EVALUATING AQUEOUS GEOCHEMICAL MODELS AND CODES

Scientific models and computer codes are useful tools in the elucidation of environmental processes and in the development and execution of regulatory requirements. However, inappropriate applications can lead to misunderstandings of science and poor policy decisions. “Models” are simplifications that represent our understanding of physical reality; they are not the same as codes which are rules to convert one type of information into another; models cannot be “validated” or “verified” in any general sense of these words. A model result can disagree with independent observations and be scientifically “correct” and, similarly, a model can agree with observations and be “incorrect.” Models can be used to examine possible future scenarios and can aid in prediction, however, scientific predictions are “logical” not “temporal,” i.e. we cannot predict the future, only the consequences of making good assumptions combined with well-established scientific principles. We must accept that policy decisions have to be made without all the answers that science might be able to provide or even to the degree of certainty that we would like. We should continue to test and evaluate models and codes and clarify their limitations without claiming that we can extrapolate their results reliably in scenarios that can never be corroborated. Aqueous geochemical models use speciation calculations to determine the free-ion concentrations and activities which are useful in estimating the mineral saturation state, redox conditions, biotoxicity, bioavailability, and sorption properties of a water sample. It is also essential to reactive-transport modeling. In mixed electrolyte solutions such as natural water it is difficult to confirm the accuracy of the speciation calculation. Several analytical techniques such as ion-selective potentiometry, voltammetry, UV-visible spectrophotometry, high-performance liquid chromatography-inductively coupled plasma-mass spectrometry, synchrotron radiation, and ion-exchange separation can often discern types of speciation. Comparisons of these analytical speciation measurements with computed speciation values have had a range of success from poor to excellent. Examples include activity measurements of ions, measurement of inorganic complexes, determinations of redox species, and determinations of organic-metal complexes. Waters known or likely to be in solubility equilibrium with well-defined minerals can be compared successfully to their calculated saturation state. Speciation calculations can also be compared to laboratory measurements of activity coefficients, activities, and solubilities. Mineral saturation indices often reflect supersaturation for minerals of moderate to low solubility such as calcite, ferrihydrite, goethite, gibbsite, barite, and fluorite. Supersaturation with respect to ferrihydrite and goethite can reflect poorly characterized solutes (insufficient filtration and ill-defined redox state). These and other examples do not indicate that the thermodynamic data are in error. Calcite seems to exhibit actual supersaturation for some groundwaters and surface waters which may be caused by kinetic processes. We cannot always assume that aqueous equilibrium speciation calculations are correct. More comparisons between computational and analytical techniques are necessary. Standard test cases should be designed to demonstrate the reliability of models and codes at several levels of sophistication from speciation to batch mass transfer to reactive transport. Multi-code comparisons would be helpful in clarifying the strengths and limitations of speciation computations.