The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
363 
known in the two coordinate systems are necessary to estimate 
the three parameter values. 
A particular procedure was performed between the first and the 
second sequences of the acquired radar images to determine the 
coordinates of three reference points in the two coordinate 
systems. Three SAR images were acquired, each of which was 
related to one of the three positions (CRI, CR2, CR3) assumed 
by an active radar calibrator in the monitored scenario (Figure 
5). 
Figure 5. Active radar calibrator (left) and the CR1, CR2, CR3 
positions in which it has been installed (right) in order to 
correctly locate the acquired SAR images 
The active radar calibrator appears as a bright point in the radar 
intensity images, therefore it was possible to easily determine 
the coordinates of the positions assumed by the calibrator in the 
SAR coordinate system. The coordinates of these positions have 
instead been determined in the DEM coordinate system through 
topographic measurements. 
Thanks to the estimation (least square method) of the three 
parameter values, it was possible to calculate the coordinates, in 
the DEM coordinate system, of the positions assumed by the 
two antennas of the SAR instrument along its rail and to 
correctly apply the focusing SAR algorithm, in this way 
obtaining radar images in which the position of each pixel is 
known in the DEM coordinate system. 
Estimation of the measurement precision 
An estimation of the radar measurement precision was 
performed. There is also an error on the pixels which have a 
higher coherence than the lower limit value (0.9 in this case) 
that has to be estimated to determine the measurements 
precision. 
The method adopted in this case considers the distribution of 
the phase difference values of the interferograms concerning 
two stable areas of the scenario (Fig. 6 left). 
Figure 6. Representation of the two stable areas used to 
estimate the radar measurement precision (left) and distribution 
of the phase difference values of the interferograms (right) 
An histogram of this distribution is represented in Fig. 6 (right). 
The histogram is similar to a gaussian distribution, with a 
standard deviation (a) of 0.35 mm. Hence, a measurement 
reliability from a statistical point of view (precision) of ± 0.70 
mm at 95 % reliability has been obtained. 
For the results of the GB-SAR campaign see section 2.4. 
Test site in Florence 
In order to estimate the accuracy of the GB-InSAR by means of 
known displacements a measurement campaign using a test site 
in Florence was performed before the measurement campaign in 
Baveno. 
The GB-SAR was installed on a stable concrete platform in the 
University of Florence park. A particular device was installed at 
a distance of about 120 m from the radar, to generate controlled 
displacements. This device consists of an electric engine, a 
worm and a reflecting metal disk oriented in the direction of the 
radar (Figure 7). 
Figure 7. Device employed to generate controlled 
displacements (Florence test site) 
Thanks to this device, several displacements (varying from 0.05 
mm to 5 mm) were imposed with a precision of 0.01 mm and 
they were measured with the GB-InSAR technique. 
The main radar parameters adopted during the measurements 
are summarized in Table 2. 
Polarisation 
VV 
Target distance 
120 m 
Transmitted Power 
-3.0 dBm 
Band 
300 MHz 
Central frequency 
5.650 GHz 
Number of transmitted frequencies 
801 
Linear scansion length 
1.9 m 
Linear scansion point number 
191 
Antenna gain 
15 dB 
Measurement time 
65 min 
Table 2. Main radar parameters used during the measurements 
in the Florence test site 
The minimum and the maximum standard deviations of the 
measurements were 0.03 mm and 0.19 mm respectively, while 
the RMS (root mean square) was 0.13 mm. These values are 
smaller than the precision value estimated in the Baveno 
campaign because a target was used in this test site which was 
optimal for the interferometric measurement because of its high 
reflectivity and correct direction with respect to the radar 
instrument.