An Artificial Neural Network Approach for the Determination of Infiltration Model Parameters

Abstract

Prediction of soil infiltration rate is of prime importance in irrigation and drainage studies. newlineInfiltration seems to be very simple, but determination of soil infiltration in field is very tedious newlineand time consuming job. The present study attempts to predict the soil infiltration rate and to newlineevaluate the soil infiltration model parameters using two infiltration models namely, Kostiakov newlineand modified Kostiakov into clay soil (Vertisols-FAO Classification) in Kopargaon region of newlineMaharashtra State. Subsequently, the Artificial Neural Network (ANN) was employed to newlineevaluate the constants of Kostiakov infiltration model. In this study the feedforward newlinebackpropagation type ANN was used. The data from the study area were generated through field newlinemeasurements of the infiltration of soils using double ring infiltrometer for two seasons namely newlinewinter and summer with existing land covers. The soil infiltration measurements were made at newline106 points over the study area of clay soil. newlineBefore conducting the field infiltration tests, the data regarding different soil properties like bulk newlinedensity, moisture content, % sand, % silt, % clay, electrical conductivity, field capacity and newlinewilting point were determined as these serves inputs for ANN models. Soil samples were taken newlinefrom the surface layer 150 -300mm thick by excavation and auger technique, from each study newlinepoint of clay soil. For determination of bulk density, undisturbed soil samples were collected, newlinewhereas for remaining soil properties disturbed soil samples were used. newlineThe infiltration model parameters were determined graphically and analytically using the Davis newlinemethod. The results of the investigation show that the cumulative infiltrations predicted by newlineKostiakov and modified Kostiakov models were very close to the field measured cumulative newlineinfiltration values of clay soil locally called as black cotton soil. The physical properties like newlinemoisture content, textural analysis and electrical conductivity affect soil infiltration rate as well newlineas the values of infiltration model par