The Inverse Modeling and Geostatistics Project
Research People Papers Software Teaching

Effective high-resolution Geological Modeling

This project is a cross-disciplinary applied/theoretical project funded from September 2013 by The Danish National Research Foundation (Højteknologifonden) (DTU Budget DKK 2,390,400).


In applied geoscience, methods for combining geo-information and setting up a geological model is today typically sequential: Each type of available geo-information is treated separately, by experts in each individual type of geo-data. A typical scenario related to setting up models of groundwater reservoirs is the following: A geophysicist may use inverse theory to translate geophysical data into a (most often smooth) model of physical properties of the subsurface. This model is then used by a geological expert to suggest a geological model, which may then again be used by for example a hydrologist as the basis for setting up a flow model of the subsurface.

There are several - currently unsolved - problems related to this practice which can be summarized as follows:

1. The amount of geo-data, from which the geological model is built, is constantly increasing. Already today, the amounts of geo-data are so great that in practice not all available geo-data are used to setup geological models.
2. There is an ever increasing focus on resolving smaller scale features of the subsurface, as such small scale features can have dramatic effect when the geo-model is used to, for example, simulate the effect of pollution spread.
3. Decisions based on geological models may not be consistent with all available information used to setup the geological model. 4. Most often, one unique geo-model is estimated with no associated uncertainty assessment.

To overcome these challenges this project aims at formulating the problem of integration of geo-information in a statistical Bayesian framework, where all information is quantified by statistical models. Based on state of the art methods developed in inverse problem theory and geostatistics, we aim to develop new theory and applications that can be used to solve the challenges listed and, at the same time, be manageable in practice. This will enable efficient geological interpretation of possibly very large geophysical data sets, and allow inclusion of geologically realistic small scale variability. Combining these methodologies will increase reliability and usability of geo-models when used in water resource evaluation and risk analysis.

(Collaborators: I-GIS, Denmark; GEUS)