Establishing Metrics to Quantify the Vulnerability of Municipal Supply Wells to Contaminants from Surface Water Sources

Abstract

This study was completed in conjunction with the Regional Municipality of Waterloo and the Southern Ontario Water Consortium, with assistance from the Canadian Water Network and Grand River Conservation Area with the main focus of the field study in the Alder Creek Watershed, near Kitchener, Ontario. The main objective of this study was to evaluate the utility of a broad range of field site characterization techniques designed to assess the vulnerability of public supply wells to water quality impacts from surface water sources. This was carried out through detailed field investigations at the site of an existing supply well, managed by the Regional Municipality of Waterloo. Focus was placed on determining which data would be most useful to collect to evaluate well vulnerability during extended pumping tests. Connections between different parameters were also important in this investigation for their potential to act as data surrogates, where easier to measure and more inexpensive parameters could advise on otherwise difficult to collect data. The main intention was to evaluate and streamline the process of field site assessment to determine well vulnerability without the need for or in concert with conventional predictive modeling approaches. A 60-day pumping test was conducted on a newly installed public supply well located within the Regional Municipality of Waterloo adjacent to a perennial stream, Alder Creek, in order to gather hydrogeological and water quality information to assess well vulnerability. A network of groundwater monitoring wells was installed and instrumented at the site in the vicinity of the supply well, which included multilevel wells at several locations and drive point piezometers in the stream bed. Additional instruments were also placed within Alder Creek itself to measure surface water characteristics. A multitude of parameters were measured during the course of the test, including hydraulic head, temperature, general chemistry, metals, stable water isotopes, electrical conductivity, turbidity, and climatic data from drive points in Alder Creek, the pumping well, surrounding monitoring and multi-level wells, along with Alder Creek itself. It is rare to have such an extensive data. Stratigraphic information from drill records indicated the subsurface was dominated by glacial sands and gravels and identified an isolated lower permeability unit of silty clay above the depth of the supply well screen separating a shallow and deeper groundwater system. The hydraulic data collected during the pumping test were processed through standard pump test analysis methods to determine hydrogeologic parameters and understand the subsurface behaviour. The analysis indicated the subsurface responded as an unconfined system suggesting that the lower permeability unit did not significantly restrict the hydraulic connection between the shallow and deep systems. In this instance, the early, intermediate, and late time drawdown response indicative of the unconfined sand aquifer required six days of pumping to become apparent, providing evidence of the value for extended time pumping tests. Both the data from the stratigraphic mapping and the aquifer test analysis indicated the potential for a high degree of vulnerability of the supply well to surface sources of contamination. The groundwater water level data illustrated a fairly rapid response to the influx of recharge following significant precipitation events throughout the entire monitored subsurface region, again suggesting a high degree of hydraulic connection. Mapping of the drawdown cone resulting from the long term pumping from the supply well based on regional hydraulic head data illustrate that Alder Creek was situated within the capture zone of the well and that the influence of pumping passed beneath the creek and was clearly observable on the side opposite to where the pumping well is situated. These combined observations based on the hydraulic head data provide more evidence of a high degree of vulnerability of the supply well. Alder Creek and shallow groundwater beneath the streambed did not respond to the pumping process and this may be due to a low permeability bed under the stream or perched conditions. A strong downward gradient was measured across the streambed that indicates downward flow below the creek; however, additional information is required to quantify the groundwater-surface water interaction in the stream. Water quality and temperature data were collected for their potential to act as tracers of groundwater flow and groundwater-surface water interaction. Based on the relatively low average concentrations of hardness and calcium in the shallow system, they were identified as shallow tracers that decreased concentration in the pumping well during the pumping test below those levels in the intermediate and deeper groundwater systems. Higher concentrations of iron and sulphate were attributed to deeper groundwater contributions as a result of aquifer materials weathering in the subsurface. These data indicate that both shallow and deep groundwaters were captured by the pumping well. Temperature was an excellent indicator of precipitation influxes, which could be observed as pulses of higher temperature water in the wells after a given time lag. Variations in groundwater temperature distributions resulting from transient groundwater flow could be correlated to geologic heterogeneity. At the site, a silt layer in the subsurface caused a difference in temperature, where multi-level ports above the silt layer were considerably warmer than the ports screened below the silt layer. Water temperature from the pumping well became colder during the test, likely a result of deep groundwater being drawn up to the well screen. Additionally, pumping caused temperature increases in the shallower multilevel ports indicating that warmed water was also been drawn downward as a result of pumping. This deep and shallow groundwater movement matches the geochemical data analysis. The multilevel well between Alder Creek and TW2-13 showed the largest degree of change in groundwater temperatures, with shallower ports becoming warmers throughout the test, which might be a result of some surface water infiltration from the creek. The 50-day time of travel distance, a common method to assess well vulnerability, was determined for the groundwater flow system; Alder Creek is contained well within this estimated distance, once again increasing the vulnerability at the site. Several different vulnerability index calculations were performed, with a mixture of results ranging from moderately to extremely vulnerable. It is evident that there is room for improvement when it comes to establishing the vulnerability of an aquifer, where there is a specific need for indexing methods which focus on well vulnerability. Correlation coefficient and covariance calculations were applied to compare the different continuous and discrete data parameters available. The statistical analyses found correlation coefficients effective in determining the surface water level and turbidity correlation, pump well water level and temperature correlation, and the inverse relationship between conductivity and turbidity for the data sets available. Once again, sodium, chloride, anions and cations, and electrical conductivity were correlated to one another. Calcium, manganese, and hardness also correlated, indicating the mineral signature of the subsurface. Manganese and iron concentrations correlated positively to each other. Correlation coefficients are helpful in indicating groundwater flow direction and water sources based on quality parameter connections, where shallow or deep groundwater systems can be attributed to having certain qualities allowing for trend analyses to indicate groundwater movement. Statistically, surface water temperature can also act as a surrogate for air temperature, however there was no data available that would act as a reasonable surrogate for precipitation data. Given its usefulness, precipitation information should be gathered during longer duration pumping tests where the groundwater system is potentially connected to the surface. These statistical analyses are extremely easy to perform on existing or newly collected data sets, allowing for quick connections at the site to be identified. The statistical analysis can provide useful additional understanding of geochemistry associated with shallow or deeper groundwaters, assist in interpreting the movement of water in the subsurface and assess any response to surface changes. Overall, lengthy data sets allow for the myriad of conclusions to be made regarding long term water quality changes and impacts of seasonality, precipitation events, and shallow and deep groundwater mixing on the vulnerability of a public supply well. In the event of a short test being run, the depth of information gathered would not have been possible. Long term monitoring, coupled with quantifiable changes and correlations are paramount in addressing well vulnerability to surface water sources. Although certain geochemical parameters are bound to be site specific, monitoring turbidity, and electrical conductivity are valuable starting points; however detailed water level, water chemistry, and temperature data, from drive points and multi-level wells, are most important in estimating groundwater-surface water interaction and well vulnerability.