James Irving

An important part of properly managing and protecting our valuable groundwater resources is the development of reliable numerical models of groundwater flow and contaminant transport. Such modeling requires the characterization of subsurface hydrological properties, and depending on the nature of the investigation, this characterization can be required at a variety of spatial scales. Over the past two decades, geophysical methods have gained increasing popularity in hydrology because of their ability to fill a significant gap between the resolving abilities of traditional characterization methods, namely drill coring and pumping / tracer tests. Whereas drill cores provide only sparse, one-dimensional profiles of subsurface geology, pumping and tracer tests yield average hydraulic properties over a relatively large region, and may not provide sufficiently detailed estimates of these properties for the particular modeling task at hand.

Many of the geophysical techniques popular in hydrological studies have been around for a long time. However, there are still a number of areas that require much research when it comes to incorporating geophysical data into groundwater investigations. One of these areas is the improved numerical modeling and inversion of geophysical data, such that the most accurate images of subsurface geophysical properties can be obtained for input into the hydrological models. In my M.Sc. and Ph.D. studies, my research focused within this domain and largely involved the ground-penetrating radar (GPR) method. For my M.Sc. degree at the University of British Columbia, I examined the problem of frequency-dependent attenuation in reflection GPR data, and developed a means of estimating and correcting for its effects using time-frequency analysis and inverse-Q filtering, respectively. The result was a substantial improvement in the resolution of reflection GPR images, and thus an improvement in our ability to resolve subsurface geological structure. For my Ph.D. degree at Stanford University, I worked on improving the numerical modeling and inversion of crosshole GPR data. One part of this research consisted of developing computer codes to accurately and efficiently simulate antenna transmission and reception for this technique. Using these codes, I then examined the effect of the antennas on crosshole GPR travel-time tomography, and proposed that a standard inversion assumption, namely that first-arriving energy always travels directly between the antenna centers, may often be invalid at high transmitter-receiver angles. This is possibly the reason why significant tomographic artifacts commonly appear when high-angle travel times are incorporated into crosshole GPR inversions. By modifying the inversion procedure to account for these findings, I was able to remove the artifacts and obtain significantly improved tomograms of subsurface EM-wave velocity, a parameter that is strongly correlated with soil water content. At the beginning of my doctorate, I also wrote some basic finite-difference time-domain (FDTD) codes in MATLAB for modeling borehole and surface-based GPR. The paper and corresponding codes are available on my publications page.

In my current post-doctoral position at the University of Lausanne, I am working on a variety of projects, all of which are geared towards the more effective use of geophysical data in hydrological investigations. One of these projects involves using conditional stochastic simulation techniques to integrate geophysical and hydrological data, the goal being the generation of multiple realizations of subsurface hydrological parameters that all fit (to within a specified degree of uncertainty) the available information. Another project that I am working on involves the investigation of whether borehole Stoneley-wave logging can be used to obtain ballpark estimates of hydraulic conductivity in unconsolidated near-surface sediments. Finally, I am looking at whether reflection GPR data can be used to invert for the statistical parameters describing the spatial variability of subsurface water content. This information could be used to better constrain unsaturated flow and transport models.