An ensemble approach for predicting future groundwater levels in the Zagreb aquifer impacted by both local recharge and upstream river flow

Abstract

This paper presents the first study assessing the climate change impact on groundwater levels of the Zagreb alluvial aquifer in Croatia by coupling climate projections under RCP8.5 for the period 2040–2070 with local scale groundwater flow modelling. As a novelty in groundwater modelling of climate change impacts, this study utilizes an ensemble of five climate models and two land surface models for providing projections of relevant boundary conditions to the groundwater model. The groundwater model is used for predicting climate driven changes in groundwater levels in the aquifer system influenced by both changes in local groundwater recharge and changes in river-flow. Both boundary conditions are obtained from the land surface models. In addition, the uncertainty contributions to groundwater levels from both climate models and land surface models were quantified using variance decomposition. The results revealed that the spatial pattern of changes in groundwater levels can be related to different processes influencing groundwater dynamics. In areas dominated by a strong groundwater-surface water interaction, close to the Sava river, the multi-model ensemble mean change in groundwater levels is negligible for low and average water levels. However, for high water levels, an increase was identified. In contrast, a decline in groundwater levels prevails in all areas where aquifer recharge is largely driven by rainfall infiltration. For the variance decomposition analysis, climate models were identified as the main source of uncertainty for groundwater levels. However, the uncertainty contribution of the choice of land surface model to the impact simulations was found to be more important in the area affected by a strong groundwater–surface water interaction and for periods of high groundwater levels. Also, specific local issues related to low and high groundwater levels that are associated with extreme hydrological regimes were found to exacerbate under climate projections.