Prediction of annual drinking water quality reduction based on Groundwater Resource Index using the artificial neural...

Drought is one of the most significant natural phenomena affecting different aspects of human life and the environment. Due to water scarcity, prediction of water quality reduction is very crucial for urban and rural communities. This study contributes by applying artificial neural network and modified fuzzy clustering techniques to estimate the drops in potential drinking water quality in the GIS environment. In this research, the probability of occurrence of adverse annual changes in the water quality of drinking water is estimated. The model was tested using real instances of the southeast aquifers, the regions of the central parts of the IRAN and especially the significant portions of the aquifers of the east area. To validate the model, the data adequacy test and the standardization of the drought index are used. The results of the lowest available water quality and the highest drought using ANNs show that the qualitative stress conditions in large part of the country's aquifers are in unfavorable conditions. Evidence from this research shows that the aquifers in these areas are expected to have severe drought stress and poor quality class status. Also, the computational results indicate that the modified clustering method increases the efficiency of the prediction model as against the previous research. The outcomes do not show a relatively favorable state of drinking water quality for some aquifers in the country. However, the conditions for quantitative changes in the depth of water, based on the predicted results of ANN, are considered critical. The generated maps demonstrate that about 64% of the study area is subjected to a severe deterioration in the quality of drinking water if the current trend continues in the exploitation of aquifers. As a result, the main finding the present study is that the probability of groundwater quality decline is significant in many aquifers in the country.