ANYLISIS OF ST. NICOLAS CHURCH BY MEANS GROUND
PENETRATING RADAR TECHNIQUE
I. Bosch, J. Gosalbez, R. Miralles and V. Genovés (Universitat Politècnica de Valencia, Spain)
The Ground Penetrating Radar (GPR) is a non-destructive testing technique based on the
transmission of high frequency electromagnetic waves into the material to be inspected.
The heterogeneities of the material cause that part of the energy of the wave is reflected
back toward the surface, while the rest of the energy continues travelling through the
material. The reflected signals are represented as a function of the distance versus time.
Therefore, as the antenna is moved spatially, a cross section of the inspected surface is
generated. Since GPR is a non-destructive testing technique, its use for testing historical
elements is very interesting because it allows the analysis of the element avoiding any
alteration and even without physical contact between the element and the sensor.
St. Nicholas Church is located in the city of Valencia and it is an example of Gothic
structure with baroque decoration. Some of the frescoes of this church were restored in
1920 while others are quite blackened and pending for a future restoration. For these
reasons, any analysis must be contactless to avoid any damage. Therefore, the use of the
GPR can be a useful tool to evaluate characteristics of the wall: thickness, moisture…
In this paper, the GPR combined with signal processing techniques are used. Firstly, 15
linear meters of the church’s dome were analysed using GPR technique with frequencies
between 0.9GHz and 2.6GHz. The main objective of this analysis was to determine the
optimum frequency range for estimating thickness and in homogeneities in the dome.
Secondly, it was analysed frescoes with 2.6GHz frequencies in order to test the feasibility
of contactless inspection.
The parochial church of “St. Nicholas of Bari and San Pedro Martir” is National Historic
Artistic Monument declared by Royal Decree 1757/1981, 5 June 1981 (BOE 10/08/1981).
It is based on a homonymous square, within urban historic site of Valencia, with certain
medieval reminiscences. It belongs to the Gothic churches that were built over an old
mosque, around mid-thirteenth century and it is one of the first Christian parochial
churches of Valencia.
A multidisciplinary team of architects, conservators and engineers from Universitat
Politècnica de València (UPV) is conducting a restoration during 2014, the main aspects of
this restoration are: consolidation of the pictorial base and layer, pictorial cleaning, gap
treatment and pictorial reintegration.
Within this work group, the Signal Processing Group (GTS) is a research group at the
Institute of Telecommunications and Multimedia Applications (iTEAM) of UPV. The GTS
group is involved in different aspects of the theory and application of statistical signal
processing. The main fields of research are optimal detection and classification, pattern
recognition, independent component analysis, higher-order statistics and time-frequency
spectral analysis. Applications include non-destructive testing (ultrasonic, impact-echo
methods, thermography, GPR ...), surveillance and monitoring (acoustic and infrared) as
well as biomedical signals. Within these applications, include the group's experience in
signal processing in the field of architectural restoration .
As it will be explained in GPR technique section, GPR is a non-contact and non-
destructive testing technique based on the transmission and reflection of electromagnetic
signals into the inspected material. Therefore, it could be useful to analyze materials or
regions where direct contact is not possible. Some examples are: blacking and deteriorated
paints, air pockets in walls,…
Taking into account the experience of GTS, this article analyzes, from the point of view of
signal processing, the different possibilities of using Ground Penetrating Radar (GPR)
technique, in the context of pictographic (wall paintings, sculptures and ornamental tiles)
and architectural restoration of the St. Nicholas Church.
GPR (Ground Penetrating Radar) is a nondestructive testing technique based on the
transmission of high frequency electromagnetic waves into the material to be inspected .
Material heterogeneities are the cause of this wave energy reflected back to the surface,
while the remaining energy continues to propagate into the material. The reflected energy
is detected by a receiver antenna, which may be the same as the transmitter or may be a
different one. The reflected signals are processed in real time and plotted as a function of
distance versus time, in this way, as the antenna is moved spatially, a cross section of the
inspected surface is generated.
The reflections are due to inhomogeneities, which must be due to changes in the electrical
properties such as dielectric constant, and may be associated with interfaces, changes of
humidity, degree of compaction, fractures, clays, foreign bodies... This implies a wide
range of application: estimating the depth, thickness, determination of groundwater levels,
archaeological studies and/or geological location of cavities or fractures.... Depending
on the location, orientation and type of heterogeneity as well as configuration parameters
(frequency, time of analysis, gain ...) used for inspection, the ease of detection will change.
Penetration capability depends mainly on two parameters: the wave frequency and the
dielectric properties of the inspected material. With low frequency antennas, higher
penetrations are achieved but at the cost of lateral and axial resolution. Materials that are
poor conductors such as dry sand, limestone, cement... offer greater penetration than most
conductive materials such as clays. Antennas of 500 MHz and 1200 MHz are able to
penetrate between 5 m and 0.5 m with a resolution between 5 cm and 0.5 cm. Therefore,
the choice of frequency of the antenna will be a compromise between penetration and
As we have mentioned, the heterogeneities cause reflections of the electromagnetic signal
energy at the surface of objects. This reflection will take time to come the surface and to be
collected by the antenna. This time is related to the depth reflector
d at which the reflector is
located and to the propagation speed mater ial
cof the material according to Equation (1).
The propagation speed depends on the dielectric properties of the material, which are a
complex combination of material composition and water content and it is possible to
synthesize in Equation (2).
is the relative dielectric constant of the material and ܿ is the speed of light.
Inspection time will determine the maximum depth and will be conditioned by ߝோ, which is
not a trivial parameter to set if the inspected area is composed by different materials or
layers. The inspection time should be set according to the penetration capacity , because,
for larger inspection times, wave will be completely attenuated and only noise will be
A GPR or radargram profile is formed by a series of pulses of radar placed one below the
other providing a profile image of the inspected area. Therefore, lateral variations of ߝோ in
the environment are detectable. Processing techniques are needed to improve the
visualization and interpretation of these images , . Some of these techniques are:
automatic gain processes, trying to keep the signal to noise constant with depth; temporal
filters for elimination of interference; or spatial deconvolution techniques that allow
recovering the original shape of the heterogeneities, which are distorted by the antenna
Background removal Improved vertical resolution Improved horizontal resolution
Figure 1. Processing of a radargram
In this section, the GPR equipment is described as well as the experimental measurements:
linear and frescoes areas.
For the measures proposed in this study, we used the measuring equipment SIR 3000
(Geophysical Survey Systems, Inc.) configured with antennas of different frequencies:
0.9GHz, 1.4GHz and 2.6GHz. Figure 2 shows the inspection equipment and the 2.6 GHz
Figure 2. a) GPR measurement equipment. b) 2.6GHz antenna and moving device
with motion sensors
The main technical characteristics of the equipment are shown in Table 1.
Table 1. Technical specifications of the SIR 3000
Acquisition rates 220 scans / sec at 256 samples / scan
120 scans / sec at 512 samples / scan
Bits/sample 8 o 16
Samples per scan 256, 512, 1024, 2048, 4096 o 8192
Gain [-20dB, 80dB] Manual or automatic
Two types of clearly different measurements were performed, depending on the inspection
areas. On one hand, a linear dome area was measured following a typical GPR
configuration with all available antennas and without restriction as to the contact surface.
On the other hand, frescoes areas of the church were measured. In this case, the challenge
was the implementation of the inspection technique because it has to take all precautions to
avoid any contact and any damage to the paintings. The choice of the right antenna was an
Linear dome area
Initially we analyzed the dome from its upper part as shown in Figure 3. The inspection
area covers a total distance of 15 m from the front door.
Figure 3. Elevation drawing of the church and inspection area circled in red.
In this case, the inspection was made from the corridor between the dome and the roof.
This area was unconstrained and thus it was possible the physical contact between the
antenna positioner and the inspected area (Figure 4); therefore, radargrams were perfectly
synchronized with the spatial positions.
Figure 4. Interior of the corridor between the dome and the church roof (left).
0.9GHz antenna in direct contact with the inspected area (right).
Three antennas with different frequencies (0.9GHz, 1.4GHz and 2.6GHz) and penetration
capabilities were used to inspect the same area with the same test parameters: 1024
samples at 16 bits / sample, with 64 measurements per second and a range of 20 ns.
Comparing the signals obtained from the three antennas, we note that the results are
consistent with each other, offering similar graphs (Figure 5). The different radargrams
present the same inhomogeneities with similar morphologies. However, as the frequency
increases, the penetration capability is less and attenuation is higher, but spatial resolution
Furthermore, as it can be appreciated, the different building elements are reflected in
radargrams: lintels are shown as areas without signal reflection due to their uniformity; key
domes are detected as early echoes as it is the thinnest part; and domes are shown as
changes in depth (Figure 5). Additionally, radargram 0.9 GHz shows an early echo
extending between 7.5 m and 10.5 m. This echo can be associated with a material change
or, more likely, to a change in its characteristics. If we compare this result with the analysis
of moisture already done, it is verified that an increase in moisture appears in the indicated
area. This moisture increase causes an increase of the reflected signal radargram, showing
the consistency between two analyses.
On the other hand, from Figure 5, we can deduce that: 0.9 GHz antenna generates the
highest signal level; meanwhile 2.6 GHz antenna provides a better signal to noise than the
1.4 GHz antenna, which offers better lateral resolution. Therefore, 1.4 GHz antenna may
be the option as a compromise between resolution and penetration capability. Furthermore,
because the dielectric constant of the material is unknown, it is not possible to establish
precise depths for the different detected construction elements.
Figure 5. Overlaid radargrams of different frequencies on building components
Measurements made in areas with frescoes are described in this paragraph. In this case, 2.6
GHz antenna was used because it offered the best resolution, with maximum theoretical
inspection capabilities between 25 and 50 cm (depending on the material to be inspected).
It was inspected from the surface of the paint into the construction element.
The main problem encountered when conducting the inspection was the imposibility to
make physical contact between the antenna positioning system and the paints. This was
due to the fact that the wheels of the positioner could move dust and dirt from one area to
another. Additionally, these wheels could damage paintings.
Because of this, less precise configuration was chosen, but very conservative with paints.
Antenna was separated from paints and moved manually through the wall at a distance of
about 1cm. The equipment was set to perform a continuous acquisition, rather than
synchronously with the positioning, as in the previous case. Thus, any physical contact
with the paintings was avoided by doing a freehand movement of the antenna. In return the
precise relationship between the spatial position and the GPR signal synchronization was
lost. Nevertheless, it has been possible to establish a partnership between the signal and the
positions of the wall.
In this case, the test parameters were 1024 samples at 16 bits / sample, with 64
measurements per second and a range between 7 ns and 10 ns, allowing a better depth
Different measures corresponding to different areas of the church, as nerves inspected from
different angles, areas near windows, etc... were performed. Each measure was repeated 2
or 3 times to check consistency and repeatability. An example of performed measurements
can be seen in Figure 6. In this figure, it is shown how different radargrams are taken for a
specific area. It is important to notice the detected material changes at the beginning and
the end of the radargram, and the detection of a air pocket in the middle of it. Additionally,
and thanks to the consistency between radargrams, it was possible to concatenate them
allowing the inspection of greater length than the original one. Figure 7 shows a
concatenation of different radargrams and a thickness change is clearly detected.
Figure 6. Example of frescoes measures and baggy detection
From this work, we can conclude that GPR is feasible technique for contactless inspection
and therefore it does not damage any delicate area, such as paints and frescoes. To avoid
any contact with paints surface, measures were taken freehand but relationship between
spatial position and GPR signal was not very precise. Non contact positioner, as lasers
ones, will be a future research line.
GPR is capable for detecting dielectric discontinuities, which can be associated to changes
of material, heterogenies, as air pockets (Figure 6), thickness changes (Figure 7), gaps or
characteristics changes, as humidity (Figure 5). It has been proven that GPR measurements
are consistent, which implies that detected artifacts are due to material changes and that
they are not due to external parameters as instrumentation noise, movements scaffold,
interference, randomness… This implies that radargrams are reliable and support
differential measurements to evaluate before and after restoration process. In addition, the
repeatability of measures allows obtaining bigger area representation by means
concatenation of different radargrams.
Different frequencies were also assessed in dome analysis and finally the 1.6 GHz antenna
has been chosen due to its compromise between penetration and resolution.
In both cases (dome and frescoes analyses), the depth axis of radargrams did not have
much precision due to the fact that ߝ
value was not known with enough accuracy. For this
was assumed to be 5 and a more comprehensive study is necessary in order to
obtain the real dielectric constant.
Figure 7. Concatenation of radargrams and thickness changes
(1) L. Vergara, I. Bosch, J. Gosalbez and A. Salazar. “Optimum detection of ultrasonic
echoes applied to the analysis of the first layer of a restored dome”, EURASIP Journal
on Advances in Signal Processing, 2007, pp.1-10.
(2) Maierhofer, C. “Nondestructive evaluation of concrete infrastructure with ground
penetrating radar”. Journal of Materials in Civil Engineering, Vol.15, No.3, 2003,
(3) Schrott, L., & Sass, O. “Application of field geophysics in geomorphology: advances
and limitations exemplified by case studies”. Geomorphology, Vol.93, No.1, 2008,
(4) A. Zhao, Y. Jiang, W. Wang. “Exploring Independent Component Analysis for GPR
Signal Processing”. Electromagnetics Research Symposium 2005, pp. 750--753. The
Electromagnetics Academy, Cambridge, USA (2005)
(5) J.X. Liu, B. Zhang, R.B.Wu. “GPR Bounce Removal Methods Based on Blind Source
Separation”, Electromagnetics Research Symposium (2006) 256-259.