FACTORS AFFECTING BIOCHEMICAL OXYGEN DEMAND CONCENTRATION IN RAYONG RIVER

This research focused on BOD estimation in Rayong river, Rayong province, Thailand. Research methodology used in this study was regression analysis technique using 9 chemical factors, dissolved oxygen, temperature in water, pH, total phosphorus, nitrate-nitrogen, nitrite-nitrogen, ammonia, suspended solids and total dissolve solution, to generate models separated four periods, January to March, April to June, July to September and October to December. Data of 319 samples was divided into two groups; training and evaluation groups. The first group was to build the regression models and the other was to validate the achievement of forecasting BOD concentration in each period. After fitting the models, assumptions of regression analysis were detected by Anderson-Darling statistic, Durbin-Watson statistic, Breusch-Pagan statistic and Variance inflation factor. Finally, the performance of forecasting BOD values was calculated by mean absolute percentage error, mean absolute error, root mean squared error and mean squared error. The results found that the regression standard errors each period model wererespectively0.6761, 0.4446, 0.2932 and 0.4743with the adjusted coefficients of determination of 0.430, 0.517, 0.306 and 0.536. The best performances of fore casting accuracy were mean absolute percentage error of 2.4278 in the first period, mean absolute error of 3.7210 in the second period, mean absolute error of 4.4017 in the third period and mean absolute percentage error of 2.5738 in the fourth period.