Prediction of Unsaturated Hydraulic Conductivity of Agricultural Soils Using Artificial Neural Network and c#

Abstract

Aims: The objective of this study was to develop an artificial neural network model and interactive application using C# application to predict unsaturated hydraulic conductivity of soil. Study Design: The actual measurements of unsaturated hydraulic conductivity of soil were obtained using the Mini Disk Infiltrometer (Decagon Devices, Inc.). Place and Duration of Study: The study was conducted in laboratory located in Community College, Huraimla, Shaqra University, Saudi Arabia during March-April 2015. Methodology: The experiments were conducted using water having electric conductivity of 2.26 dS/m and sodium adsorption ratio of 4.8. Unsaturated hydraulic conductivity of different soil textures (sand, sandy loam, loam and loamy sand) was determined at suction of -6 cm using Mini Disk Infiltrometer. The soil samples were taken from depth of 0-20 cm and repacked in a plastic 1000 cm3 container. Results: The predicted unsaturated hydraulic conductivity of soils compared favorably with the actual measurements in testing stage, however, mean relative error was 4.184% and coefficient of determination (R2) was 0.9979. In general, artificial neural network model gave considerable results but more data is still necessary. The main equations for C# application were obtained from the trained artificial neural network model.  Conclusion: It could be concluded that the developed interactive application is recommended for estimating unsaturated hydraulic conductivity of agricultural soils within the range of the studied variables to provide data for water management in Saudi Arabia.