Effect of grain size variation on strontium sorption to heterogeneous aquifer sediments.

Strontium-90 (90Sr) is a major contaminant at nuclear legacy sites. The mobility of 90Sr is primarily governed by sorption reactions with sediments controlled by high surface area phases such as clay and iron oxides. Sr2+ adsorption was investigated in heterogeneous unconsolidated aquifer sediments, analogous to those underlying the UK Sellafield nuclear site, with grainsizes ranging from gravels to clays. Batch sorption tests showed that a linear Kd adsorption model was applicable to all grainsize fractions up to equilibrium [Sr] of 0.28 mmol L-1. Sr2+ sorption values (Kd; Langmuir qmax) correlated well with bulk sediment properties such as cation exchange capacity and surface area. Electron microscopy showed that heterogeneous sediments contained porous sandstone clasts with clay minerals (i.e. chlorite) providing an additional adsorption capacity. Therefore, gravel corrections that assumed that the > 2 mm fractions are inert were not appropriate and underestimated Kd(bulk) adsorption coefficients. However, Kd (<2 mm) values measured from sieved sediment fractions, were effectively adjusted to within error of Kd (bulk) using a surface area dependant gravel correction based on particle size distribution data. Amphoteric pH dependent Sr2+ sorption behaviour observed in batch experiments was consistent with cation exchange modelling between pH 2-7 derived from the measured cation exchange capacities. Above pH 7 model fits were improved by invoking a coupled cation exchange/surface complexation which allowed for addition sorption to iron oxide phases. The overall trends in Sr2+ sorption (at pH 6.5-7) produced by increasing solution ionic strength was also reproduced in cation exchange models. Overall, the results showed that Sr2+ sorption to heterogeneous sediment units could be estimated from Kd (<2 mm) data using appropriate gravel corrections, and effectively modelled using coupled cation exchange and surface complexation processes.

Comparing the Fate and Transport of MS2 Bacteriophage and Sodium Fluorescein in a Karstic Chalk Aquifer.

Groundwater flow and contaminant migration tracing is a vital method of identifying and characterising pollutant source-pathway-receptor linkages in karst aquifers. Bacteriophages are an attractive alternative tracer to non-reactive fluorescent dye tracers, as high titres (>1012 pfu mL-1) can be safely released into the aquifer, offering improved tracer detectability. However, the interpretation of bacteriophage tracer breakthrough curves is complicated as their fate and transport are impacted by aquifer physicochemical conditions. A comparative tracer migration experiment was conducted in a peri-urban catchment in southeast England to characterise the behaviour of MS2 bacteriophage relative to sodium fluorescein dye in a karstic chalk aquifer. Tracers were released into a stream sink and detected at two abstraction boreholes located 3 km and 10 km away. At both sites, the loss of MS2 phage greatly exceeded that of the solute tracer. In contrast, the qualitative shape of the dye and phage breakthrough curves were visually very similar, suggesting that the bacteriophage arriving at each site was governed by comparable transport parameters to the non-reactive dye tracer. The colloid filtration theory was applied to explain the apparent contradiction of comparable tracer breakthrough patterns despite massive phage losses in the subsurface. One-dimensional transport models were also fitted to each breakthrough curve to facilitate a quantitative comparison of the transport parameter values. The model results suggest that the bacteriophage migrates through the conduit system slightly faster than the fluorescent dye, but that the former is significantly less dispersed. These results suggest that whilst the bacteriophage tracer cannot be used to predict receptor concentrations from transport via karstic flow paths, it can provide estimates for groundwater flow and solute contaminant transit times. This study also provides insight into the attenuation and transport of pathogenic viruses in karstic chalk aquifers.

Groundwater flow velocities in karst aquifers; importance of spatial observation scale and hydraulic testing for contaminant transport prediction.

We review scale dependence of hydraulic conductivities and effective porosities for prediction of contaminant transport in four UK karst aquifers. Approaches for obtaining hydraulic parameters include core plug, slug, pumping and pulse tests, calibration of groundwater flow models and spring recession curves. Core plug and slug tests are unsuitable because they do not characterize a large enough volume to include a representative fracture network. Pumping test values match regional-scale hydraulic conductivities from flow modelling for the less intensively karstified aquifers: Magnesian Limestone, Jurassic Limestone and Cretaceous Chalks. Reliable bulk hydraulic conductivities were not available for the intensively karstified Carboniferous Limestone due to dominance of flow through pipe conduits in Mendips. Here, the only hydraulic conductivity value found from spring recession is one order of magnitude higher than that indicated by pumping tests. For all four carbonate aquifers, effective porosities assumed for transport modelling are two orders of magnitude higher than those found from tracer and hydrogeophysical tests. Thus, a combination of low hydraulic conductivities and assumed flowing porosities resulted in underestimated flow velocities. The UK karst aquifers are characterized by a range of hydraulic behaviours that fit those of karst aquifers worldwide. Indeed, underestimation of flow velocity due to inappropriate parameter selection is common to intensively karstified aquifers of southern France, north-western Germany and Italy. Similar issues arise for the Canadian Silurian carbonates where the use of high effective porosities (e.g. 5%) in transport models leads to underestimation of groundwater velocities. We recommend values in the range of 0.01-1% for such aquifers.

Prediction of contaminant transport in fractured carbonate aquifer types: a case study of the Permian Magnesian Limestone Group (NE England, UK).

Viruses and bacteria which are characterized by finite lives in the subsurface are rapidly transported via fractures and cavities in fractured and karst aquifers. Here, we demonstrate how the coupling of a robust outcrop characterization and hydrogeophysical borehole testing is essential for prediction of contaminant velocities and hence wellhead protection areas. To show this, we use the dolostones of the Permian Magnesian Limestone aquifer in NE England, where we incorporated such information in a groundwater flow and particle tracking model. Within this aquifer, flow in relatively narrow (mechanical aperture of ~ 10-1-1 mm) fractures is coupled with that in pipe cavities (~ 0.20-m diameter) following normal faults. Karstic cavities and narrow fractures are hydraulically very different. Thus, the solutional features are represented within the model by a pipe network (which accounts for turbulence) embedded within an equivalent porous medium representing Darcian flowing fractures. Incorporation of fault conduits in a groundwater model shows that they strongly influence particle tracking results. Despite this, away from faulted areas, the effective flow porosity of the equivalent porous medium remains a crucial parameter. Here, we recommend as most appropriate a relatively low value of effective porosity (of 2.8 × 10-4) based on borehole hydrogeophysical testing. This contrasts with earlier studies using particle tracking analyses on analogous carbonate aquifers, which used much higher values of effective porosity, typically ~ 102 times higher than our value, resulting in highly non-conservative estimates of aquifer vulnerability. Low values of effective flow porosities yield modelled flow velocities ranging from ~ 100 up to ~ 500 m/day in un-faulted areas. However, the high fracturing density and presence of karstic cavities yield modelled flow velocities up to ~ 9000 m/day in fault zones. The combination of such flow velocities along particle traces results in 400-day particle traces up to 8-km length, implying the need for large well protection areas and high aquifer vulnerability to slowly degrading contaminants.