Forecasting Monthly Water Quality Index Using a Seasonal ARIMA Model for Tigris River at Al-Rashediya Water Station in Baghdad City

Abstract

In this study, the quality of TGRIS River is studied at the intake of Al-Rashediya Water Station using time series analysis. 14 measured parameters of water quality, daily periods for 9 years (2013-2021), monthly mean averaged were studied which are: K+, Na+, T.S.S, T.D.S, SO42-, Cl-, Mg2+, Ca2+, T.H, Alk., E.C, pH, Turb, and Temp., from which WQI was calculated. Investigation of observed WQI time series shows that there is a simple seasonal behavior. The order of model for WQI time series was determined using auto correlation function (ACF) and partial auto correlation function (PACF). ARIMA (0, 1, 1) (autoregressive, integrated, moving average) model was found suitable to generate and forecast the quality of the river water. The fit statistic for, Stationary R-squared, R-squared, RMSE, MAPE, MaxAPE, MAE, MaxAE, and Normalized BIC criteria were used for evaluating the generation and forecasting results. Their MEAN generated for the model fit were 0.250, 0.338, 106.248, 43.119, 217.295, 73.758, 355.509, 9.419, respectively. The model statistics result for Ljung-Box Q (18) (statistics, DF, and Sig.) were 17.156,17, and 0.444 respectively.
 The above results show that time series modeling is quite capable of water quality forecasting.
 In this study of the Forecasted WQI model of the becoming 24 months for the years (2022 and 2023) were predicted, shows an increasing trend, which must be considered and managed.