Evaluation of spatiotemporal changes in water quality using statistical methods: Acisu Creek application

The aim of this study was to reveal the spatiotemporal changes in water quality and the main factors causing these changes by using the results of water quality monitoring studies carried out in the Acısu Creek between 2020-2021. Hierarchical Cluster Analysis (HCA) and Principal Component Analysis (PCA) were performed using the measurement and analysis results of 37 water quality parameters obtained from 12 different monitoring points for 12 months in the Acısu Creek. In addition, Self-Organizing Maps (Kohonen Networks-SOM) applications were performed using the data sets of 30 water quality parameters. As a result of the HCA, both sampling points and sampling dates were grouped according to their similarity. HCA clustered the sampling points into two different clusters with similar type and degree of water pollution. HCA revealed the water quality heterogeneity in the monitoring results and grouped the sampling dates into two main clusters. According to the PCA results, 69.74% of the total variance in the Acısu Creek water quality was explained by the first 3 axes and 91.22% of the total variance was explained by the first 8 axes. Kohonen Networks-SOM analysis was effective to form clusters according to the physicochemical and bacteriological water quality and anion-cation groups. As a result, the SOM analysis results were confirmed by the PCA results. This study demonstrated that multivariate statistical methods and artificial neural networks are powerful tools to evaluate spatiotemporal variation of water quality and the main factors for the changes.