687
Data concerning Gl and G2 sub-frames are estimated
and extrapolated since no drill has ever been tried.
The top of the reservoir is rather shallow at site
LG which can be referenced as surely affected by ge
othermal influence. The following cross section, a-
fter Calore et al. (1979), illustrates the geothermal
situation schematically.
Green and Red which correspond to the reflected e-
nergy Green, Red and Infrared.
Moreover, since the object of our investigation is
to outline fine anomalies within visually homogeneous
vegetated zones, we also have applied to the band
ratios IR/Red, IR/Green and the so called Normalized
Difference Vegetation Index (NDVl) where:
. Impervious
' complex
Main aquifer
Basal complex
-1500 —* t
Jtfm
J Statute mile
NDVI = —
IR + Red
On the basis of frequency distribution of DN present
in the NDVI image and in agreement with the actual
values of NDVI, we have operated a suitable level
slicing as coded as follows.
Table 4. Correlation between DN intervals and actual
values of NDVI represented on the image.
DN intervals Grey Level Code NDVI Values
Figure 3. Geothermal cross section of the Travale
field. Arrow outlines location of the sub-frames Gl,
G2 and MA.
The top of the reservoir dips north-eastward, benea
th the cover of an impervious complex. The sub-frames
G1,G2 and MA are sited in correspondance of a tecto
nic feature -the graben of Travale- where associated
fractures are the easy vehicle of the geothermal fl
uid. This structural remark confirms the geothermal
potentiality of the suspected zones.
3 DIGITAL ANALYSIS OF THE I.R. AERIAL PHOTOGRAPHS
The investigation has continued on the basis of spe
ctral signature of vegetation, analysed on the IR
False Colour stereo pairs.
3.1 Image processing
In order to avoid faults due to aberration of the
camera lenses used and to the film developing pro
cesses we selected the frames so that each test area
occupies the central portion of them. Frames no. 035,
077 and 101 resulted to be in good agreement with
the requirements.
The 4 sub-frames previously identified were scan
ned, at a 100 pm resolution, and digitized by means
of a drum microdensitometer Optronics 0-4500,
Table 3. Dimensions of the windows digitized from
the original transparencies.
Site
Frame no.
Dimensions
(no. pixels)
Area
(ha)
Gl
077
212 X 173
5.7
G2
077
188 X 164
4.8
LG
035
119 X 137
2.5
MA
101
98 X 135
2.0
* 1 pixel = 1.25 m x 1.25 m
Each window was scanned under separate reading of
the 3 colour components of the film layers Blue,
0 -
- 126
0
<
0.
.0
127 -
- l4l
30
0.00
-
0.l4
142 -
- 156
70
0.15
-
0.29
157 -
- 171
110
0.30
-
0.44
172 -
- 186
150
0.45
-
0.59
187 -
- 201
190
0.60
-
0.74
202 -
- 216
230
0.75
-
0.89
217 -
- 255
255
>
0.
.9
The band ratios IR/Red and IR/Green correlate the
highest values of reflectance to the lowest and to
the Green peak of vegetation, respectively.
NDVI enhances the different stages of the vegetation
vigour.
The additive view of the 3 ratios also has been
attempted by different coding for visual interpre
tation. The most useful colour combination has
resulted: IR/Red —> Blue, Green/IR -> Green and IR
-> Red.
The results obtained by the digitization of IR
component and their elaborations have been imaged
in a plate (see fig. 4).
3.2 Output analysis
Data deriving from the image processing have been
represented under the form of images and statisti
cal parameters.
Since the sub-frames correspond to small and well
confined areas, the statistical approach has seemed
to be useful and more suitable than others.
■f. In particular, the arithmetic mean m, the standard
deviation a and the coefficient of skewness s have
been considered.
Values of standard deviations confirm that the
sub-frame MA and Gl are the most uniform units.
The values of a for IR/Green and NDVI images ap
pear to be smaller than in the other cases.
The skewness of ratios remains almost negative;
that means the baricenters of the histograms are
shifted to high DN of representation in relation to
the primitive images.
Mean values have been plotted and correlated.
