Artificial neural network optimization to predict saturated hydraulic conductivity in arid and semi-arid regions

•Soil saturated hydraulic conductivity (Ksat) is spatially variable and heterogeneous.•Artificial neural work (ANNs) are feasible option relating soil physical properties to Ksat.•Generalized Regression Neural Networks (GRNN) models were built with different soil features.•Soil electrical conductivity (EC) was the most important feature after soil texture features.•Sand and clay contents with EC explain 86% of Ksat variability.