Development of Irrigation Water Quality Index Using Artificial Neural Network

Abstract

The data-driven Artificial Intelligence (AI) techniques revealed specific relevance for the treatment of nonlinear relations and predicting the behaviour of complex systems, as a promising application in hydrology and water quality problems. The goal of this study is to build a developed model to forecast the quality of irrigation water by estimating its Water Quality Index using Artificial Neural Network (ANN). The developed model is applied to predict a data-based Irrigation Water Quality Index (IWQI) for groundwater usability in a desert reach pilot area in Egypt. The raw data for the model were the results of the main ion-causing irrigation hazards: (Salinity & Infiltration rate& Specific Toxics and Miscellaneous effects) for seventy-seven groundwater samples. The effectiveness of the model was achieved through the standardized coefficient of input variables. Revealing that the developed ANN model has a high agreement between measured and calculated IWQI (R2= 0.963, RMSE=0.0693) and becomes satisfactory verified for predicting the overall quality of groundwater in the research region, which is based on individual measurements rated according to their sensitivity. Moreover, the newly developed model can overcome the problem of missing some sample index parameters when one or more of the parameters are missing.

Keywords

• irrigation water quality index
• prediction
• artificial neural network
• regressions
• groundwater.