| Numerical reservoir models are constructed by cell-based techniques or by stochastically placed geometric objects. Description of spatial structures needs three-point or even higher order of statistics, but traditional geostatistical tools are limited to reproduce one and two-point statistics. Stochastic surface-based modeling allows for improved integration of geological information in deepwater clastic turbidite reservoir models. Surface-based methods model stratigraphic layers to fill available accommodation space. Stacking patterns and hierarchies of trends related to sedimentary processes are reproduced by construction.; Deepwater surface-based methods are not mature. This thesis documents new developments, such as automatic surface-picks identification, deterministic and stochastic surface placement, global and local base levels modeling, improved hierarchical trend modeling and, global and depositional erosion events modeling, which result in more practical workflows and greater integration of deepwater geologic information. The result is improved numerical reservoir models of deepwater systems and, therefore, an expectation of improved reservoir performance forecasting and management. |
