Characterising groundwater drought: Using standard indices and cluster analysis to quantify drought response and propagation across aquifer typology and identify areas of resilience

Groundwater is a resilient resource that is vital for local water supplies and maintaining baseflow in rivers and streams, particularly during low flow periods. Due to global climate change, the occurrence of meteorological drought is increasing in both frequency and severity, causing or amplifying water scarcity events even in areas considered water rich, like Scotland. While groundwater is generally more resilient to drought than surface water sources, the specific impacts on the resilience of the resource is influenced by local hydrogeology, geography, and climate. To evaluate differences in groundwater response between sites and assess vulnerability and resilience at varying timescales, it is important have standardised parameters to evaluate.

The Standardised Groundwater Index (SGI) is a normalisation procedure that can be applied to groundwater level data from observation boreholes to compare drought response more easily between locations. To standardise groundwater data, typical methods use a specific probability distribution which is unlikely to represent variability over large, diverse regions across different seasons or empirical probabilities requiring large sample sizes. Here, in the transformation to standardised units, probability distributions are optimised using Akaike Information Criterion (AIC) to select the most appropriate distribution for each season at each location. Incorporating model fit statistics for each time and site reduces uncertainty in calculations, particularly at the tails of the distribution which is vital for drought studies.

Groundwater storage and memory is evaluated across 33 sites in Scotland through the autocorrelation function of the SGI time series and correlated with Standardised Precipitation Evapotranspiration Index (SPEI) to evaluate the time scale of groundwater drought propagation at each location and better characterise storage properties for aquifers of different lithologies and dominant flow types (e.g. intergranular, fractured). Autocorrelation lengths of less than 5 months are common in the fractured flow systems compared to lag periods of 9 months or greater in more highly transmissive aquifers likely dominated by intergranular flow.

Hierarchical cluster analysis of the SGI time series provides an added line of evidence to the differential response between hydrogeological units and to identify areas where local changes in geology, structure, or surface water connections could be influencing groundwater response. Characterisation of the groundwater drought response can reveal areas of greater groundwater resilience and provide water managers better metrics to assess the spatiotemporal controls of groundwater drought propagation, along with modelling the timing and magnitude of seasonal groundwater minima.