Agriculture Reference
In-Depth Information
In situ spectral analyses were carried out to characterize and separate the red dust optical
spectral features and shapes from other soils and backgrounds as well as to construct a
spectral library of different materials useful for calibrating and validating the remote
sensing data acquired within the study area. At ground reflectance spectra in the 350-2500
nm range were acquired on red mud and red dust widespread on nearby soils using a field
portable spectrometer (ASD), from a height of 1 m using a field of view of 25° and a
spectralon panel in the same geometry (i.e. at-nadir on the samples) as a white reference to
enable directly conversion of the measurement data into reflectance values.
Moreover, several dry red mud samples from the red mud impoundment and soil samples
with different concentrations of surfacing red dust were collected for laboratory analyses.
The mineral composition of the collected samples was determined by XRD. Pure dried red
mud samples and red dust polluted soils were analysed for bulk mineralogy on randomly
oriented powders of whole rocks (Srodon et al., 2001). The samples were scanned from 2° to
70° 2θ. The XRD patterns of the pure red mud samples has confirmed the main presence of
hematite, goethite, gibbsite and boehmite, rutile and sodium as sodium aluminium silicates
or hydrosilicates, and of different red dust polluted soil samples exhibit, instead, the main
presence of quartz, phyllosilicates, feldspar and carbonates, and goethite/hematite are also
present.
The chemical abundance of major and trace elements (e.g., Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Ga,
As, Rb, Sr, Y, Zr, Se and Pb) in the samples were measured using tube excited EDXRF. Using
this system, optimum sensitivity for a particular element is achieved having the dual effect
of improving sensitivity and reducing background.
Finally, MIVIS hyperspectral airborne remote sensing data collected over the study area
were calibrated to radiance and atmospherically and geometrically corrected to obtain
reflectances from the Visible to the Short Wave Infrared (0.4-2.5 m) (Bassani et al., 2006).
The reflectance spectral library analysis of the collected red dust samples showed that the
dominant spectral features in the VIS-NIR (0.4-1.5) range are primarily related to the iron
oxides (hematite - Fe
2
O
3
and/or goethite - Fe
3+
O(OH)) absorption features. Phyllosilicates
(clay minerals) are also largely present in the samples in the form of gibbsite and their
strong spectral absorption feature in the SWIR region due to a combination of the O-H
stretching fundamental with the Al-O-H bending mode (Clark, 1999) severely influences the
SWIR spectral behaviour of the red dust polluted soil samples.
Figure 6 shows an example of reflectance spectra (black line) acquired by ASD spectrometer
under controlled conditions in our laboratory on a representative soil sample with a high
concentration of red dust on its surface in comparison with the pure USGS spectra (available
at: speclab.cr.usgs.gov/spectral-lib.html) of the main minerals constituent of the red dust
samples as derived from XRD analysis. This comparison was performed for the each
spectrum acquired in the field in order to individuate which mineral constituent primarily
influences the red dust spectrum shape and absorption features. Results attained for the
analyzed polluted soil samples show that, even on naturally red soils, the red dust
deposited from wind is spectrally detectable if a high level of red dust is present on it.
In Figure 7 is presented the map of the red dust distribution on soils as obtained by
classifying airborne hyperspectral reflectance data using a simple and fast spectral-shape
based algorithm (i.e. the Spectral Feature Fitting procedure, see Segl et al., 2003 and the
references therein). The reference soil field spectra in the 0.4-2.5 m spectral region were
scaled to match the image spectra after they were normalized with the continuum removal
approach to allow the comparison of individual absorption features using a common
