Abstract
The accuracy of predictive water quality modelling of pit lakes depends upon the availability and quality of input data. Quality control of input datasets is therefore crucial when undertaking geochemical modelling to ensure there is confidence in the predictions that inform operational and/or closure related management decisions. The accurate prediction of pH depends on an accurate and realistic understanding of the behaviour of acid and alkalinity generating materials: over or under estimation of either may result in inaccurate or even misleading model predictions.
For this study, data with unrealistically high dissolved alkalinity (>10,000 mg/kg as CaCO3) was identified in dilute static leachate samples used to represent surface water runoff in a PHREEQC pit lake model. The measured alkalinity was inconsistent with measured and calculated parameters (significant charge imbalance and highly oversaturated CO2). The use of this data (unaltered) as an input into water quality models resulted in unrealistically alkaline conditions, driven by excessive CO2 degassing of overestimated dissolved inorganic carbon (DIC). The origin of this issue was identified as a discrepancy in the method used by the laboratory - the alkalinity titration for some historic samples had been undertaken on unfiltered liquors and therefore accounted for additional ANC from the dissolution of mineral solids with carbonate minerals. This issue was resolved in future analyses, and all samples are now filtered, however historic data remained unsuitable for inclusion in future assessments.
To recover alkalinity data for the affected historic samples, PHREEQC was used to remodel dissolved alkalinity in the leachate samples where suitable data was available (pH, major cations and anions). The modelled alkalinity was verified with a control dataset finding excellent correlation and allowing development of robust QA/QC criteria. The redefined alkalinity data was used as a model input resulting in more defensible pH predictions of the modelled pit lake.