Multi-model approach to quantify groundwater-level prediction uncertainty using an ensemble of global climate models and multiple abstraction scenarios

Abstract

Worldwide, groundwater resources are under a constant threat of overexploitation and pollution due to anthropogenic and climatic pressures. For sustainable management and policy making a reliable prediction of groundwater levels for different future scenarios is necessary. Uncertainties are present in these groundwater-level predictions and originate from greenhouse gas scenarios, climate models, conceptual hydro(geo)logical models (CHMs) and groundwater abstraction scenarios. The aim of this study is to quantify the individual uncertainty contributions using an ensemble of 2 greenhouse gas scenarios (representative concentration pathways 4.5 and 8.5), 22 global climate models, 15 alternative CHMs and 5 groundwater abstraction scenarios. This multi-model ensemble approach was applied to a drought-prone study area in Bangladesh. Findings of this study, firstly, point to the strong dependence of the groundwater levels on the CHMs considered. All groundwater abstraction scenarios showed a significant decrease in groundwater levels. If the current groundwater abstraction trend continues, the groundwater level is predicted to decline about 5 to 6 times faster for the future period 2026–2047 compared to the baseline period (1985–2006). Even with a 30 % lower groundwater abstraction rate, the mean monthly groundwater level would decrease by up to 14 m in the southwestern part of the study area. The groundwater abstraction in the northwestern part of Bangladesh has to decrease by 60 % of the current abstraction to ensure sustainable use of groundwater. Finally, the difference in abstraction scenarios was identified as the dominant uncertainty source. CHM uncertainty contributed about 23 % of total uncertainty. The alternative CHM uncertainty contribution is higher than the recharge scenario uncertainty contribution, including the greenhouse gas scenario and climate model uncertainty contributions. It is recommended that future groundwater-level prediction studies should use multi-model and multiple climate and abstraction scenarios.