Assessing water quality impacts and cleanup effectiveness in streams dominated by episodic mercury discharges

Accurate pollutant mass budgets are needed for identifying contaminant sources and establishing cleanup goals. We monitored mercury discharges from an abandoned mine site in northern California with the objectives of: (1) estimating the mass loading of mercury from the site; (2) evaluating the factors that control the mercury discharges; (3) assessing the significance of peak flows in transporting contaminants; and (4) developing methods for measuring the effectiveness of cleanup efforts. We sampled water downstream from the mine site over a wide range of streamflows. Mercury concentrations varied over 2000-fold, from 485 to 1040000 ng/l, grossly exceeding the regulatory water quality objective of 12 ng/l at all times. Particulate mercury represented over 99.97% of the total mercury, and mercury concentrations were closely correlated to suspended sediment concentrations (r = 0.98). Thus, we can use suspended sediment concentrations as a proxy for mercury concentrations, and calculate a continuous record of mercury flux from continuous monitoring of streamflow (using a small flume) and turbidity (using an optical backscatter sensor). Mercury fluxes inferred in this way are consistent with fluxes estimated from field samples. In January and February of 1998, our small abandoned mine site released approximately 82 kg of mercury to downstream waters. Most of the mercury was released during brief intense rainstorms. For example, in one 200-min period we recorded 3.4 cm of rain, a 2.6-fold increase in streamflow (460-1120 l/s), and an 82-fold increase in mercury flux (1.2-99 g/min). Over 75% of the total mercury flux during this 2-month period occurred in less than 10% of the total time. In systems such as this one, where contaminant transport is highly episodic, sampling programs that miss the high-flow episodes may greatly underestimate the actual water quality threat. In addition, changes in pollutant fluxes or concentrations in receiving waters may not reflect changes in pollutant sources (such as an environmental cleanup) if the stochastic forcing (e.g. intense rainstorms) varies through time. We propose that water quality trends can be more accurately measured by changes in the relationship between contaminant flux and stochastic driving factors, as expressed by contaminant rating curves.