Application of SARIMA for Prediction of Water Storage Levels for a Metropolitan Area: Chennai, a Case Study

摘要

In 21st Century, one of the major problems in many metropolitan cities is water scarcity. Water supply and demand is a challenging issue that is international in scope and context. Therefore, a suitable forecasting is required to supply sufficient water to any major city worldwide. Water Supply in insufficient amount against the increasing demand is a challenge for many countries including developing countries like India due to different reasons. Chennai, in India, is a city which has perennial water problem. This paper considers Chennai, as a case study and presents forecasting of water levels of the four major reservoirs from which large amount of water flows into Chennai Metropolitan Area using Time Series Approach. This work is based on historical 5.7 years of data on water reservoir level and applied Seasonal Autoregressive Integrated Moving Average (SARIMA) model of Time Series Analysis for individual reservoirs to forecast the possible curve indicating the varying trend affected by seasonality of water level of these reservoirs for future. The adequate models were selected based on criteria- Akaike Information Criteria, Root Mean Squared Error. The result shows performance of SARIMA is better than ARIMA due to seasonal influence. The objective of this work is to present forecasted water storage decreasing trend to guide the policy makers to have suitable planning for future to meet the increasing demand given the present scenario.