Optimization of Parameters in a Water Quality Index Model Using Principal Component Analysis

The Water Quality Index (WQI) model is a simple method for evaluating water quality and determining pollution levels. This approach is becoming increasingly popular as a result of its simple architecture when compared to other hydrological models. The development and implementation processes of WQI are also simple. Parameter selection is a critical step towards the development of any hydrological model, including the WQI model. According to many researchers, model parameterization is one of the key sources of uncertainty in the WQI models. Across literature, a range of techniques are used to select essential water quality features - expert opinion and literature review are the most common tools. This study aims to establish a technique for selecting important water quality parameters. The principal component analysis (PCA) was applied to filter the relevant water quality parameters. A dataset from water quality monitoring in Cork Harbour was used to test a suitability of PCA. The PCA1 component explained 41.5 percent of the total variance with strong positive loading on dissolved inorganic nitrogen (DIN), total organic nitrogen (TON), Molybdate Reactive Phosphorus (MRP), and strong negative loading on pH, dissolved oxygen (DO), and water transparency. Water biological oxygen demand (BOD), chlorophyll, and total organic nitrogen (TOC) all contributed to 25.8% of the data variance. In total, the PCA technique recommended nine significant water quality parameters from the list of eleven in terms of their relative importance. The reliability of the PCA method was assessed by intercomparison to a boxplot technique. A significant difference between the PCA and boxplot results was found to suggest that the PCA technique may not be a reliable tool for selecting crucial water quality parameters for the WQI model of coastal waters.