Mixing effects on apparent reaction rates and isotope fractionation during denitrification in a heterogeneous aquifer

Gradients in contaminant concentrations and isotopic compositions commonly are used to derive reaction parameters for natural attenuation in aquifers. Differences between field-scale (apparent) estimated reaction rates and isotopic fractionations and local-scale (intrinsic) effects are poorly understood for complex natural systems. For a heterogeneous alluvial fan aquifer, numerical models and field observations were used to study the effects of physical heterogeneity on reaction parameter estimates. Field measurements included major ions, age tracers, stable isotopes, and dissolved gases. Parameters were estimated for the O-2 reduction rate, denitrification rate, O-2 threshold for denitrification, and stable N isotope fractionation during denitrification. For multiple geostatistical realizations of the aquifer, inverse modeling was used to establish reactive transport simulations that were consistent with field observations and served as a basis for numerical experiments to compare sample-based estimates of apparent parameters with true (intrinsic) values. For this aquifer, non-Gaussian dispersion reduced the magnitudes of apparent reaction rates and isotope fractionations to a greater extent than Gaussian mixing alone. Apparent and true rate constants and fractionation parameters can differ by an order of magnitude or more, especially for samples subject to slow transport, long travel times, or rapid reactions. The effect of mixing on apparent N isotope fractionation potentially explains differences between previous laboratory and field estimates. Similarly, predicted effects on apparent O-2 threshold values for denitrification are consistent with previous reports of higher values in aquifers than in the laboratory. These results show that hydrogeological complexity substantially influences the interpretation and prediction of reactive transport.