Présentations Colloques

Oral Presentation
Session 8.05: Karst aquifers
Gill Laurence
Groundwater flooding in a lowland karst network in response to extreme rainfall and tidal event synchronicity
The objective of this research was to develop a black box transfer function model of a lowland karst network in order to assess groundwater flooding in the area in response to combinations of extreme rainfall and tidal events. The study has focused on a lowland karst network in the west of Ireland fed by allogenic runoff from low permeability Devonian mountains and discharges into a bay below mean sea level. The temporal dynamics of localised groundwater-surface water interactions have been studied for several years to yield information about the nature of the hydraulic connections beneath the ground. From this a deterministic hydraulic hydrological distributed pipe network model of the system has been developed. This model predicts the outflow from the main spring into the sea which has been validated against discharge estimates using conductivity profiles and radon concentrations. **In response to severe flooding events in November 2009 and December 2015, this current research has now characterised the hydrogeology of whole karst network by a single transfer function to investigate the impact of the two main drivers on flooding (rainfall and tidal level) in the area. The data used to develop the black box model was a 12 year time series of flooded storage, rainfall and tide levels (all from monitoring) and the spring outflow discharge (from the calibrated pipe-network model). Frequency analysis of the data sets was then carried out using Fast Fourier transform analysis and a transfer function based upon a discrete wavelet function has been derived to characterise this inherently non-stationary behaviour of the flooding in the karst system. The results suggest that the extent of flooding is related to the synchronicity of heavy rainfall and perigean (i.e. maximum) spring tides Historical flooding of the area back to the 1900s have then been compared with the predictions of the model. This knowledge can be used to make more reliable flood management predictions in the future in order to help to protect local communities.**