| Multiple point statistics has surfaced as the answer to many of the limitations of traditional variogram based facies modeling. Research has explored implementation difficulties facing a successful application of multiple point statistics; however, there are no readily available, geologically realistic training images for mineral deposits. Various sources of training images are considered including: using data sets as training images; generating training images for vein type deposits from pseudo genetic mimicking of the geological processes that formed the deposit; and, using currently available geostatistical techniques to generate the desired nonlinear features observed in weathered deposits.; The selection of a training image for use in multiple point statistics can be difficult. Often, there is little objective evidence to select a specific training image. A methodology to rank training images based on a multiple point statistical comparison to exploration data is presented as objective support for selecting an appropriate training image. |
