Abstract:
Swelling soils are a major geological hazard worldwide and their
identification is a major concern. Swelling is due to the presence of clay minerals
with a potential to accommodate water within their crystal structure. Three
common groups, namely smectite, illite and kaolinite, have been identified as
indicators of high, moderate and low swelling potential soils where in
abundance. This study assessed the potential of spectral data as a tool to
recognize the presence of these minerals in soils of varying swelling potential
collected from the Antequera area of southern Spain, with the aim of
establishing spectral parameters that would in future be applied in remote
sensing applications for mapping swelling soils. The coefficient of linear
extensibility (COLE) and the linear extensibility (LErod), which are established
methods for assessing swelling potential and tentative clay mineral allocation,
were used as the controlling methods to classify the soils into swelling potential
groups and to assign them to dominant clay mineral types. Several spectral
parameters were identified as giving a good basis for assigning the soils to
domination by either of these clay minerals, and thus to a swelling potential class
based on their diagnostic assignment to hydroxyl (OH) and molecular water
vibrational processes in clay minerals. The parameters consisted of asymmetries
(1400 nm and 2200 nm) that were strong in the presence of the more ordered
kaolinite, wavelength position and depth at 1900 nm in abundant smectites, first
and second derivative peaks at or near the molecular water feature and the
unique presence of absorptions at 2170 nm and 2340 nm in abundant kaolinite
and illite. The results show the potential for using reflectance spectroscopy as a
tool in the classification of soils into domination by these indicator minerals and
thus into swelling potential classes. However, high organic matter and the
presence of moisture were found to affect area and depth intensities and would
require consideration in such applications.
