All Models are wrong, but some are useful. George Box
Before model results are applied with confidence, an measurement of the range of possibilities and closeness of fit to the observations must be made. This can be performed at two levels. The first level measures the difference between the modelled and observed depths, interval thickness, paleobathymetry, porosity, etc. Comparison plots are made to illustrate the locations of the largest deviations. The second level performs a series of model runs evaluating the model results over a range of values for a single parameter. Each model changes the value of the parameter in question and converges on a best solution for that parameter set.
Once the program is within the best-fit model space defined by the observations, a series of model runs are performed. At the end of each run the probability of encountering one or more lithofacies types or physical condition (such as porosity greater than 15%) is measured by overlying a grid with user-defined thickness. In this case, the model was looking for sand prone lithofacies (medium sand, coarse grained sand, interbedded sand and silt) within a 5 meter thick interval. At the end of 100 or more runs, the table is plotted where each grid node value represents the weighted probability of encountering the state. In this case, we were evaluating the sensitivity of the distribution of sand to variations in the controls on gravity flow sedimentation. The orange regions indicate areas where sands occurred in 98% of the cases. In other words, deposition of sand in this location was insensitive to subtle variations in subsidence history, sediment supply, and the factors controlling gravity-flow sedimentation.
Weight Probability = (encounters/number of model runs) / Closeness of Fit Measure
There are several factors which influence the result. One factor is the shape of the sand bodies. Sands deposited in barrier island settings tend to be narrower and located in isolated patches. Basin-floor fan sands tend to be widely distributed but thin. Another factor is the final depth of the location. If the final location of a sand is moving up or down in the section due to differences in subsidence rates or thickness of the substrata, the probability will be distributed over a wider section of the column. A third factor is the sensitivity to variations in the factors controlling the processes. If the sand location doesn't change over the likely range of a variable, then the location of the sand body is insensitive to the factor.
These cross-sections are helpful in evaluating the probability of encountering the necessary structure, reservoir, and seal relationships required to trap hydrocarbons.
15 of 45
Comments for the Webmaster.