The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
be found in Brereten (2000), Martens and Naes (1989), Wold et 
al., (2001), Yeniay and Goktas (2002), http://www.camo.com. 
3. RESULTS AND DISCUSSIONS 
3.1 ENGINEERING AND SPECTRAL PARAMETERS 
Values of engineering parameters indicate that the soil samples 
have a wide range of variability in their expansion potential (e.g. 
Figure 2 plasticity chart). Distribution of samples on plasticity 
chart is useful in getting indication of soil expansion potential 
(Dakshanamurty and Raman, 1973). Consistency or Atterberg 
limits are ranges of consistency (the ease with which a soil can 
be deformed) of cohesive soils as a function of changes in 
moisture content (Perloff and Baron, 1976). They represent 
empirical boundaries which divide various states that cohesive 
soils exhibit with varying amount of moisture content; solid, 
semisolid, plastic and semi-liquid states. Since water has a 
significant effect on engineering behaviour of clayey soils in 
such a way that clayey soils with higher moisture content are 
weaker and easily deformable than their same varieties with 
lower moisture content (Lambe and Whitman, 1979; Mitchell, 
1993; Perloff and Baron, 1976), determination of consistency 
limit values of such soils has proved to be useful in engineering 
applications. 
Figure 3. Distribution of soil samples on plasticity chart. 
Differences in spectral characteristics among spectra of 
different soil samples were used in differentiating various clay 
mineral types present in the soil samples. Position of absorption 
features, their shapes, types and number, depth intensity and 
asymmetry; shape of spectral curves, differences in slopes of 
spectral curves and variations in reflectance intensity of spectra 
were some of the important qualitative parameters that helped 
to identify spectrally dominant clay mineral from the soil 
reflectance spectra. Up on spectral interpretation, spectra of soil 
samples were grouped into three major classes of mineralogical 
composition; smectites, mixtures and kaolinites (Figure 3). 
Among smectite classes are montmorillonite and nontronite and 
of kaolnite groups are halloysite and kaolinites. Those that are 
grouped under mixtures are a mixture of smectites, kaolinites 
and others. 
Figure 4. Variability in spectral characteristics of different soil 
samples (no offset). Note the differences in shapes of spectral 
curves; overall reflectance intensity, shape, position and 
number of absorption bands among the spectra. 
Relationships between measured engineering parameters and 
mineralogical classes obtained upon spectral interpretation with 
respect to the magnitude of relationship each mineralogical 
classes show with absorption feature parameters were examined 
(Figure 4). 
Figure 5. Scatter plots showing the relationship between depth 
at -1900 nm and liquid limit of different clay mineral 
categories; showing the magnitude of the linear relationship per 
mineralogical groups (kaolinites, mixtures & smectites). 
The magnitude of relationship between absorption feature depth 
at ~ 1900 nm and liquid limit of samples is highest for smectites 
that exhibit strong absorption band at ~ 1900 nm due to 
adsorbed water in their structure, followed by mixtures. 
Kaolinites that show no (kaolin) or less resolved (halloysite) 
absorption at ~ 1900 nm show the lowest correlation between 
liquid limit and depth of absorption feature at ~ 1900 nm. 
3.2 PARTIAL LEAST SQUARES (PLSR) PREDICTION 
MODELS 
Absorption feature parameters (position, depth, area, width of 
absorption feature) calculated from absorption bands at -1400 
1321