Breakthrough dynamics of s-metolachlor metabolites in drinking water wells: Transport pathways and time to trend reversal.

We present the results of a two years study on the contamination of the Luxembourg Sandstone aquifer by metolachlor-ESA and metolachlor-OXA, two major transformation products of s-metolachlor. The aim of the study was twofold: (i) assess whether elevated concentrations of both transformation products (up to 1000 ng/l) were due to fast flow breakthough events of short duration or the signs of a contamination of the entire aquifer and (ii) estimate the time to trend reversal once the parent compound was withdrawn from the market. These two questions were addressed by a combined use of groundwater monitoring, laboratory experiments and numerical simulations of the fate of the degradation products in the subsurface. Twelve springs were sampled weekly over an eighteen month period, and the degradation rates of both the parent compound and its transformation products were measured on a representative soil in the laboratory using a radiolabeled precursor. Modelling with the numeric code PEARL simulating pesticide fate in soil coupled to a simple transfer function model for the aquifer compartment, and calibrated from the field and laboratory data, predicts a significant damping by the aquifer of the peaks of concentration of both metolachlor-ESA and -OXA leached from the soil. The time to trend reversal following the ban of s-metolachlor in spring protection zones should be observed before the end of the decade, while the return of contaminant concentrations below the drinking water limit of 100 ng/l however is expected to last up to twelve years. The calculated contribution to total water discharge of the fast-flow component from cropland and short-circuiting the aquifer was small in most springs (median of 1.2%), but sufficient to cause additional peaks of concentration of several hundred nanograms per litre in spring water. These peaks are superimposed on the more steady contamination sustained by the base flow, and should cease immediately once application of the parent compound stops.

On Using Lumped Parameter Models and Temperature Cycles in Heterogeneous Aquifers.

The use of the annual cycles of stable isotopes to estimate the parameters of transit time distribution (TTD) functions has been recently criticized by Kirchner who showed that if the catchment is heterogeneous the mean residence time calculated from the damping of the amplitude of the input signal is often underestimated by large factors. We modify Kirchner's initial analysis and show that the estimation bias is a function of the degree of heterogeneity. In low heterogeneity cases corresponding to permeable aquifers, the TTD of a simple exponential model developed assuming a homogeneous medium deviates only slightly from the true TTD, and thus estimated and true mean transit times are close. We also develop and test for a sandstone aquifer a model predicting the temperature amplitude of water at the outlet of a groundwater system from the annual air temperature cycle, and show that even though temperature is not a conservative tracer, it can be used to calibrate a lumped parameter model describing the TTD of the aquifer.

Estimating Pesticide Attenuation From Water Dating and the Ratio of Metabolite to Parent Compound.

Although pesticides are primarily degraded in the topsoil, significant attenuation can be expected in groundwater systems where the transit time of pesticides usually are orders of magnitude longer than in the soil. Because degradation and transport processes in the subsurface take place at time scales of months to years or even decades, direct measurements of natural attenuation are hampered by practical and logistical limitations (for instance the limited duration of sampling or a correct estimation of the pesticide flux into groundwater). Indirect methods such as measuring the changes in the ratio of degradation product to parent compound as a function of transit time in the aquifer, along a flow line provide a possible alternative. This paper presents a simple mathematical formulation of the relationship between transit time in the subsurface and changes in that ratio, and allows estimating the transformation rate of both parent compound and degradation product. The applicability of the method is illustrated in a case study investigating atrazine attenuation in a fractured sandstone aquifer.

A case-study on the accuracy of mass balances for xenobiotics in full-scale wastewater treatment plants.

Removal efficiencies of micropollutants in wastewater treatment plants (WWTPs) are usually evaluated from mass balance calculations using a small number of observations drawn from short sampling campaigns. Since micropollutant loads can vary greatly in both influent and effluent and reactor tanks exhibit specific hydraulic residence times, these short-term approaches are particularly prone to yield erroneous removal values. A detailed investigation of micropollutant transit times at full-scale and on how this affects mass balancing results was still lacking. The present study used hydraulic residence time distributions to scrutinize the match of influent loads to effluent loads of 10 polar micropollutants with different influent dynamics in a full-scale WWTP. Prior hydraulic modeling indicated that a load sampled over one day in the effluent is composed of influent load fractions of five preceding days. Results showed that the error of the mass balance can be reduced with increasing influent sampling duration. The approach presented leads to a more reliable estimation of the removal efficiencies of those micropollutants which can be constantly detected in influents, such as pharmaceuticals, but provides no advantage for pesticides due to their sporadic occurrence. The mismatch between sampled influent and effluent loads was identified as a major error source and an explanation was provided for the occurrence of negative mass balances regularly reported. This study indicates that the accurate determination of global removal values is only feasible in full-scale investigations with sampling durations much longer than 1 day. In any case, the uncertainty of these values needs to be reported when used in removal assessment, model selection or validation.