APPLICATION OF STATISTICAL TECHNIQUES AND DESIGN OF EXPERIMENTS IN DETERMINING RIVER WATER QUALITY

Abstract

Water is essential to life, and its contamination affects all living beings on earth. Dissolved oxygen (DO) is one of the most important water quality parameter not only for human beings but also for the survival of aquatic life. Discharge of organic, industrial, agricultural, biodegradable wastewater into rivers, results in decrease of DO concentration in downstream waters. In this work, we have considered point sources of pollution and their effect on the river water quality. Qual2K model is used for analysis of river water quality. The water quality parameters included in the model were DO, BOD, nitrogen, phosphorus and chlorophyll-a, among others. In real situations, the water quality monitoring stations are located at some distance and many point sources might discharge water into the river between two monitoring stations. In such situations, the contributions of the various point sources to the degradation in water quality become difficult to ascertain. In this work, we have tried to assess if a quantitative determination of the contribution from various point sources is feasible. This study is carried out in simulation using two point sources and DO as the water quality parameter of interest. The Qual2K model has proved to be especially useful in predicting the impact of point sources on DO in downstream water quality. The experiments for this study were chosen using statistical designs and the factors of interest were the DO of the point source and its distance from the point of sampling. The effect of the point source on water quality was measured as the decrease in DO from the pure river water DO. This decrease in DO was studied for one point source to first determine individual effect and for two point sources to find the combined effect. This is the advantage of using simulations as individual effects can be calculated which is not possible in real data. The data from simulations were analyzed using analysis of variance (ANOVA) and explicit and implicit regression models were obtained to explain the data using simple equations. Two test cases were then run to validate the model and very good agreement was found between regression model and Qual2K model.