Use of seasonal trend decomposition to understand groundwater behaviour in the Permo-Triassic Sandstone aquifer, Eden Valley, UK; [Utilisation d’une procédure de décomposition des séries temporelles en saisonnalité et tendance pour comprendre les variations de niveau piézométrique dans l’aquifère des grès Permo-Triassique, Eden Valley, Royaume-Uni]; [Uso da decomposição de tendência sazonal para entender o comportamento das águas subterrâneas no aquífero Permo-Triássico, Vale do Éden, Reino Unido]; [Uso de la descomposición de la tendencia estacional para comprender el comportamiento del agua subterránea en el acuífero de Arenisca Pérmico-Triásico, Eden Valley, Reino Unido]

The daily groundwater level (GWL) response in the Permo-Triassic Sandstone aquifers in the Eden Valley, England (UK), has been studied using the seasonal trend decomposition by LOESS (STL) technique. The hydrographs from 18 boreholes in the Permo-Triassic Sandstone were decomposed into three components: seasonality, general trend and remainder. The decomposition was analysed first visually, then using tools involving a variance ratio, time-series hierarchical clustering and correlation analysis. Differences and similarities in decomposition pattern were explained using the physical and hydrogeological information associated with each borehole. The Penrith Sandstone exhibits vertical and horizontal heterogeneity, whereas the more homogeneous St Bees Sandstone groundwater hydrographs characterize a well-identified seasonality; however, exceptions can be identified. A stronger trend component is obtained in the silicified parts of the northern Penrith Sandstone, while the southern Penrith, containing Brockram (breccias) Formation, shows a greater relative variability of the seasonal component. Other boreholes drilled as shallow/deep pairs show differences in responses, revealing the potential vertical heterogeneities within the Penrith Sandstone. The differences in bedrock characteristics between and within the Penrith and St Bees Sandstone formations appear to influence the GWL response. The de-seasonalized and de-trended GWL time series were then used to characterize the response, for example in terms of memory effect (autocorrelation analysis). By applying the STL method, it is possible to analyse GWL hydrographs leading to better conceptual understanding of the groundwater flow. Thus, variation in groundwater response can be used to gain insight into the aquifer physical properties and understand differences in groundwater behaviour. © 2015, The Author(s).