Spatial-environmental Assessment of Drinking Water Quality Under Salinity Stress: An Integrated WQI, Removal Efficiency and ANN Approach in Basra, Iraq

A spatial-environmental assessment of the quality of treated drinking water at Al-Baradiyah Water Treatment Plant, Basra (Iraq), in response to increasing salinity intrusion into the Shatt al-Arab River Data-driven based integration methodology were investigated by employing Water Quality Index (WQI), removal efficiency (RE) analysis, GIS-based spatial mapping and artificial neural network (ANN) modelling. Monthly water samples from both raw and treated sources were collected over 2024–2025 for major physicochemical parameters analysis. Results indicated a WQI of 90.83 for treated water which can be classified into good status; however, an extended month-to-month examination of WQI shows values varying from 58.66 to108.72 that demonstrate that mean based calculations mask important time-sensitive declines present in salinity peaks. Salinity-parameters (TDS, EC, chloride, sulphate and total hardness) consistently surpassed the WHO guideline values indicating chronic hadrochemical stress. Suspended constituents (e.g., turbidity and TSS) were routinely removed more effectively than dissolved ionic parameters, for which removal was moderate during unsalted periods or variable/occasional negative in higher-salinity conditions. GIS mapping-based spatial analysis showed a downstream increase in salinity indicators along the Shatt al-Arab, identifying a disproportionate up-estuary (uniproxy) salinisation ‘hotspot’ at station SH4 alone within Al-Faw area and linking plant-scale performance to river-streaming processes. The turbidity prediction model which was created with artificial neural networks (ANN) performed with acceptably accuracy, presented by determination coefficient R=0.9125; R²=0.8326, and thus validated adaptive water management under nonlinear hadrochemical conditions. The results reinforce that WQI itself cannot sufficiently evaluate water quality under salt stress but must be combined with spatiotemporal, parameter-specific and predictive analysis-held approaches for robust water resource planning.