Abstract

The water crisis is a concern for Chilean agriculture. Testing new methods based on computer simulations is urgent to optimize irrigation. This study aimed to assess the model efficiency of HYDRUS 2D/3D simulations of volumetric water content (theta) and pore electrical conductivity (ECp) in an olive tree variety Kalamata under desert conditions. The model efficiency was assessed by comparing model simulations against observations of B and ECp in five frequency domain reflectometry (FDR) sensors installed in the soil profile. Model simulations were improved by calibration using PEST software. Global sensitivity analysis was performed before calibration, analyzing both theta and ECp model outputs. Outcomes of sensitivity analysis indicate that the surface area associated with transpiration (S-T) and the slope of the stress response function (s) are relevant parameters for theta(s) and ECp. Both parameters were calibrated along with the saturated water content (theta(s)) and pore size distribution (n) parameters of the second material. Calibration improved HYDRUS 2D/3D simulations for theta but not substantially for ECp. However, the Nash-Sutcliffe and the root mean square error (RMSE) are comparable with previous research for both variables. Furthermore, the s parameter decreases after calibration, indicating that Kalamata variety is tolerant to salt, which is in line with previous research. HYDRUS 2D/3D represents the theta variation in time and space with acceptable precision for olive trees under desert conditions. Additionally, subsequent studies should focus on the value of s, which is variety dependent, and S-T, which requires the actual root volume.