Analysis and Prediction of Groundwater Level Trends Using Four Variations of Mann Kendall Tests and ARIMA Modelling

Abstract

In this study monthly, annual and seasonal groundwater variation trends of Warangal district (2000–2015) were examined for forty observation wells using four variations of non-parametric Mann Kendal (MK) methods. Magnitudes of trends were computed using the Sen’s slope estimator. Results conclude that on a monthly time series, three observation wells (31, 32 and 37) experienced significant positive trends (positive Z-statistics), whereas other wells experienced significant negative trends among forty. Wells 2, 17, 22, 33, 38, 39, 40 noticed strong significant negative trends. Using MK-1/MK-2/MK-3/MK-4 for monthly series, around 23/33/23/23% of cases exhibited positive trends (both significant and insignificant) and 77/67/77/77 % of cases exhibited negative trends. Median of trend slopes for groundwater levels (GWLs) were negative in monthly, seasonal and annual timescale. Seasonally trend slope was observed as a decreasing trend in the pre-monsoon period. Annually trend slope was varying approximately from −0.4 mm/year to +0.4 mm/year. ARIMA models were fitted for each well to predict the parameters of the models and to forecast (2015) using monthly GWLs from 2000 to 2014. ARIMA(2,1,2), ARIMA(2,0,1), ARIMA(4,0,0), ARIMA(1,1,1), ARIMA(1,0,1), ARIMA(1,0,2) and ARIMA(2,0,0) fitted for most of the Wells. Using these models GWLs were predicted and the performance of the predicted levels was analyzed using the correlation coefficient (R2) and Akaike Information Criteria (AIC) values.