Nonlinear analysis of groundwater levels: investigating trends and the impact of El Niño on groundwater drought in a southern region of India

Abstract

The expansion of groundwater irrigation and the cultivation of water-intensive sugarcane, combined with low rainfall, have exacerbated groundwater depletion and intensified droughts in the semi-arid Upper Krishna basin, India. This study employs an iterative singular spectrum analysis (iterative SSA) approach to impute missing groundwater level data from 25 monitoring wells. Cross-validation results show that iterative SSA effectively preserves the overall data structure when missing data was random, achieving good performance metrics with NSE > 0.79, R² > 0.8 and RMSE < 0.88 under optimal parameters (L = 12 and k = 5). The reconstructed groundwater levels were then used to identify nonlinear trends with a 180-month smoothing SSA window and to investigate the impact of strong El Niño events on groundwater drought through cross-wavelet transform (XWT) and wavelet coherence (WTC) analyses between 1983 and 2017. The nonlinear trends revealed short-term deviations in groundwater levels during 1991–2000, 2002–2003, and 2015–2017. These deviations were corroborated by significant cross-wavelet power and high wavelet coherence between the Niño 3.4 SST Index and groundwater drought, particularly under low rainfall conditions, indicating stress on the region’s groundwater system. Although the study effectively captures the nonlinear nature of groundwater levels and the influence of climate variability on drought, the complexity of the groundwater system in the region persists due to physical water scarcity and high groundwater extraction for irrigation. This study highlights the importance of imputing missing data and applying nonlinear trend and wavelet analyses to detect short-term deviations caused by severe droughts, driven by strong El Niño events and high irrigation demands.