Investigation of water quality with kohonen networks and multivariate statistical methods: Application in Antalya Aksu region

Sustainable use and protection of surface and groundwater resources is a priority for integrated water management. Evaluating and interpreting monitoring data regarding both surface and groundwater quality in temporal and spatial terms is very important for protecting and improving water quality. In recent years, the use of multivariate statistical analysis methods has become widespread because they enable complex data sets to be analyzed and visualized easily. In this thesis study, it is aimed to examine the quality of Aksu Stream surface water and Altınova region groundwater using Principal Component Analysis (PCA), Hierarchical Clustering Analysis (HCA) and Kohonen Networks - Self-Organizing Maps (SOM) methods, which are among the multivariate statistical methods. Although water quality monitoring studies have been carried out before in the specified study areas, temporal and spatial data analysis has not been carried out using multivariate statistical analysis methods. In addition, statistical analyses were conducted by combining the water quality monitoring results of Altınova region with index results of vulnerability to groundwater pollution. The thesis study contains originality in terms of the statistical methods used. With the results obtained, the water quality of the two study areas was comprehensively evaluated. With HCA, water quality data is grouped according to certain or similar characteristics. With PCA, the relevant main components were revealed by simplifying the data sets in order to reduce the complexity of the factors affecting water quality. Similarities of changes in water quality parameters have been revealed with Kohonen Networks. KEYWORDS: Hierarchical Clustering Analysis, Kohonen Networks, Principal Component Analysis, Statistical analysis, Water quality index