Prediction and assessment of water quality index using neural network model and gis case study: Tigris river in Baghdad city

In this research, nine sampling locations were selected for Tigris River to assess water quality by analyzing Water Quality Index (WQI). The analysis includes different water quality parameters: Turbidity (Tur.), Alkalinity (Alk.), pH, Total hardness (T.H), Magnesium (Mg), Calcium (Ca), Chloride (Cl), Iron (Fe), Conductivity (EC), Sulfate (SO4), Ammonia (NH3), Nitrate (NO3), Orthophosphate (PO4) and Total solids (T.S) for the period 2011 to 2017. The techniques used in this research for water quality assessment involve WQI, Artificial Neural Networks (ANNs) and Geographical Information System (GIS). The values of maximum for WQI reached to 632.8 and 862.8 at Al-Wathba WTP in 2011 and 2013 respectively (WTP at the middle part of Baghdad). High iron concentrations have been found in the river and are the main reason for the increase in WQI values. Neural network model was used to predict changes in the condition of Tigris River in Baghdad city. GIS was used to correlate the WQI results with the satellite image for developing colored analytical maps which classifies water quality of the river for public usage. For the NNM, the confusion matrix gave overall prediction rate about 93% for model training, 90% for testing and 70% for holdout sample. The general conclusion from this model may predict that river quality is of a poor to polluted condition, which are the same results as GIS.