Stepwise conditional transformation for multivariate geostatistical simulation

Numerical models are most effective when they account for all sources and types of data. Real data exhibit complex multivariate features such as non-linear, constraint, and heteroscedastic relations. Current geostatistical simulation methods that allow for modeling of multiple variables rely on simple statistical models that are sometimes inappropriate or unable to account for realistic complexity in the multivariate relations.; This dissertation develops the stepwise conditional transformation technique for use as a pre- and post-processing tool for multivariate Gaussian simulation. The back transformation enforces reproduction of the original complex multivariate features. The methodology and underlying assumptions are explained. Several petroleum and mining examples are used to show features of the transformation and implementation details.; Application to the Red Dog zinc deposit showed an increase in profit from the simulation approach relative to the common practice of kriging. A comparative study of multivariate simulation using stepwise conditionally transformed variables against conventional simulation approaches is also shown. Further, the stepwise transform can also be used to account for multivariate features resulting from trend modeling.