Modelling Golok River Water Quality using Multiple Linear Regression

Water is an important source for the continuation of life for the living organisms and for the smooth functioning of ecosystems, communities and economics. The water quality has been an issue for quite a long time and to monitor the water quality, Department of Environment (DOE) in Malaysia uses the Water Quality Index (WQI) to ensure that it is safe for use. However, the parameters included in the WQI does not include the salinity, turbidity and temperature that could be the factors polluting the water. This research aimed to predict the water quality at Sungai Golok using Multiple Linear Regression and to compare the water quality models using secondary data and real-time sampling data in predicting the water quality. Real-time samplings were done at three points at Sungai Golok twice a month during July and August 2019 to obtain the value for real-time sampling data. In developing models, Statistical Package for Social Science (SPSS) 20.0 software were used using Stepwise method. The result for secondary data model shows that as one value of COD increase, the WQI value will decrease for 0.180 in value, as one value of DO increase, 2.620 value of WQI will increase, as one value of NH3-N increase, 10.447 value of WQI will decrease, as one value of SS increase, 0.044 value of WQI will decrease, as one value of BOD increase, 0.737 value of WQI will decrease and as one value of temperature increase, 0.158 value of WQI will increase. For real-time sampling data, the result shows that as one value of SS increase, 1.053 value of WQI decrease, as one value of BOD increase, 1.500 value of WQI decreases and as one value of TEMP increase, 0.868 value of WQI increase. The water quality model developed from the secondary data was also found to be the best model in predicting the water quality at Sungai Golok as the variables in the model represent a larger percentage which is 97% compared to the model using real-time sampling data which is only 90.6%. Other than that, the variables included in the model using secondary data is more that the model using real-time sampling data making it to be more suitable in predicting water quality as it gives more accurate results.