240 
The analysis of the contour maps reveals the unreliability 
of some algorithms, for the following reasons: 
the method N.N. produces contour lines with a 
course to irregular grid, with lines of discontinuity 
conformed in real precipices, absolutely not 
correspondents to the real morphology of the ground; 
the method P.R., being a method of global 
interpolation, confirms its unsuitable to represent the 
physical surface of the ground, also with the use of a 
polynomial of degree 10; 
the R.B.F. and the SH.’s method produce graphics 
with numerous anomalous peaks (bull's eyes); the R.B.F. 
especially in the low part of the landslide, the SH. in all 
its extension. 
The remainders four methods elaborate DEMs that 
produce contour line graphics without apparent 
anomalies. 
For the most meaningful methods and for which was 
possible, the options of default have been varied to verify 
the behaviour of the algorithms and to choose the best 
options to interpolate the set of data. The tests have been 
performed only on SA20. 
To varying the disposition of the data (simple, quadrant, 
octant and all data) the values of the residuals and their 
statistic parameters didn't change significantly; therefore 
the options of default have been maintained and some 
other free parameters have been varied. 
In the following the fulfilled tests are described. 
- Inverse Distance to a Power 
Power from 1 to 5 has been tested while the default 
option is equal to 2. 
In table 3 you can note that values of the parameters are 
slightly dependent from the power, sort exception for the 
interpolation of power 1 that furnishes minimum, 
maximum and standard deviation values of residual very 
high. From 3 r power on the differences are very small. 
Kriging 
The performed tests have concerned the statement of 
different functions of variogram. Changing the options, 
the standard deviation values and the residual maximum 
and minimum stay substantially unchanged. 
Minimum Curvature 
The only option of the method concerns the maximum 
value of the residual and the number of the iterations. 
Varying these values the residuals increase notably, as the 
average and the standard deviation. 
Radial Basis Functions 
Changing type of function in comparison to that of 
default (Multiquadric) it doesn't change so much in terms 
of residuals and statistic parameters. 
Value of the default smoothing parameter R 2 is equal to 
139; in table 4 you can note that setting the value equal to 
zero, the parameters of interest worsen notably while an 
intermediate value in comparison to that of default 
doesn't produce significant variations. 
- Shepard's Method 
Changing the value of the smoothing parameter with 
respect to that of default (equal to zero), the results 
worsen notably and they become unacceptable, as it is 
visible in table 5. 
SA20 
Average 
St. Dev. 
; Minimum 
Maximum 
l.D. 
[cm] 
Power 1 
5.9 
186.4 
-1039.5 
794.2 
ID. 
[cm] 
Power 2 
0.5 
37.9 
-343.2 
231.8 
l.D. 
[cm] 
Power 3 
0.4 
23.3 
-250.7 
191.3 
l.D. 
[cm] 
Power 4 
0.4 
21.1 
-251.5 
173.3 
l.D. 
[cm] 
Power 5 
0.4 
20.4 
I -253.8 
156.1 
Table 3 - SA20 landslide. Test with Inverse Distance to a Power method. 
SA20 
Average 
St. Dev. 
Minimum 
Maximum 
R.B.F. 
[cm] 
R 2 = 139 
-1.2 
8.9 
-67.7 
74.5 
R.B.F. 
[cm] 
R 2 = 70 
-1.1 
8.2 
-65.0 
79.2 
R.B.F. 
[cm] 
r 2 = o 
-0.4 
35.3 
-226.6 
289.6 
Table 4 - SA20 landslide. Test with Radial Basis Functions method. 
SA20 
Average 
St. Dev. 
Minimum 
Maximum 
SH. 
[cm] \ Smooth = 0 
-6.9 
54.4 
-444.0 
249.6 
SH. 
[cm] \ Smooth = 1 
-568.6 
1498.3 
-11642.8 
5354.4 
SH. 
[cm] : Smooth = 2 
-590.8 
1602.2 
-14394.9 
5950.9 
Table 5 - SA20 landslide. Test with Shepard’s method