Paper No. 13
Presentation Time: 11:15 AM


MACCORMACK, Kelsey E., Alberta Geological Survey, Alberta Energy Regulator, 402 Twin Attria Building, 4999 - 98 Avenue, Edmonton, AB T6B 2X3, Canada,

In the past, society has taken the information portrayed in geologic models at face value and assumed that attribute accuracy is equivalent to the digital accuracy recorded by the computer. This often led to overconfidence in the model results. Society is now more accepting that models are simply versions of reality that contain errors and uncertainty. This acceptance has led to the important recognition that error and uncertainty are not bad, and should instead be quantified and understood rather than ignored. Consquently, scientists are now able to explore multi-dimensional models in a new light and move away from the concept that attributes need to be precisely right across the entire model domain. In other words, being vaguely right is more preferable than being precisely wrong.

The time has passed for modellers to aggregate and input as much data as possible into a black-box (the computer program), and subsequently assess the model output on screen. Producing geological models has become much more sophisticated as we have become aware of the implications of incorporating low quality data (garbage in = garbage out), assuming that the modelling program is doing exactly what we expect, and that all our models can be generated using a conveyer belt approach. In short, modellers have become acutely aware of the effects of assumptions on model predictions, which can profoundly impact decisions based on the output results.

Today's models are constructed on sophisticated platforms, capable of integrating a variety of data from multiple sources to produce multi-scale, interdisciplinary models with built-in feedback mechanisms, allowing the model to adapt and evolve as additional knowledge is incorporated. Increased requirements for transparent and fully-documented modelling procedures has necessitated the development of holistic grid metadata systems to capture and store model input parameters and output statistics. It can be argued that current models are producing substantially more data than ever before, and ensuring that this information is tied back to the geology and communicated effectively has become increasingly more important.

This presentation will discuss past successes and failures of 3D geologic models, the lessons learned, and recent advances leading geological models into a new multi-dimensional era.