Trend Identification and Stochastic Modeling of Groundwater Levels in Sagar District of Madhya Pradesh

The study investigates the application of time series methods for identification of trends and subsequent forecasting of groundwater levels in few blocks of Sagar district, which faces severe water scarcity owing to declining ground water levels. A series of tri-annual ground water level observations made for 15 to 17 years is used. The available data was divided into two sets i.e. the last two years for forecasting and the remaining initial years for evaluating the parameters of the autoregressive model. Kendal's rank correlation test has been applied to identify the trend persisting in the data and the linear regression test is used to identify the significance of the slope. Univariate time series modeling was attempted by applying the auto-regressive model and the resulting accuracy is compared. The analysis indicates that the time series of groundwater levels are cyclical, with characteristics of seasonal variation in all the blocks, coupled with a declining trend at Sagar, Khurai and Bina. The study attempts to model the two different types of groundwater level data; one with trend persisting and the other having no trend with the autoregressive (AR) models. It is well demonstrated that AR models can be effectively used to model the time series of groundwater levels in Sagar district of Madhya Pradesh.