Kibbey, Tohren C.G.Yan, Shang2019-11-262019-11-262019https://hdl.handle.net/11244/322821The air-water interface in porous media has been studied for decades, and is important for understanding the distribution and movement of water, solutes and suspended colloids and nanoparticles in the unsaturated zone. The spatial configuration of individual grain surfaces plays a central role in defining the shapes of capillary-held water films in porous media, as well as influencing how flow and transport occur in unsaturated media. An accurate description of grain surface is essential for the simulation of the configuration of air-water interface. The objectives of this work were to explore algorithms related to reconstruction of grain surfaces from scanning electron microscopy (SEM) images, and then to simulate configuration and flow in unsaturated media at different scales. To obtain grain surface configuration, a new hybrid method was developed to reconstruct three-dimensional (3D) grain surfaces from two dimensional (2D) images taken by scanning electron microscopy (SEM). The hybrid method combines stereoscopic reconstruction with shape-from-shading calculations, and is able to capture detail from complex natural surfaces. shadows are universal in SEM images, and can complicate reconstruction. A machine-learning method based on boosted decision trees was used to identify shadows in SEM images based on a training set of shadows in photographic images. The influence of shadows on the hybrid reconstruction method was analyzed. Previous studies examining the configuration of air-water interfaces have been based on calculations for surfaces whose elevations can be defined by z=f(x,y). This made existing methods unsuitable for calculations in more complex geometries, such as around the outsides of grains. A numerical simulation method was developed to calculate air-water interface in three-dimensional space around the outsides of grains. The method was then extended to allow simulation of flow during evaporation. Finally, work was conducted to explore the flow and transport of nanoparticles in a cluster of sand grains experiencing evaporation. Experiments involved filling a cluster of sand grains with water containing fluorescein sodium and sulfate-modified polystyrene nanospheres, and using confocal microscopy to image water flow between sand grains and deposition of fluorescent nanoparticles during evaporation. A model was developed to explain the flow behavior observed.3D reconstructionshadow detectionair-water interface around grainsnanoparticle deposition at pore scale