Identification and modelling of water quality in river streams

River pollution is a major environmental problem negatively effects on life. In order to prevent this environmental problem, first of all sources of pollution must be determined by monitoring water quality in river basins. Investigating the effect of determined pollution sources, the necessary measures must then be taken for control of environmental pollution. Creation of models and computer simulation of water quality are very important for prediction of the potential adverse effects of pollution sources along the length of stream, and thus determining of the possible effects of investments to stream. In the water quality models, the parameters, determined through optimization without the need for trial and error method, is required to ensure the reliability of the model. In this thesis, studies on the field and in the laboratory, water quality parameters of Beylerderesi River located in the province of Malatya, were analyzed. For the dynamic modeling of the stream, continuous stirred tank reactor (CSTR) approach was used. Beylerderesi stream is modeled as a dynamic model, the kinetic parameters were found by optimization. Determination of the model parameters was made according to minimization of an objective function generated by the sum of the squares of the differences between predictive values and experimental values, all the observations taken during the sampling points of a dynamic state variables. For the step of optimization, SQP ('Sequential Quadratic Programming') method was used. The model, based on the measurements of 9 status variable of streams and pollution loads (ammonium, nitrite, nitrate, and organic nitrogens, and dissolved organic phosphors, 5-day biochemical oxygen demand, dissolved oxygen, and chlorine), was simulated. The mean absolute percentage errors by comparing simulation results with the experimental measurements, along the length of 20 km of stream, were calculated. According to the results the model predictions with the experimental data has been observed that a significant degree of alignment. One of the most important indicators of stream biochemical oxygen demand and dissolved oxygen in the average absolute error values, for the variables calculated, were obtained as 1.39% and 0.95%, respectively. Keywords: River Modeling, Dynamic Simulation, Parameter Optimization.