Author

Date of Award

8-17-2007

Document Type

Thesis

Abstract

The three-dimensional (3D) high-resolution digitized snow microstructure (pixel size 6 micron) was obtained by X-ray microtomography. The experimental result was verified by measuring the density of the snow sample. Statistical characteristics (porosity, local porosity, two-point correlation function) were extracted from cross-sectional images. The one-level-cut Gaussian random field model was used to stochastically reconstruct snow microstructure from X-ray microtomography images. Efficient computer programs were developed in MATLAB for the whole stochastic reconstruction procedure, including the numerical inversion of the correlation function and the generation of 3D large-scale Gaussian random fields by 3D inverse fast Fourier transform. The quality of the reconstruction was assessed by comparing the two-point correlation function and cross-sectional images.

Handle

http://hdl.handle.net/11122/5854

Share

COinS