Using Artificial Neural Networks to estimate saturated hydraulic conductivity from easily available soil properties

Soil saturated hydraulic conductivity is the most important physical properties that has particular importance in identifying, investigating and modeling the water, salts and pollutants transport in the porous medium. Despite numerous research, measuring saturated hydraulic conductivity with direct methods are still costly, time consuming and professional. Therefore estimating saturated hydraulic conductivity with rapid and low cost methods (pedo-transfer functions) with acceptable accuracy is essential. The purpose of this research was to estimate saturated hydraulic conductivity using easily accessible parameters such as particle size distribution, bulk density, total porosity, effective porosity, water content retained at -0.3 and -15 bar matric potentials, %CCE, %OM, pH and EC with Artificial Neural Networks. Saturated hydraulic conductivity was measured from 73 selected points at three depths (10-35, 15-35 and 20-35) with Guelph permeameter and soil samples were taken from same points. Easily accessible parameters were measured in laboratory and preliminary results were obtained. Selected parameters according to sensitivity analysis were sand and clay contents, water content at -0.3 bar matric potential, total porosity and geometric mean diameter of soil particles. Using sensitive parameters, a rapid and low cost method was selected from different designed models. Input parameters were logaritmic geometric mean diameter, total porosity, sand and clay contents with this model.