Surface water quality prediction using data mining method (Case study: Rivers of northern side of Sahand Mountain)

Monitoring and assessment of surface water quality are a very expensive and time consuming process, thus finding cheap, simple and relatively excat methods which determine water quality class based on minimum parametrs would be very useful. Decision tree as one of the data mining techniques classify data sets based on a tree structure. In this study, the decision tree method was used to classify water quality in some hydrometry stations located at northern side of Sahand Mountain, including Bostanabad, PoleSenikh, Lighvan and Vanyar. The water quality classes was defined based on if-then rules. For every considered river, the discharge and 12 hydrochemical parameters, including Ca2, Mg2, Cl-, HCO3-, Na%, pH, SO42-, total anions, total cations, total dissolved solids (TDS), sodium adsorption ratio (SAR) and Electrical conductivity (EC) were collected and used for developing desicion tree model. The results showed that the desicion tree model could evaluate water quality class with high accuracy base on only four parameters: EC, pH, SAR and Na. As the error of developed models in testing phase for Bostanabad, Vanyar, PoleSenikh and Lighvan stations were 3.4, 8.1, 22.9 and 1.6%, respectively.