A Stochastic Modelling Technique for Predicting Groundwater Table Fluctuations with Time Series Analysis

Abstract

This paper demonstrates a methodology for modeling the groundwater table fluctuations observed in a shallow unconfined aquifer of Bangladesh. For this purpose, Kushtia district of Bangladesh has been selected as a case study area. A groundwater table monitoring well is selected in each upazilla (sub-district) under the study area. The time series water table observations collected on a weekly basis during the period from 1999 to 2006 from each well is used for experiments. Empirical result demonstrates that observed data of groundwater level exhibits cyclic patterns and shows an annual periodicity. Box and Jenkins univariate stochastic models widely known as ARIMA, are applied to simulate groundwater table fluctuations in all monitoring wells under consideration in the study area. The predicted data using the best models are compared to the observed time series and the resulting accuracy is checked using different error parameters. The simulation revealed that ARIMA models generate reasonably accurate forecasts in terms of numerical accuracy. The results also showed that the predicted data represent the actual data very well for each monitoring well.