The impacts of climate change on groundwater recharge in low storativity fractured bedrock aquifers

Abstract

The ongoing climate change is expected to cause an intensification of the hydrological cycle by enhancing the extreme events, thereby challenging the resilience of the groundwater resources. Additionally, it is well acknowledged that the magnitude of the possible effects of climate change depends on the local hydrogeological settings. Irish hydrogeology is characterised by fractured bedrock aquifers with a negligible primary porosity but a relatively well-developed secondary porosity which results into a low storage and throughput capacity of the aquifers. Hence, about two-thirds of Irish aquifers are regarded as Poorly Productive. Moreover, these poor aquifers are often overlined by till-derived soils and subsoils which are characterised by having a low permeability. The influence of these features of both hydrogeological and climate variables are been investigated by carrying out a sensitivity analysis. The results shown that the hydrogeological features ? and particularly the storage capacity - have a larger control on groundwater recharge than the meteorological variables examined. A recharge characterisation was conducted in the two main study catchments; a range of recharge calculation methods were applied to constrain the inherent uncertainties of groundwater recharge which was used as a proxy to assess the aquifer?s specific yield. The results obtained through the application of the water table fluctuation method evidenced the limited storage capacity of these aquifers, as groundwater levels often present a maximum level that is not surpassed, indicating that are unable to accept further recharge. Additionally, to better understand the effect of the overlying soils and subsoils, the infiltration capacity of two additional study sites was approximated through the implementation of Hydrus 1D in both sites. The connection between the the meteorological variables and goundwater levels was also investigated; First, the relationship between rainfall and groundwater levels was examined through the application of Wavelet Coherence (WTC) at daily resolution. The results revealed important differences on the hydrogeological behaviour depending on their structural differences. The effect of Teleconnection indices on groundwater levels was also assessed with the WTCs as they are know to explain a significant percentage of the climate variability in Ireland. The results show how some of these indices can affect groundwater levels not just at long-term but also at seasonal scale. In order to represent the hydrogeological features described above under future climate conditions, a method combining wavelet transforms (WT) and neural networks forecasts was applied: the input time series are decomposed with maximum overlap discrete wavelet transforms (MODWT) that are aggregated according to different criteria and then fed into nonlinear autoregressive neural networks (NARX). Overall, the results indicate that the decomposition of the inputs improve the performance of the model if overfitting does not occur. Furthermore, the performance is also dependant on the complexity of the structure of the catchment. Finally, the NAM rainfall-runoff model and the trainned WT-NARX models were forced with climate projections to simulate the effects of climate change on groundwater resources. The contrasting nature of the two approaches applied provide different insights on the problematic whereas the NAM provides long-term estimates and a statistic characterisation of the future possible conditions, the WT-NARX shows the suceptibility of the groundwater levels to climate variability and consitutes a useful tool to stress-test the aquifer system and identify threshold values. Despite the fundamental differences, both methods agree on forecasting an increase on groundwater flow and levels.