Simulation of the spatio-temporal extent of groundwater flooding using statistical methods of hydrograph classification and lumped parameter models

This article presents the development of a relatively low cost and rapidly applicable methodology to simulate the spatio-temporal occurrence of groundwater flooding in chalk catchments. In winter 2000/2001 extreme rainfall resulted in anomalously high groundwater levels and groundwater flooding in many chalk catchments of northern Europe and the southern United Kingdom. Groundwater flooding was extensive and prolonged, occurring in areas where it had not been recently observed and, in places, lasting for 6 months. In many of these catchments, the prediction of groundwater flooding is hindered by the lack of an appropriate tool, such as a distributed groundwater model, or the inability of models to simulate extremes adequately. A set of groundwater hydrographs is simulated using a simple lumped parameter groundwater model. The number of models required is minimized through the classification and grouping of groundwater level time-series using principal component analysis and cluster analysis. One representative hydrograph is modelled then transposed to other observed hydrographs in the same group by the process of quantile mapping. Time-variant groundwater level surfaces, generated using the discrete set of modelled hydrographs and river elevation data, are overlain on a digital terrain model to predict the spatial extent of groundwater flooding. The methodology is applied to the Pang and Lambourn catchments in southern England for which monthly groundwater level time-series exist for 52 observation boreholes covering the period 1975–2004. The results are validated against observed groundwater flood extent data obtained from aerial surveys and field mapping. The method is shown to simulate the spatial and temporal occurrence of flooding during the 2000/2001 flood event accurately