Evaluation of Process Oriented Water Quality Predictions for Pit Lakes

Reliable prediction of water quality is essential to meet the official targets set for effluent composition and usage of pit lakes, and to identify the appropriate remediation technology. For this purpose, a complex conceptual model was established and tested on two pit lakes in central Germany with highly different acid loads. To assess the reliability of the predictions, we compared monitoring data from the past 8–10 years with previous water quality predictions. In the case of Lake Schladitz, with low acid load and sufficient buffering capacity, a simple setup with few model elements appeared adequate, and no readjustment of model settings was necessary. For six (pseudo-) conservative ions, the total average deviation between measured and predicted values was − 9%, as opposed to − 2.7% during calibration. For the repeatedly conditioned Lake Bockwitz, a model parameterized with more elements and specific process parameters determined from field and laboratory investigations proved adequate for calculating the technical alkalinity demand. During calibration, the total average deviation between measured and modelled conservative ions was +0.2%; it reached +4.5% during the prediction evaluation period. With continued reacidification of the lake, the predicted concentrations of dissolved Fe and Al increasingly deviated from lake measurements. Deviations diminished after the solubility constants of schwertmannite and hydrobasaluminite were included in the thermodynamic database. The developed conceptual workflow offers a tool to improve water quality predictions for other lake settings.