Development of an Artificial Neural Network model for predicting the infiltration rate of soils in the Chhattisgarh plains

Abstract

The infiltration rate is an important element in research concerning soil, hydrology, ecology, and agriculture. It serves as the fundamental input variable for models that examine water movement. Its significance lies in being the main input for water flow modeling. This study assessed the stable infiltration rate across various soil types in the Chhattisgarh Plains by using a multilayer artificial neural network, known as “ANN,” implemented through the back propagation algorithm. The data collected included bulk density, moisture content, and the proportions of sand, silt, and clay. An unseen dataset was utilized to evaluate the performance of the ANN models after the training phase. Philip created a model incorporating sand, silt, bulk density, and infiltration rate to ascertain the infiltration rate for assessment purposes. The findings from this research demonstrate a notable correspondence between the gathered data and the ANN simulations. The figures for the root mean square error and the coefficient of determination were marginally lower in the ANN steady infiltration rate model than in Philip’s model used for determining the infiltration rate. While the outcomes of these evaluations support the practical use of ANN, having a comprehensive local soil database from various locations would strengthen the assessment of the ANN models.