K-means method for clustering groundwater chemical parameters of Konya Closed basin

Abstract

Cluster analysis, respectively K-Means method, is used to classify groundwater from Konya Closed Basin, located in Central Anatolia Region of Turkey. Datasets include the beginning (MB) and end (MS) of the season values for pH, Na and HCO3 spanning 2015–2022 period. There are 163 water wells (stations) in Konya Closed Basin where chemical properties were measured. Additionally, an EDA and statistical analysis was performed. The international standards for pH, Na and HCO3 in drinking water and irrigation water are generally respected. To evaluate the success of cluster separation using the K-Means technique, we estimated the silhouette coefficient. The results reveal that, generally, 2 to 4 clusters were determined for all models proposed. This approach can help to identify those wells that exceed the standard values of the investigated chemical parameters for drinking and irrigation water. In this way, the water management department could take improvement measures over time.