Characterising uncertainty in geological maps using variability and geodiversity analysis.

Lachlan Grose and Laurent Ailleres and Gautier Laurent and Peter Betts. ( 2014 )
in: Proc. 34th Gocad Meeting, Nancy

Abstract

Geological uncertainty can be identified during all stages of the geological workflow, from data collection to interpretation. The creation of three dimensional (3D) models often involves interpolating a series of two dimensional (2D) datasets (e.g. maps and cross sections). Uncertainties in the 2D datasets will propagate into the 3D model. This study investigates the uncertainties associated with lithological contacts in geological maps and will compare these results to other studies investigating the uncertainty associated with orientation datasets. Geological maps are created from a combination of observation and human interpolation between outcrop observations. This study uses geological variability as a proxy for geological uncertainty and we present methods for identifying and analyzing geological variability surrounding geological structures. We use geodiversity to characterize map differences by a number of geometrical measurements assessing geologically significant map properties. Geodiversity analysis is performed using Self-Organising Maps (SOMs) which allow for n-dimensional datasets to be reduced to 2 dimensions for visualization and analysis. Using k-means clustering on SOMs, the varying maps can be categorised into groups with similar geological characteristics. We use 40 maps from a field course in Broken Hill, Australia, as a case study for a complex geological terrane. The maps cluster in 3 different groups according to their geological variability. Stratigraphic variability maps identify regions of high variability associated with the overall geological structure. Information entropy maps allow for the identification of uncertainty associated with varying geological interpretation. The local gradient of both variability maps and the geometry of highly variable regions can be used to assess the source of variability: varying location, differing interpretations and/or reliability of observations. Is it possible to retrieve karst network geometry and properties using an inverse technique?

Download / Links

BibTeX Reference

@inproceedings{GroseGM2014,
 abstract = { Geological uncertainty can be identified during all stages of the geological workflow, from data collection to interpretation. The creation of three dimensional (3D) models often involves interpolating a series of two dimensional (2D) datasets (e.g. maps and cross sections). Uncertainties in the 2D datasets will propagate into the 3D model. This study investigates the uncertainties associated with lithological contacts in geological maps and will compare these results to other studies investigating the uncertainty associated with orientation datasets. Geological maps are created from a combination of observation and human interpolation between outcrop observations. This study uses geological variability as a proxy for geological uncertainty and we present methods for identifying and analyzing geological variability surrounding geological structures. We use geodiversity to characterize map differences by a number of geometrical measurements assessing geologically significant map properties. Geodiversity analysis is performed using Self-Organising Maps (SOMs) which allow for n-dimensional datasets to be reduced to 2 dimensions for visualization and analysis. Using k-means clustering on SOMs, the varying maps can be categorised into groups with similar geological characteristics. We use 40 maps from a field course in Broken Hill, Australia, as a case study for a complex geological terrane. The maps cluster in 3 different groups according to their geological variability. Stratigraphic variability maps identify regions of high variability associated with the overall geological structure. Information entropy maps allow for the identification of uncertainty associated with varying geological interpretation. The local gradient of both variability maps and the geometry of highly variable regions can be used to assess the source of variability: varying location, differing interpretations and/or reliability of observations.
Is it possible to retrieve karst network geometry and properties using an inverse technique? },
 author = { Grose, Lachlan AND Ailleres, Laurent AND Laurent, Gautier AND Betts, Peter },
 booktitle = { Proc. 34th Gocad Meeting, Nancy },
 title = { Characterising uncertainty in geological maps using variability and geodiversity analysis. },
 year = { 2014 }
}