Nonlinear time series based forecast modelling of groundwater flooding at karst region in Ireland

The intensity and occurrence of groundwater flooding have been found to increase considerably in the past two decades. In general, groundwater flooding occurs over a larger area and for a longer duration, when compared to other types of flooding. The associated damage and disruption caused by groundwater flooding led researchers to focus on the development of groundwater flood forecasting models that can be used for flood risk assessment and development of flood mitigation planning and strategies. This study develops a nonlinear time series model to predict total flooded volume (TFV) in a lowland karst area of Ireland. A Nonlinear Autoregressive model with Exogenous variables (NARX) was developed in the karst region, where rainfall and tidal amplitude had been considered to influence the TFV. The developed NARX model was found to predict TFV with considerable accuracy up to 30 days ahead, with a Kling-Gupta Efficiency (KGE) value of 0.9 or above. The efficiency deteriorates beyond 30 days ahead prediction and becomes 0.81 KGE when the prediction window is 90 days. Comparison of the developed NARX model with a linear time series model in TFV forecast indicates the importance of considering the nonlinear terms while developing the forecasted model. The developed NARX model has the potential to create an early warning system for flooding. The model has further been used to predict freshwater discharge from the inter-tidal spring into the Atlantic Ocean in southern Ireland.