Modeling of Selected River Water Quality Indicators using Autoregressive Integrated Moving Average (ARIMA) Techniques

Abstract

This study applies the Autoregressive Integrated Moving Average (ARIMA) modelling technique to Predict key water quality indicators of the river Benue in Jimeta, Yola, Nigeria. Utilizing a ten-year dataset (2011–2021) obtained from the Adamawa State Ministry of Water Resources. The research focused on three essential parameters: pH, calcium (mg/L), and iron (mg/L). Following the Box-Jenkins methodology, the data were analyzed for stationarity for test using the Augmented Dickey-Fuller test, model identification via ACF/PACF analysis, parameter estimation, diagnostic checking, and forecasting. Results indicated that the ARIMA (0,0,1) model best fits the pH and iron data showing values are stable, primarily influenced by short-term random shocks (MA process)., while the ARIMA (1,0,0) model suits calcium, indicating values fluctuate more, influenced by their immediate past value (AR process). Forecasting results showed a stable average pH of 7.23, fluctuating calcium levels averaging around 61.85 mg/L, and a consistent iron concentration of approximately 0.1523 mg/L over the projected ten-year period (2023–2032). Diagnostic checks confirmed that all selected models were stable, stationary, and invertible, with no unit roots. These findings demonstrate ARIMA’s effectiveness in capturing temporal dynamics in water quality and provide a reliable foundation for proactive environmental monitoring, planning, and decision-making.