Learning from models: Insights into the behavior of water in unsaturated fractured-porous media

Thursday, 18 December 2014: 8:00 AM
Andrew M Ireson, University of Saskatchewan, Saskatoon, SK, Canada and Adrian P Butler, Imperial College London, London, SW7, United Kingdom
The way that water is stored within and transmitted through structured porous media can be challenging to infer from available observation techniques, in particular at field scales. Fractured porous media are an example of such structured media, comprising a fracture domain and a matrix domain. Flow processes depend on how water is both stored and transmitted within each domain, and how water exchanges between the domains. In unsaturated conditions the problem is confounded as hydraulic properties (e.g. hydraulic conductivity) in each domain become dynamic, as a function of saturation. One domain may not contribute significantly to the flow, or may not provide significant amounts of storage, depending of the properties of both domains. The two domains may or may not respond in equilibrium with one another, depending on both the properties and on driving boundary conditions (i.e. infiltration intensity). Physically based models can in theory be used to falsify hypotheses about the role of the fractures versus the matrix in transmitting flow, within an inverse modelling framework. However, there are two problems: firstly, even the most detailed and complex physically based model will still be a simplification of reality, unable to resolve heterogeneities that exist; and secondly the observations available may not contain sufficient information to constrain the model. In this study we explore how models and observations have been applied to understand the flow processes in the unsaturated Chalk in field sites in south east England. We look at what can and cannot be inferred from various observations, including the water table response, soil moisture and matric potential, and the power of these observations to constrain models. We look at how models provide a framework through which to interpret the observations, and provide insights into the dominant controls on behavior. In the case of the unsaturated Chalk, we have demonstrated that groundwater recharge is transmitted through both the fractures and the matrix. Matrix saturated hydraulic conductivity is found to be the dominant parametric control on partitioning of flow. The characteristics of individual rainfall events, combined with near surface soil moisture content, are shown to be good predictors of rapid, preferential recharge.