Water supply in hard rock coastal regions The effect of heterogeneity and kinematic porosity

Sammanfattning: Water resources in hard rock terrain are difficult to characterize due to heterogeneity and anisotropy in the fracture network, low porosities and limited recharge volumes available during the summer season. Three methods were developed and evaluated in order to assist in water supply planning. A groundwater resources potential index was estimated using multivariate statistics, where physical and geological variables were classified using Analysis of Variance and Fisher's Least Significant Difference tests according to their effect on hydraulic properties. Principal component analysis was used to assign weights to the different classed variables. Classes and weights were used to produce an index referred to as groundwater resources potential (GRP), which correlated significantly with well data. Nearly 80% of the wells with less than median specific capacity values also had GRP values at those locations of zero or lower. Non-stationary variance was observed in specific capacity sub-samples taken from the Geological Survey of Sweden's well archive, despite homogeneous geology and topography. Spatial statistical analyses showed that spatial correlations were weak in well archive samples, implying that regional approximations based on sparse point data are highly error prone. Kinematic porosity estimated using superficial fracture measurements correlated significantly with well archive data. However, low correlation coefficients indicated that well data is likely not a suitable method for predicting water supply characteristics. This approach is an efficient method which shows promise in preliminary estimations of groundwater storage in heterogenic terrains. A groundwater balance model which describes seasonal groundwater storage changes was created in order to better approximate the groundwater situation often found in Swedish urbanized and semi-urbanized hard rock terrains. The model was based on a water budget approach at the pixel scale, and allows for approximation of well extraction which is not uniformly distributed in space. The model showed that in specific regions groundwater extraction may lead to severe decreases in groundwater level, where these impacts may not otherwise be expected. Dry season modelling with 10% increased evapotranspiration showed that in several areas groundwater reservoir depletion may be influenced by more than 50%.