Pilot description and assessment : Devonian / Carboniferous Limestone aquifer aquifer (United Kingdom)

This report describes the work undertaken by the British Geological Survey (BGS/UKRI) as a part of TACTIC WP4 to calculate historical and future groundwater recharge across the outcrop of Devonian / Carboniferous Limestone aquifer and at selected observation boreholes within these aquifers. Multiple tools, selected from the TACTIC toolbox that is developed undert WP2 of the TACTIC project, have been used for this purpose. The Carboniferous Limestone aquifer in England and Wales include a wide variety of rock types that are fractured and well developed as aquifers in some areas. Where it is a major aquifer, the Carboniferous Limestone aquifer exhibit ‘karstic’ hydrogeological behaviour. The Devonian sediments of south Wales and the Welsh borderlands are continental deposits known as the Old Red Sandstone facies. In Scotland, the Devonian sandstones in Fife and eastern Scotland are one of the most productive bedrock aquifers in Scotland (Macdonald et al., 2005). Three tools have been used to estimate the recharge values. These are the lumped parameter computer model AquiMod (Mackay et al., 2014a), the transfer function-noise model Metran (Zaadnoordijk et al., 2019), and the distributed recharge model is developed ZOODRM (Mansour and Hughes, 2004). Future climate scenarios are developed based on the ISIMIP (Inter Sectoral Impact Model Inter-comparison Project (www.isimip.org) datasets. The resolution of the data is 0.5°x0.5°C global grid and at daily time steps. As part of ISIMIP, much effort has been made to standardise the climate data (e.g. bias correction). The estimation of the recharge model using the lumped model AquiMod is achieved by running the model in Monte Carlo mode. This produces many runs that are equally acceptable and consequently the uncertainty in the estimated recharge values can be assessed. The application of additional tools provides an additional mean to assess this uncertainty. Groundwater data at three boreholes are used in this model. The differences between the 75th and 25th percentile recharge values are found to be between 15% and 26%, which indicates a relatively high degree of uncertainty. In addition, the recharge values estimated using the distributed recharge model are found to be significantly higher than those estimated using the lumped model. It must be noted that the distributed recharge model calculates potential recharge while the lumped model calculates actual recharge. The absolute recharge values calculated by the transfer functionnoise model Metran are also different from those calculated by the lumped model. The transfer function model estimates lower values at two boreholes and higher values at the third borehole. Future recharge values calculated using the projected rainfall and potential evaporation values are -3.5 to 12.5% different from historical values on average. The 3o Max scenario, the wettest used in this work, produces values that are very different from the historical ones. This is observed in the output of both the lumped and the distributed models. Finally, future estimates are discussed in this report using long term average recharge values. It is recommended to carry out further analysis to these output in order to understand the temporal changes in recharge values in future, especially over the different seasons. In addition, it is recommended that the values and conclusion produced from this work should be compared to those obtained from different studies that applies future climate data obtained from different climate models.