Parameterization and Validation of Numerical Transport Models using Hydrogeophysical Estimation Approaches
- Author(s): Linde, Niklas;
- et al.
Numerical modeling of fluid flow and transport is often used to test hypotheses and to guide resource management. For example, numerical models are used to simulate contaminant plume transport over time and to design effective remediation plans, and to simulate aquifer depletion versus recharge over time and to design sustainable resource extraction programs. Effective hydrological hypothesis testing using mathematical modeling requires accurate and sufficient parameterization. It is well recognized that physical and chemical properties of aquifer systems (such as hydraulic conductivity and reactive iron oxide surface area) is great and can have multiple scales of spatial variability. As these parameters exert significant control on transport, natural attenuation, and active remediation processes, successful modeling requires sufficient information about the distribution of these properties. Given the difficulty of obtaining characterization data for model parameterization using conventional wellbore approaches, the predictive capability of these models for practically guiding water resource or contaminant management is limited. Geophysical methods hold potential for providing quantitative information about subsurface hydrogeological parameters or processes that can be used to constrain or validate transport models. However, to use integrated geophysical and hydrogeological information for such purposes, we must confront issues of scale, non-uniqueness, and uncertainty; we must have petrophysical relationships available that link the disparate measurements; we must have an understanding about the spatial distribution of measurement and parameter estimate uncertainties; and we must have estimation approaches that can be used for integration. In this presentation, we will first review examples of the estimation of hydrological-geochemical properties needed for model parameterization using geophysical data and stochastic approaches, such as Bayesian and Monte Carlo Mar