Morris, Kevin Peter; (1995) Mapping three-dimensional geological features from remotely-sensed images and digital elevation models. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Mapping_three-dimensional_geol.pdf Download (44MB) | Preview

Abstract

Accurate mapping of geological structures is important in numerous applications, ranging from mineral exploration through to hydrogeological modelling. Remotely sensed data can provide synoptic views of study areas enabling mapping of geological units within the area. Structural information may be derived from such data using standard manual photo-geologic interpretation techniques, although these are often inaccurate and incomplete. The aim of this thesis is, therefore, to compile a suite of automated and interactive computer-based analysis routines, designed to help a the user map geological structure. These are examined and integrated in the context of an expert system. The data used in this study include Digital Elevation Model (DEM) and Airborne Thematic Mapper images, both with a spatial resolution of 5m, for a 5 x 5 km area surrounding Llyn Cow lyd, Snowdonia, North Wales. The geology of this area comprises folded and faulted Ordo vician sediments intruded throughout by dolerite sills, providing a stringent test for the automated and semi-automated procedures. The DEM is used to highlight geomorphological features which may represent surface expressions of the sub-surface geology. The DEM is created from digitized contours, for which kriging is found to provide the best interpolation routine, based on a number of quantitative measures. Lambertian shading and the creation of slope and change of slope datasets are shown to provide the most successful enhancement of DEMs, in terms of highlighting a range of key geomorphological features. The digital image data are used to identify rock outcrops as well as lithologically controlled features in the land cover. To this end, a series of standard spectral enhancements of the images is examined. In this respect, the least correlated 3 band composite and a principal component composite are shown to give the best visual discrimination of geological and vegetation cover types. Automatic edge detection (followed by line thinning and extraction) and manual interpretation techniques are used to identify a set of 'geological primitives' (linear or arc features representing lithological boundaries) within these data. Inclusion of the DEM data provides the three-dimensional co-ordinates of these primitives enabling a least-squares fit to be employed to calculate dip and strike values, based, initially, on the assumption of a simple, linearly dipping structural model. A very large number of scene 'primitives' is identified using these procedures, only some of which have geological significance. Knowledge-based rules are therefore used to identify the relevant. For example, rules are developed to identify lake edges, forest boundaries, forest tracks, rock-vegetation boundaries, and areas of geomorphological interest. Confidence in the geological significance of some of the geological primitives is increased where they are found independently in both the DEM and remotely sensed data. The dip and strike values derived in this way are compared to information taken from the published geological map for this area, as well as measurements taken in the field. Many results are shown to correspond closely to those taken from the map and in the field, with an error of < 1°. These data and rules are incorporated into an expert system which, initially, produces a simple model of the geological structure. The system also provides a graphical user interface for manual control and interpretation, where necessary. Although the system currently only allows a relatively simple structural model (linearly dipping with faulting), in the future it will be possible to extend the system to model more complex features, such as anticlines, synclines, thrusts, nappes, and igneous intrusions.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as XMLJSONCSV

Downloads by country - last 12 months

Export as CSVJSONXML

Archive Staff Only

View Item