Abstract:
The fact that expansive soils are a major engineering problem makes their study an
important research aspect due to the accruing cost involved in terms of economic
loss when construction is undertaken without due consideration to the probability
of their presence. Though there are several methods that have been used to
recognise their presence a need of a fast and relatively cheaper methods continues
to be a necessary undertaking. In this thesis new methods are explored where the
various aspects of swelling soil properties are investigated consisting of
engineering, geophysical, reflectance spectroscopy and remote sensing where data
from two study areas one in central Kenya and the other in southern Spain were
used.
The study relied on the establishment of indicator spectral parameters as to the
presence of three clay minerals commonly used in soil classification to swelling
potential classes namely; smectites, illites and kaolinites. This was through several
reflectance spectra analysis techniques among which are; absorption feature
mapping, derivative analysis, optical density and band normalised with centre. This
was followed by the establishment of correlations between these parameters and
the commonly used physicochemical indices of Atterberg limits, Cation Exchange
Capacity (CEC), Saturated Paste (SP), and Coefficient of Linear Extensibility
(COLE) tests. In this, their widely accepted thresholds within which soils are
assigned to a swelling potential class and provisionally to a dominant clay mineral
were used.
This was followed by analysis of airborne hyperspectral data, in the Spain case
and Landsat Thematic Mapper (TM) image data for the Kenyan study area to
establish similar compositional differences. In both cases the lower spectral
information content was taken into consideration where not only the spectral
characteristics were used but also the surface expression of the soil compositional
differences in the Kenyan area in the form of gilgai microrelief. This information
was integrated with field based data consisting of topography, drainage and vegetation differences and correlations with field based soil classification maps to
establish the potential of remote sensing in the mapping of swelling potential.
Finally a prove of concept as to the potential discrimination of swelling soils under
buried non-swelling soils was also explored where two geophysical methods
consisting of Induced Polarisation (IP) and Nuclear Magnetic Resonance (NMR)
known to give indication as to the CEC, moisture and clay mineralogy differences were used at a laboratory scale with the aim of identifying the problems to be
overcome for such methods to be applicable in a field setting where they would
provide faster ways of establishing the swelling potential characteristics based on
the fact that these soil properties are the key to their swelling behaviour.
From the engineering methods, three of the indices were established to best
represent the potential volume change and consisted of those directly related to the
clay mineralogy type i.e. CEC and the Atterberg limits of liquid limit and plastic
index. This was interpreted to show them as most suited from an engineering
perspective in the identification of swelling soils and as best suited at the
exploration as to other methods capability at identifying these soils.
The absorption feature mapping technique established several feature parameters to
be diagnostic as to the dominance of these minerals in soils where significant
presence of kaolinite enhanced the hydroxyl features whereas substantial amounts
of smectite enhanced molecular water features. Manipulations of the spectral
curves in the form of first and second derivatives were also observed to give
similar information whereas other manipulations such as the optical density and the
band normalised with centre were not as promising. The potential of spectral data
to discriminate the soils based on the clay mineralogy differences was thus
concluded to show spectroscopy to have a potential at mapping swelling soils an
assumption that was finally confirmed through correlations between the spectral
parameters and the established swelling potential indices. This was also confirmed
based on spectral information from the hyperspectral image data where derivatives
established several wavelength positions to give strong indications of such a
potential. Landsat image data on the other hand added a new dimension to the potential
identification and mapping of swelling soils other than the spectral differences in
the form of recognition of gilgai topography pattern exclusively present in these
soils. This provides a possibility of carrying out more detailed analysis of the
potential differences among the swelling soils where spatial analysis of the gilgai
patterns in the form of parameters such as homogeneity index can be used to relate
soils in different regions based on the similarity of such indices and their
association to environmental factors. This when coupled with spectral differences
in the form of such indices as the soil brightness index were established capable in
assigning soils in the Kenyan study area to swell potential categories upon
integration with other field based information such as topography and drainage
patterns coupled with land cover differences. The geophysical methods (IP and
NMR) though laboratory based were also established as potentially useful in the
study of these soils and could be useful in their recognition in places where they could be buried based on good correlations obtained between the normalized IP
and the CEC on the one hand and moisture and grain size distributions and the
NMR parameters on the other. However, for the two methods to become
operational there are some factors that require to be addressed one of which is the
influence of salinity on the obtained IP and a proper calibration of the NMR to
measure the various properties on which it has a potential to give information on.
Reflectance spectroscopy was therefore concluded to have a potential application
in swelling soil identification and to offer a fast non-destructive method based on
diagnostic spectral features. Low spectral but high spatial resolution images were
found useful in the recognition of these soils and to offer a potentially important
tool for the comparison of the soils over wide areas through the established gilgai
pattern analysis based on their expression of underlying forces resulting from
differential swelling. Geophysical methods of NMR and IP were also established to
have a potential to non-intrusive identification of these soils if some influencing
factors could be overcome. Conclusions were therefore drawn that though soil
swelling is a complex phenomenon involving several underlying factors i.e.
compositional differences, structure and moisture regime etc, clay mineralogy
plays a central role in determining it and thus has an overall controlling influence
to most of the soil properties making their use in its estimation possible. The
research however established that there are still more handles to be overcome as to
the full operation of these new techniques.