AbstractThe London Clay Formation (LCF) underlies much of the London and Hampshire Basins and is well known for being susceptible to slope instability and shrink-swell. The engineering behaviour of this unit is closely related to lithology and mineralogy, both of which vary laterally and vertically. Much variation is controlled by a complex stratigraphy that is difficult to recognise due to a monotonous appearance, limited exposure and weathering. Robust stratigraphic identification requires fossil evidence, though clay mineralogy, geotechnical characteristics or natural water content have been used.
This study used Visible-Near-Infrared Spectroscopy (VNIRS) to examine 10240 samples of the LCF taken from a continuous coastal exposure at Whitecliff Bay, Isle of Wight, at a sampling resolution of 10mm in clay units and 1m in sand units. This was supported by the results of 100 X-Ray Diffraction (XRD) spectra and 100 particle size distribution (PSD) tests carried out on samples taken at 1m sampling throughout the succession. Two sets of spectral end members were derived from the 100 well-characterised samples, one from clay mineral library data and a second using XRD and PSD. In addition, a new measure, termed ‘volumetric clay content’, was derived that estimates the volume of specific clay species within a sample. It is suggested that it is more closely linked to the relative volume and, therefore, the chromosphoric effect of clay species within a soil sample.
Statistical analyses of all 10240 samples using Spectral Angle Mapper (SAM) were to derive measures of spectral similarity to all end-members. Results are presented as a wireline log that agrees with existing stratigraphic models with some variation. Further analyses of the spectral end-member data showed that the 100 well-characterised samples could be classified by plotting the similarity of samples to the end members representing Volumetric Illite-Smectite (V-IS) and Silt content. Threshold values were established (0.977 and 0.987, respectively), samples with values lower than both were found to have less than 30% content clay and silt (and considered coarse grained). Thus, using spectral data alone, it was possible to consistently classify fine and coarse grained soils and characterise the whole LCF accordingly. Further, it was found that a simple statistical treatment, indicating the number of samples per m length with values above both thresholds (termed Counts per m High V-IS/ High Silt) could be used to uniquely identify units within the established stratigraphic succession (A-E). This technique, with development, could lead to more efficient and rapid litho-stratigraphical identification in the LCF and other monotonous sedimentary sequences worldwide.
|Date of Award||Oct 2021|
|Supervisor||Andy Gibson (Supervisor), Nick Koor (Supervisor) & Andy Gale (Supervisor)|