Stochastic assimilation of geophysical and hydrological data for the estimation of hydraulic conductivity

点击翻译按钮获取中文摘要

Geophysical methods offer much potential in hydrology because of their ability to image subsurface properties with a resolution and degree of spatial coverage unattainable with traditional hydrological measurements. This has resulted in a recent and rapid expansion of the new and interdisciplinary field of "hydrogeophysics". The overall goal of the proposed research program is to address a key issue in current hydrogeophysical research, which is how to best assimilate diverse sources of geophysical, hydrological, and geological data in order to estimate subsurface hydrological properties within a framework of uncertainty. Two student projects are proposed to begin work in this domain at the University of Guelph, both of which have the general goal of estimating the spatial configuration of hydraulic conductivity in heterogeneous alluvial aquifers, which is the dominant control on groundwater flow and contaminant transport. The first project will work at the local (~10m) scale, and will involve the stochastic assimilation of tracer-test concentration measurements and time-lapse electrical resistivity data using a Bayesian Markov-chain-Monte-Carlo methodology. Here, the geophysical and hydrological process models will be coupled to provide a link between the resistivity measurements and hydraulic conductivity. The second project will work at a larger (~100m) scale, and will involve the use of multiple geophysical methods and cluster analysis techniques to first zone the subsurface into distinct units that are assumed to have constant hydrological properties. Then, using larger-scale hydrological data, the properties of the zones and their uncertainties will be estimated. Work on both of these projects will begin with numerical "synthetic" studies, and will be fully tested in 2-D before moving to 3-D applications. All results obtained will be evaluated in terms of their ability to reduce hydrological prediction uncertainty. Over the longer term, this research will lead to additional student projects involving testing of the proposed methodologies in the field as well as detailed investigation of issues expected to be crucial for meaningful uncertainty estimates, such as the reliable assessment of data and model structural errors.