Characterisation of Groundwater Drought Using Distributed Modelling, Standardised Indices, and Principal Component Analysis.

Abstract

A modelling framework was developed to characterise groundwater drought at a catchment scale in the absence of adequate observational records. The framework was used to characterise historical groundwater drought events for a Chalk aquifer in southern England over the period 1971-2004 during which three major drought events occurred. A numerical groundwater model was used to simulate the groundwater level fluctuations driven by historical time-variable and spatially non-uniform recharge inputs. The standardised groundwater level index (SGI) was applied to the simulated groundwater levels to evaluate the spatial pattern of groundwater drought and of their severity and duration. A dimensionality reduction method, namely principal component analysis (PCA), was applied to the SGI dataset and to the standardised precipitation index (SPI) to further explore the spatio-temporal drought characteristics. The analysis showed inconsistency in the spatial distribution of the duration and severity among the three studied events. PCA indicated that the SPI was not a good predictor of groundwater drought during the extreme European heatwave of 2003 whereas the proposed modelling framework correctly identified the resilience of the groundwater system to that event and in line with observations. Furthermore, significant differences were observed between the spatial patterns obtained from SPI and SGI datasets highlighting the important role that hydrological and hydrogeological features of a catchment have in groundwater drought development.