Evaluating the Reliability of Groundwater Samples Taken from Wells Potentially Impacted by Residual Drilling Fluids Using Multiple Reduction-Oxidation Indicators

Historic activities at Los Alamos National Lab located in Northern New Mexico have led to the release of potentially harmful contaminants to the groundwater. Regional and intermediate groundwater wells have been installed in and around laboratory property to characterize the hydrogeologic setting and identify potential contaminant pathways. In some of these wells the quality of groundwater data has been compromised due to incomplete removal of drilling fluids during well development (2005b, ESD-ERSP). Native groundwater is characterized as oxic: low but measurable nitrate and sulfate; negligible dissolved iron and manganese. However, biodegradation of residual organic drilling fluids by in situ microbes initiates a sequence of geochemical events that affects the concentrations of the above stated redox indicators: NO₃^- (aq) is reduced to N₂(g), MnO₂(s) is reduced to Mn ²⁺ (aq), FeO(OH)(s) is reduced to Fe ²⁺ (aq), and finally, SO₄^2- (aq) is reduced to S^2-(aq). The work described in this poster builds on initial efforts to use geochemical indicators to identify water samples impacted by drilling-activities documented in the Well Screen Analysis Report (LANL 2005b, ESD-ERSP). Any given groundwater sample should have concentrations of the redox indicator species that reflect the stage of reduction. For example if a sample exhibit elevated iron concentrations it should also have elevated manganese concentrations, and lowered nitrate concentrations relative to background. An inconsistent result, using the previous example, could have elevated concentrations of all three. A dataset of 348 sample events was tested for consistency amongst redox indicators. The test was run three times using three sets of threshold values. The initial run was run using background values from the Background Investigations Report (2005a, ESD-ERSP). Using these values 28.4% of sampling events were inconsistent. A statistical reevaluation the dataset from the Background Investigations Report was used to establish thresholds. The second run of the redox test using revised thresholds resulted in 32.8% inconsistencies. A final redox test was applied using thresholds adjusted using a conceptual model, which resulted in 23.6% inconsistencies. The inconsistencies can be diagnosed as: conceptual model inadequacy (45.1%), analytical error (29.3%), unknown (14.6%), combination of model and trend (4.9%), trend (3.7%), groundwater contamination (2.4%). Although not completed in time for this presentation, geochemical speciation modeling will be used to further refine our conceptual model and thresholds.