ınbul 2004 
cause, as 
ons on the 
ual to the 
cessary to 
n is partly 
behaviour 
he 
ed 
d 
=1-p; 
of 
p 
ors, while 
brated in 
order to 
antity of 
> image’s 
the mid 
image is 
quisitions 
'alculated 
nal bands 
ue of the 
ining the 
hes to a 
when the 
- contrast 
value of 
ing area. 
| because 
thermal 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
  
where 
T(x,y) is the local value of the temperature 
T,,(x,y) is the average value of the surrounding area. 
The dimension of the area in which you average the temperature 
value. is chosen in relation to the deduced dimensions of the 
underground structure. The average value was extracted using a 
convolution filter, with a Kernel of 3 pixels. However in the 
average value, the value of the temperature of the Kernel's 
central pixel was not taken into account. This choice is justified 
by the fact that if the temperature of such pixel is not much 
removed from the pixel, which it calculates, that is if there are 
no anomalies, its value will have no relevant influence on the 
average value. 
[n the case of notable deviations, also taking into account the 
temperature of the central pixel, the average would be closer to 
the value of the central pixel temperature reducing DN value of 
the resulting image therefore the contrast of the image making 
the photointerpretation operation more difficult. 
5.4. Thermal Conductivity 
Thermal Conductivity is a parameter, which enables the 
discovery of humid areas, in which water is an optimum heat 
conductor. Having available the two successive elevations of 
temperature in the reheating stage it is possible to extract 
indications regarding thermal conductivity. Supposing that the 
image is composed of many prisms of indefinite lengths and 
that the lateral faces of the prism are in contact with those of the 
adjacent prism, under the effects of a surge of heat applied to 
the superficies of the prism the transmission of the heat from the 
lateral faces of the adjacent prisms is more or less rapidly 
depending on its thermal conductivity. If the conductivity is nil, 
the thermal gradient G between the prisms, dependant on the 
source of heat and the adjacent prisms, increases in proportion 
to the intensity of the heat surge. 
This implies the relation between the temperature gradient and 
the value of the temperature of the prism, depending on the heat 
surge (normalized gradient G,), is constant, independent of the 
intensity of the heat (A.M. Tonelli, 1997) 
G = Abs(T(x+1,y)-T(<,y)+Abs(T(<,y+1)-T(<,y)) 
G,-Abs(T(x,*ly)- T(x.y)- Abs(T(x,y--D-T(x,y Y T(xy) 
Whereby if the intensity of the surge increases from t, to t; the 
value of the normalised gradient is maintained as a constant 
Gn(t, )-Gn(t,)=0 
For elements with a thermal conductivity other than zero, the 
expression takes the name Space Temporal Variation (STV) and 
assumes values as near zero as the thermal conductivity is less. 
In this experimentation the Thermal Conductivity was 
calculated for both aerial swaths. 
5.5. Synthesis map 
The images relative to the above mentioned parameters studied 
were put together so one unique map was produced where 
lineation of the presence of humidity could be individuated. To 
generate this map a greater importance was given to the 
parameters more sensible to humidity; in particular the presence 
of water has a great influence on the Thermal Conductivity, on 
the Thermal Inertia, on the Thermal Deviation and less on the 
NDVI. Putting together all these considerations the following 
amplifying factors were chosen (Tab. 1). In figure 5 the images 
of some zones, which are characterised by the anomalies 
comparable to the possible presence of underground structures, 
were compared with real colour image. The Villa del Casale is 
evident within the magenta circle. 
Each colour of rectangle corresponds to an area characterised by 
the possible presence of underground structures. 
  
  
  
  
  
  
  
Parametro investigato | Fattore moltiplicativo 
Conduttivita Termica 10 
Inerzia Termica 7 
Scostamento Termico 5 
NDVI l 
  
  
Table 1. Amplifying factors 
  
  
  
  
  
   
  
    
   
m 
P 
= 
1 
3 
5 
aee 
YS 
Á 
"e. 
"e 
  
  
  
Figure 5. Zones characterised by anomalies 
495