Download files
Abstract
Stochastic simulation allows the creation of multiple realistic realizations of an unknown geological domain, with conditioning to existing observations and reproduction of a given spatial continuity. Some of the most successful methods in this field are multiple-point geostatistics, which define spatial continuity based on the concept of training image (TI). One such method, image quilting (IQ) initially proposed in computer graphics, is introduced in this thesis and adapted for hydrogeological applications. This thesis first modifies the original method to accommodate conditional data, stochastic modelling, parameter sensitivity analysis and 3-D simulation. The performance of IQ is tested for a variety of hydrogeological test cases. The results, when compared with previous multiple point statistics (MPS) methods, indicate significant improvement in the CPU time and memory requirements, and do not come at the cost of degraded connectivity or patterns reproduction.
The application of IQ in a multivariate format is presented to integrate multiple scales of hydraulic conductivity measurements, which is the first implementation of a multivariate MPS algorithm in the field of patch-based techniques. It has been widely demonstrated that the hydraulic conductivity of an aquifer increases with a larger portion of the aquifer tested. This poses a challenge when different hydraulic conductivity measurements coexist in a field study and have to be integrated simultaneously. This study addresses this issue in the context of MPS simulation.
The second part of the thesis utilizes the spatial array of automated cave drip monitoring in Golgotha Cave, SW Australia, to understand infiltration water movement via the relationships between infiltration, stalactite morphology and groundwater recharge. Terrestrial LiDAR measurements are used to analyze stalactite morphology to characterize possible flow locations in the cave. The study then identifies the stalactites feeding the drip loggers and classifies each as matrix, fracture or combined-flow based on stalactite morphology. These morphology-based classifications are compared with flow characteristics from the drip logger time series. The total estimated discharge is compared with infiltration estimates to better understand flow from the surface to the cave ceilings of the studied areas. Finally, a methodology of using LiDAR imagery to produce TIs for MPS is introduced.