Practical Implementation Details of Multiple Point Statistical Simulation

• Author / Creator

  Jiang, Chenyu

• Best practice and recommendations for multiple point statistics (MPS) simulation are presented. Three main contributions are: (1) assessing the stationarity of training images (TI) and determining the characteristics of TI's that result in realizations that reproduce features found in the TI; (2) determining optimal input parameters of MPS simulation; (3) summarizing categorical merging rules when implementing MPS simulation in a hierarchical MPS methodology. Specifically, the first contribution is determining what type of TI is suitable for generating MPS realizations that reproduce TI features. The quality of a TI is considered here by the quality of TI features reproduced in the MPS realizations. Stationarity is assessed through dividing a TI into zones and comparing the distribution of categories, oriental features and the Euclidean distance matrix. Ten 2D TIs and six 3D TIs are assessed to determine the optimal input parameters to use in SNESIM. The optimal settings found depend on the dimensionality of the TI and include: using a 40-70-point template in 2D cases; 50 (or more) point templates with 3D TIs; square template shapes; and use of 4 or more multiple-grids. Because the quality of the realizations generated with SNESIM are found to depend on the stationarity of the TI, cut-offs for determining the stationarity level of a TI is provided and is based on its statistical assessment and predicts expected realization quality. Finally, if the TI is too complex to provide decent feature reproduction because of the number of categories, a hierarchical methodology is recommended. The TI is re-coded to a TI with fewer categories, and is simulated in multiple steps. Each simulation step is based on the results of the previous steps. Rules for combining categories are discussed: categories that share a little contact area, or are far from each other should be lumped, to keep the isolated category separated, and lumping two categories that are completely connected. With the popularity of MPS simulation, it is important to ensure a proper implementation of the methodology. The TIs analyzed are diverse and allow for the generalization of the findings in this thesis. Ten 2D TIs and six 3D TIs are assessed to determine the optimal input parameters to use in SNESIM. The optimal settings found depend on the dimensionality of the TI and include: using a 40-70-point template in 2D cases; 50 (or more) point templates for 3D non-channel-type TIs; square template shapes; and use of 4 or more multiple-grids. Because the quality of the realizations generated with SNESIM is found to depend on the stationarity of the TI, the expected realization quality can be predicted by cut-offs that determines the stationarity level of a TI. Finally, if the TI is too complex to provide decent feature reproduction because of the number of categories, a hierarchical methodology is recommended. The TI is re-coded to a TI with fewer categories, and is simulated in multiple steps. Each simulation step is based on the results of the previous steps. Rules for combining categories are discussed: categories that share a little contact area, or are far from each other should be lumped or not to be simulated together, to keep the isolated category separated, and to lump two categories that are completely connected. With the popularity of MPS simulation, it is important to ensure a proper implementation of the methodology. The TIs analyzed are diverse and allow for the generalization of the findings in this thesis.

• Subjects / Keywords

  • Multiple Point Statistical Simulation
  • Training Image

• Graduation date

  Fall 2016

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/R3Z60C71S

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.