CAPTURE ZONE ASSESSMENT FOR CONTAMINATED MUNICIPAL WELLS IN A CARBONATE AQUIFER

Delineation of contributing areas in carbonate aquifers is particularly challenging because of the effects of bedrock dissolution, which are difficult to assess but which can result in both rapid groundwater velocities and in well capture zones that deviate significantly from those estimated from porous medium numerical models. Such a challenge occurred at Walkerton, Ontario, in May 2000, following the deaths of seven people and the illnesses of 2300 others as a result of bacterial contamination of the municipal water supply. Subsequent epidemiological investigations indicated that the contamination of the water supply had occurred within hours or days at most after heavy rain.

A major hydrogeological investigation was conducted on behalf of the town of Walkerton in the summer of 2000. A numerical model of flow incorporating the results of this investigation (assuming an equivalent porous medium and using MODFLOW) indicated that the 30-day capture zones extended less than 300 m from each of the three municipal wells. However, comparison between the bacteria in the human cases and in cattle manure showed that some of the pathogenic bacteria came from a more distant source. The pathway for these latter cases was difficult to explain if the aquifer behaved as an equivalent porous medium.

Evidence came forward during the year-long public hearings of a judicial inquiry into the events at Walkerton that the carbonate aquifer was in fact karstified, implying that the 30-day capture zones could be larger than model simulations had indicated. Subsequent spring monitoring confirmed that the well capture zones were far more extensive than the MODFLOW model had shown. Convergent flow tracer testing to one well then showed that groundwater velocities were some 80 times faster than model simulation results. Calculations using the tracing results and downhole video observations indicated that these fast pathways were via channels with apertures of 5 - 20 mm.

These results demonstrate two crucial considerations for hydrogeological assessments of carbonate aquifers. First, it is wise to take a precautionary approach and assume that karstic channels are present. Second, measuring groundwater velocities in the field by means of tracer testing yields the most reliable data on groundwater velocities.