Framework for multiple hypothesis testing improves the use of legacy data in structural geological modeling

Geological models, as structural representations of the subsurface, are increasingly used for regional scale geological analyses and research studies. In this context, it is often essential to use geological legacy data, for example in the form of printed well logs, seismic sections, or maps and interpreted cross-sections from previous reports. A problem when using this type of data is that standard modeling methods and workflows are optimized towards applications in hydrocarbon and mineral exploration where data are usually newly acquired and of a high quality. Although recent developments address the modeling side for regional models with novel concepts and ideas, the possibility to change the workflow on a conceptual level has, to date, not been addressed.

We examine here how we can use legacy data more efficiently and sustainably, in a model construction workflow that leaves the typical sequential path of model development. In the common approach, a single best-fit model is continuously updated or refined when additional data become available. We test here the application of a parallel type of model construction where multiple models can be generated on the basis of different input data sets. Geological data and models are strictly separated, and this allows us to (a) use geological models to test quickly the spatial consistency of different geological data sets, and (b) to allow for an approach where we finally obtain multiple geological models as different hypotheses about the subsurface structural setting. Both aspects are especially important for the application of legacy data, as the data quality is always difficult to assess.

The concept is applied to a geological model project of the Perth Basin, Australia, where we show how it enables us to quickly revise and update the (previously constructed) model with additional data (e.g. newly available digitized legacy data), to evaluate the spatial consistency between different legacy data sets and interpretations, and to test different hypotheses. In our point of view, this is an important aspect towards a sustainable approach for geological modeling as it allows a very flexible and transparent use of different data sets for model construction – and therefore a more sustainable use of legacy data itself in the increasing use of subsurface representations using 3D geological models.