TY  - BOOK
AU  - Kim,Ahram
ED  - Universiti Teknologi Brunei
TI  - Compositional kriging for geological heterogeneity : : enhanced spatial orediction framework 
PY  - 2025///
CY  - Bandar Seri Begawan
PB  - Universiti Teknologi Brunei
KW  - Unversiti Teknologi Brunei
KW  - Thesis
KW  - Universiti Teknologi Brunei
KW  - Final Year Report
KW  - Dissertation, Academic
KW  - Thesis writing
KW  - Dissertation Universiti Teknologi Brunei
KW  - Engineering
KW  - Petroleum and Chemical
N1  - Submitted in fulfilment of the requirements for the degree of Doctor in Philosophy; ABSRACT
This thesis presents an advanced spatial prediction framework based on compositional kriging for the analvsis of datasets that exhibit both compositional constraints and spatial heterogeneity. By integrating isometric log-ratio (il) transformations, the method preserves the geometry of the simplex space, ensuring statistically coherent spatial modelling of compositional data. The study applies this framework across four diverse case studies to demonstrate its adaptability and accuracy: demographic compositions in Texas, sediment facies on Long Island, coal seam thickness in Queensland, and grain size distributions from outcrops in Brunei. In the Long Island case, compositional kriging was compared with ordinary kriging and showed improved spatial interpolation accuracy, as validated by metrics such as Root Mean Square Error and Mean Squared Error. For the other datasets, compositional kriging produced accurate and reliable spatial predictions, with interpolated distribution shapes closelv matching the original data and effectivelv capturing spatial variability. The interpolated distribution shapes closely matched the original data in each case, underscoring the method's reliability. It makes a significant contribution to spatial statistics by integrating compositional data analysis with geostatistical modelling. The proposed approach provides a robust, accurate, and flexible tool for geoscientific prediction tasks, particularly where data are multimodal and spatially dependent. The results highlight compositional kriging's potential as a preferred method for future studies involving complex spatial datasets; Dissertation (Doctor of Philosophy) - Universiti Teknologi Brunei (2025); Includes references and appendix
ER  -