Developing a Data-Fused Water Quality Index Based on Artificial Intelligence Models to Mitigate Conflicts between GQI and GWQI

The study of groundwater quality is typically conducted using water indices such as the Groundwater Quality Index (GQI) or GroundWater (GWQI). are calculated field data and a scoring system that uses ratios constituents to prescribed standards weights based on each constituent’s relative importance. results obtained by this procedure suffer from inherent subjectivity, consequently may have some conflicts between different indices. An innovative feature drives research mitigate in GQI GWQI predictive power artificial intelligence (AI) models integration multiple indicators into one representative index concept fusion through catastrophe theory. This employed two-level AI modeling strategy. In Level 1, three were calculated: GQI, GWQI, data-fusion four pollutants including manganese (Mn), arsenic (As), lead (Pb), iron (Fe). Further was applied at 2 supervised learning methods, Mamdani fuzzy logic (MFL), support vector machine (SVM), neural network (ANN), random forest (RF), with 1 inputs, data-fused target values derived targets. We these methods Gulfepe-Zarinabad subbasin northwest Iran. show all performed reasonably well, difference negligible root mean square errors (RMSE), coefficient determination (r2) metrics. RF (r2 = 0.995 RMSE 0.006 test phase) MFL (r 0.921 0.022 had best worst performances, respectively. indicate results. method presented can also be other aquifers.