Professional Documents
Culture Documents
Tomografia Elettrica
Tomografia Elettrica
ABSTRACT
Cavities located in shallow layers of the subsoil represent a major hazard especially in sites located
in town centres. For this reason, it is necessary to determine their location and size in order to
evaluate the risk of subsidence and to draw up plans for restoration and safety.
In this study, electrical resistivity tomography was carried out, both to detect cavities and to test
a new inversion algorithm that we developed. The cavities are located in a volcanic formation in a
public recreation park where there are municipal buildings. The survey was performed using two
different arrays: dipole-dipole and Wenner. A new approach to adopting an initial model and ine-
quality constraints was used and the results from different algorithms were compared. The effective
range of the dimensions of cavities was determined in order to make an approximate evaluation of
ground stability and any potential subsidence hazard.
© 2006 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2006, 4, 387-392
ERT to detect buried cavities in Rome 389
or other direct surveys, to limit the number of unknowns and Two different surveys were performed in the past. In the first,
reduce the ill-posedness of the problem. Furthermore, the possi- drill-holes were made in a small area of the park; cavities of dif-
bility of defining a discretization of the domain to conform ferent sizes and located at different depths were detected in this
exactly to the underground structures or of reducing the size of area (Table 1). In the second, a geophysical survey was carried
the mesh of the grid certainly provides an opportunity to obtain out using ERT and electromagnetic techniques. In this second
improved solutions. Our code allows the option of introducing survey, high-resistivity anomalies were detected that can pre-
nodes at any (x, z) coordinate of the domain. Such flexibility in sumably be linked to the presence of cavities.
building the grid is important, both in the modelling step, to
represent complex geometry structures, and in the inversion step;
the possibility of varying the block thickness and width ensures
a better fit to underground structures. In addition, the smoothness
matrix can be corrected in order to penalize roughness in either
the x- or y-direction (see Appendix). An empirical method of
assigning the regularization parameter (the damping factor) is
defined, but it can also be assigned by the user at each iteration.
An appropriate tool was then constructed to control the inver-
sion results, for example to correct reconstructed models and to
check the effects of such changes on the calculated apparent
resistivity.
TABLE 1
(a) Results of drilling survey; (b) percentage of holes in which different
numbers of cavities were detected
(a)
Result of drilling survey Percentage
Drillings with cavities 60%
Drillings without cavities 40%
(b)
Number of cavities detected
Percentage FIGURE 2
by drilling survey
One cavity 51% Inversion results for Line 2. (a) Resistivity distribution obtained by
standard inversion; (b) resistivity model; (c) resistivity distribution
Two cavities 37%
obtained using the VERDI algorithm with the starting model; (d) plot of
Three cavities 12%
RMS versus number of iterations.
© 2006 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2006, 4, 387-392
390 E. Cardarelli, G. Di Filippo and E. Tuccinardi
FIGURE 3
Inversion results for Line 5. (a) Resistivity distribution obtained by
standard inversion; (b), (c), (d), (e), (f) resistivity distributions obtained
using the VERDI algorithm with different starting models; (g) plot of
RMS versus number of iterations.
FIELD SURVEY
Because previous surveys had not covered all the area of the
park, we surveyed additional 11 electrical profiles located as
shown in Fig. 1. Blue lines represent the new survey that was
performed with Wenner and dipole-dipole arrays using 48 elec-
trodes with 2 m or 3 m spacing between electrodes. The length
of the lines varies between 94 m and 141 m. Red lines in Fig. 1
indicate the earlier survey. Wenner arrays with 2 m or 3 m elec-
trode-spacings were performed using a maximum of 16 or 15
electrode-spacings, respectively; the maximum number of meas-
urements for each ERT line was 360. Dipole-dipole arrays were
performed using a maximum dipole separation of 6 electrode-
spacings; the maximum number of measurements was 476.
RESULTS
Resistivity sections, and the results obtained with standard inver-
sion and with VERDI are compared for three representative
profiles. The resistivity distribution of Line 2, obtained starting
from a half-space of 100 Ωm is shown in Fig. 2(a). The results
obtained for Line 2, starting from an initial model characterized
by a bounded anomaly (Fig. 2b), are shown in Fig. 2(c). In this
case, the resistivity of the anomalies could vary between 500
and1000 Ωm (inequality constraints); this interval was suggested
because in some cases these cavities are partially filled due to
collapse of the vault. In such a case, if the resistivity values in the
bounded anomaly vary between two limits (e.g. 500 and
1000 Ωm), it should be possible to recognize the above-described
condition (i.e. partially filled cavities). At the same time, a mean
value of 100 Ωm was used for the surrounding area. The model
that we used to invert field data with VERDI was suggested by
the results of the previous inversion (standard inversion). Only
the largest resistivity anomaly was taken into account in the
model. We assumed that the anomaly on the right in the profile
was due to lateral resistivity contact. A comparison of root-mean-
square (RMS) errors of both inversion procedures is seen in
Fig. 2(d). Note that the results obtained for the model in Fig. 2(c)
show lower RMS values.
The results for Line 5 are shown in Fig. 3(a, b c, d, e, f). The
results obtained, starting from a half-space of 100 Ωm and with
no constraints, are shown in Fig. 3(a). Figure 3(b–f) shows the
results obtained from different models, characterized by different
sizes and locations of the bounded anomalies, and using inequal-
ity constraints. In the bounded anomalies, the resistivity ρ could
be in the ranges 500 Ωm<ρ1<3000 Ωm and 100 Ωm<ρ2<1000 Ωm
for the higher and lower anomalies, respectively. In all models,
the same damping factor was used, starting from a value of 0.1
and halving it at each iteration. Figure 3(g) shows a plot of RMS
© 2006 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2006, 4, 387-392
ERT to detect buried cavities in Rome 391
FIGURE 5
Assonometric view of the resistivity profiles
ties, with respect to both the speed of execution and the low cost
of surveys. To reduce the high degree of indeterminacy of ERT,
we applied a new inversion method. By using VERDI, we were
able to determine accurately the probable sizes, shapes and loca-
tions of the cavities. Also, the use of a standard smoothness-con-
strained inversion program was important in defining the starting
model that we used for the new approach (a posteriori informa-
tion).
The possibility of using an algorithm that permits us to insert
the shape and size of anomalies (in this case cavities) into the
starting model, and that includes the possibility of adding resis-
tivity values in bounded anomalies ranging within a closed
FIGURE 4 interval, is certainly the best way to obtain a solution close to the
Inversion results for Line 4. (a) Resistivity distribution obtained by physical reality of the subsoil.
standard inversion; (b) resistivity distribution obtained using the VERDI Furthermore, a method of estimating the shape and dimen-
algorithm with the starting model; (c) plot of RMS versus number of sions of the cavities is a very useful tool for town planners, mak-
iterations. ing it possible to evaluate the hazard of cave-in and to plan the
reclamation or infilling of the cavities.
versus the number iterations. Note that the same minimum val-
ues are reached in models B and C. This means that the true size REFERENCES
and location of the cavities are probably somewhere between Barker R. 1992. A simple algorithm for electrical imaging of the subsur-
face. First Break 10, 53-62.
these two models, in accordance with the average size of the
Benson A.K. 1995. Applications of ground penetrating radar in assessing
cavities detected in this area. The results for Line 4 are shown in some geophysical hazards: examples of groundwater contamination,
Fig. 4. In this case, the smoothness-constrained inversion also faults, cavities. Journal of Applied Geophysics 33, 177–193.
detected a high-resistivity anomaly that was modelled subse- Bernabini M., Brizzolari E., Monna D., Padula G., Piro S. and Versino L.
quently using the VERDI algorithm. The inversion results are 1982. Individuazione di cavità sepolte mediante prospezione geoelet-
trica. Esempio di applicazione: ricerca di tombe nella necropoli sabina
shown in Fig. 4(a, b). In Fig. 4(c), the RMS errors are plotted
di Colle del Forno nei pressi di Montelibretti (Roma). Bollettino del
versus the number of iterations; here, too, better results are Servizio Geologico d’Italia.
obtained using a starting model characterized by bounded Borelli G.B. 1999. Metodologia di indagine per ricerca e mappatura di
anomalies and using inequality constraints. cavità sotterranee, con particolare riferimento alle aree urbane.
Finally, a map summarizing the results of the whole survey is Proceedings of ASSONET Meeting, “Le cavità sotterranee nell’area
urbana di Roma e nella provincia. Problemi di pericolosità e ges-
shown in Fig. 5. In this map, a good correspondence between the
tione”.
sections can be noted in terms of size and depth of anomalies. Cardarelli E. and Fishanger F. 2003. 2D data modelling by electrical
resistivity tomography for complex subsurface geology. (Is it possible
CONCLUSIONS to improve interpretation of electric resistivity measurements with
Electrical resistivity tomography (ERT) is, without doubt, one of multielectrodic devices in undetermined cases?) Proceedings of 9th
EEG Meeting, Prague, Czech Republic, Inv-01.
the most suitable geophysical techniques to detect buried cavi-
© 2006 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2006, 4, 387-392
392 E. Cardarelli, G. Di Filippo and E. Tuccinardi
© 2006 European Association of Geoscientists & Engineers, Near Surface Geophysics, 2006, 4, 387-392