Water and solute transport through the unsaturated zone have a major
quantitative impact on the high-quality groundwater resources available for
exploitation. The protection of these valuable resources requires an
accurate understanding of the unsaturated processes in order to produce
reliable decision tools. This Ph.D. thesis was initiated with the objective of
determining unsaturated hydraulic parameters at a scale more appropriate
than traditional laboratory derived parameterization. Unsaturated flow and
transport processes were therefore studied at a field site in Denmark. The
field site was characterized by ~30 m thick unsaturated zone consisting
mainly of sands of varying coarseness. Following an instrumentation of 16
boreholes two geophysical methods (cross-borehole ground penetrating
radar and electrical resistivity tomography) were applied during natural
precipitation and forced infiltration. The methods provided estimates of soil
moisture content and electrical resistivity variations among 12 m deep
boreholes located 5 – 7 m apart.
The moisture content change following natural precipitation was observed to
be practically negligible, providing minimal information to constrain the
dynamic properties of the subsurface. On the other hand, volumetric
moisture content variations of up to 5% were observed during a 20-day
long forced infiltration experiment. The cross-borehole electrical resistance
tomography and ground penetrating radar data collected during this
experiment were subsequently combined to produce estimates of tracer
concentration profiles and images suitable for moment analysis. In spite of
the extensive use of the cross-borehole geophysical methods, there are still
some limitations and uncertainties associated with tomographic images
resulting from the methods. To avoid these effects, a framework to estimate
unsaturated hydraulic parameters using multiple data types was developed.
In this methodology, the collected geophysical data was used directly
without producing tomographic images. Unfortunately, the data did not
provide sufficient information to constrain all the parameters of the
parametric function describing the unsaturated hydraulic properties. Only
the saturated hydraulic conductivity values of the top 7 m were partially
constrained.
In order to improve the tomographic estimates obtained through inversion,
two additional methodologies were investigated. (1) A method was
developed that incorporated information regarding correlated data errors in
the inversion. Unwanted artefacts were in this way dampened considerably
without reducing information of the subsurface. (2) A stochastic inversion
technique was evaluated that maintained the variability of the subsurface
physical properties. This method also provided estimates of the subsurface
correlation structures which may serve as input for stochastic simulation
techniques.