterization was based on the determination of soil hydraulic and physical properties using field and laboratory methods, including soil water retention curves described by the van Genuchten–Durner dual-porosity model and saturated hydraulic conductivity. Model performance was assessed using RMSE, RSR, R², NSE, and PBIAS statistics. Overall, HYDRUS-1D showed high accuracy in predicting the temporal soil moisture dynamics across different soil classes and depths, indicating its suitability for applications in Cerrado soils. In particular, the use of the dual-porosity approach for describing the soil water retention curve contributed to improved prediction of soil hydraulic behavior. However, some statistical metrics indicated lower performance for the Rhodic Ferralsol in the subsurface layer, influenced by the sensitivity of the metrics to low data variability. The results highlight the importance of soil-specific determination of soil hydraulic parameters for the model performance. Despite the potential limitations, the study shows that the HYDRUS-1D model can be a valuable tool for soil water and land management in the Brazilian Cerrado and in the Federal District.

Keywords: dual-porosity, HYDRUS-1D, soil moisture, soil water retention curve, vadose zone.