Content uploaded by Rubén Iglesias
Author content
All content in this area was uploaded by Rubén Iglesias on Feb 25, 2014
Content may be subject to copyright.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
1
Persistent Scatterer Interferometry: potential, limits
and initial C- and X-band comparison
DESCRIPTION OF THE PAPER’S CONTENT
Potentials and limits of a powerful satellite-based technique to monitor land
deformation phenomena.
ABSTRACT
This paper is focused on Persistent Scatterer Interferometry (PSI), a powerful remote
sensing technique used to measure and monitor deformation phenomena. It only refers
to satellite-based PSI techniques, focusing in particular on the most important sources of
C-band SAR data: ERS-1/2 and Envisat. In addition, it compares C- and X-band results,
considering data from the high resolution TerraSAR-X sensor. The paper begins with a
concise description of the main characteristics of PSI. It then discusses the most
important PSI products and their performances, analysing in detail their spatial
sampling, the so-called residual topographic error and PSI geocoding, the average
displacement rates and the deformation time series. As C-band products are concerned,
the paper reports some relevant PSI validation results, which come from the ESA-
funded Terrafirma Validation Project. Regarding the X-band, it describes the results
obtained over the city of Barcelona by processing 13 StripMap TerraSAR-X images.
The last part of the paper discusses the main limits of PSI.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
2
1. Introduction
Persistent Scatterer Interferometry (PSI) is a radar-based remote-sensing technique to
measure and monitor land deformation. It is the most advanced class of Differential
Interferometric Synthetic Aperture Radar techniques (DInSAR) based on data acquired
by spaceborne SAR sensors. Theoretically, PSI techniques can be used with data from
terrestrial (see Luzi et al. (2004) and Crosetto et al. (2009)), or airborne SAR sensors.
However, spaceborne SAR sensors are by far the most important PSI data source. For a
general review of SAR interferometry and DInSAR, see Rosen et al. (2000) and
Crosetto et al. (2005).
PSI started with the so-called Permanent Scatterer technique proposed by Ferretti et al.
(2000). After this initial work, other approaches were proposed in the following years
(see Ferretti et al. (2001), Berardino et al. (2002), Colesanti et al. (2003), Mora et al.
(2003), Lanari et al. (2004), Hooper et al. (2004), Kampes and Hanssen (2004), Crosetto
et al. (2005), Pepe et al. (2005), Crosetto et al. (2008a), Hanssen and van Leijen
(2008)). Although these techniques were initially called “Permanent Scatterer
techniques”, now all of them, including the original Permanent Scatterer technique, are
usually called “PSI techniques”, while the term “Permanent Scatterers” is exclusively
associated with the original technique patented by Ferretti et al.
There are two main differences between DInSAR and PSI techniques: first, the number
of required SAR images (PSI uses large series of SAR images, typically more than 15-
20), and second, the implementation of suitable data modelling and analysis procedures
that allow one to obtain the following key PSI products: (i) the time series of the
deformation; (ii) the average displacement rates over the observed period; (iii) the
atmospheric phase component of each SAR image; (iv) and the so-called residual
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
3
topographic error. From the application viewpoint, the main products of any PSI
analysis are given by the map of the average displacement rates, and the deformation
time series of each measured Persistent Scatterer (PS).
The first part of the paper describes the main PSI products and their performances,
mentioning, among others, some of the most important PSI validation results achieved
in recent years. This includes discussion of the PSI area coverage and spatial sampling,
the characteristics of the residual topographic error and its impact on geocoding, the
sensitivity of PSI to small deformations, and the characteristics of the main PSI
products: maps of the average displacement rates, and deformation time series. The
second part of the paper discusses the key limitations of PSI at present and open
technical issues, which include PS spatial sampling, the limitation in measuring “fast”
deformation phenomena, the impact of the linear deformation model assumption made
in many PSI approaches, the tilts or trends in the PSI deformation rates, and the line-of-
sight PSI measurements. This is followed by conclusions.
2. PSI products and performances
PSI techniques have remarkably increased their capability as deformation measurement
and monitoring tools in recent years. In parallel, different efforts have been made to
validate the PSI techniques by assessing the performances of their products. In the
following sections we mainly refer to the validation results obtained in the ESA-funded
Terrafirma project (www.terrafirma.eu.com), a project of the GMES (Global
Monitoring for Environment and Security) Service Element Programme aiming to
establish a long-term market for PSI products. The Terrafirma Validation Project was
focused on the four PSI commercial service providers: Telerilevamento Europa
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
4
(www.treuropa.com), Altamira Information (www.altamira-information.com), Gamma
Remote Sensing (www.gamma-rs.ch) and Fugro NPA (www.npagroup.com). It
included the inter-comparison of the different providers’ processed outputs (to test their
“equivalence”) and product validation against ground truth (levelling and other
topographic data). The PSI products concerned two test sites located in Holland, which
were studied using ERS-1/2 (1992–2000, 83 images) and ASAR-Envisat data (2003–
2007, 39 images): the Alkmaar area, which includes spatially correlated deformation
fields due to gas extraction, and the city of Amsterdam, which includes autonomous and
mainly spatially uncorrelated terrain displacements. More details on these results can be
found in Crosetto et al (2008b) and other associated documents, all of which are
available at www.terrafirma.eu.com/Terrafirma_validation.htm.
2.1 Area coverage and spatial sampling
SAR imagery offers wide-area coverage, which is associated with a relatively high
spatial resolution. ERS and Envisat standard imagery, for instance, covers 100 by 100
km with an approximate pixel footprint of 20 by 4 m, while the StripMap TerraSAR-X
imagery used in this work (35.5º off-nadir angle) covers 30 by 50 km with a pixel
footprint of about 1.9 by 1.6 m. These characteristics make it possible to get a
comprehensive outlook of the deformation phenomena occurring in relatively wide
areas, while maintaining the capability to measure individual features, like
infrastructures and buildings, in good conditions. However, it is worth underscoring that
the actual PSI spatial sampling capability is much lower than that of the original SAR
imagery. This mostly depends on the characteristics of the observed scene. In fact, PSI
is an “opportunistic deformation measurement method”, which is able to measure
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
5
deformation only over the available PSs, i.e. the points where PSI phases maintain good
quality over time to get reliable deformation estimates. Using C-band ERS and Envisat
data the PS density is relatively high in urban areas, where densities of up to 800-1000
PS/km
2
can be achieved, while it is usually low or zero in vegetated and forested areas,
over low-reflectivity areas (very smooth surfaces), and steep terrain. It is worth noting
that the location of the PSs cannot be known prior to the PSI processing: this can
sometimes hamper the PSI applicability. The most straightforward advantage of the X-
band TerraSAR-X data is surely its higher spatial resolution. Figure 1 compares the PSI
C- and X-band results over an urban area: there are 107 C-band PSs, while with X-band
one gets 5070 PSs, 47 times more and about 39000 PS/km
2
. With the X-band data one
may get a very dense sampling of single buildings, which opens new application
perspectives. Importantly, this high sampling is partially due to the relatively small
observation period considered in this work (the results were achieved using 13
StripMap TerraSAR-X images covering a 9 month period from December 2007 to
September 2008): the PS density could slightly decrease is longer periods were
considered. Note that the two datasets do not cover exactly the same period (Envisat
covers 14 months, from November 2007 to January 2009): the differences in terms of
the estimated average displacement rates are not relevant in this context.
Figure 2 shows another interesting example of the X-band PSI sampling capability,
which refers to a part of the Barcelona airport. The left image shows the SAR amplitude
over this area, which includes a portion of the terminal and some runways, while the
right image depicts the available PS. The most relevant feature is the high PS density
over the runways (in this case there are more than 25000 PSs), which appear to be rough
at 3.1 cm X-band, causing a good and persistent response over time, while they are
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
6
basically smooth at 5.6 cm C-band, resulting in a weak response towards the SAR and
hence very few PSs over the same area.
Figure 3 shows another example of an area that includes large buildings, like the
Olympic Stadium and the Palau Sant Jordi (sports arena), and other forested and
vegetated areas in the bottom-right part of the image. In this case, we find a good PS
density over the buildings and other infrastructures, while there is almost no PS in the
forested and vegetated areas. The same phenomenon occurs with C-band data: neither
the C- nor the X-band produce signals that can be exploited for PSI purposes. This is a
severe limitation to the PSI monitoring of vegetated and forested areas at these
wavelengths.
2.2 Residual topographic error
The importance of the residual topographic error (RTE) is two-fold: it plays a key role
in the accurate modelling of the PSI observations (i.e. the PSI phases), and in
geocoding. The magnitude of the topographic phase component of the PSI phase is
usually reduced by simulating a synthetic topographic phase using a DEM of the
observed scene. However, any difference between the true height of a PS and the DEM
height generates the so-called residual topographic phase component, which has to be
properly modelled, estimated and separated from the PSI deformation phase component.
Additionally, the estimated RTE is used to get an improved geocoding of the PSI
products. In fact, the standard PSI geocoding only employs the external DEM (used in
the processing) to geocode its products: it does not consider the difference between the
true PS height and the DEM height, resulting in geocoding location errors. By using the
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
7
estimated RTE this kind of error can be largely reduced, and more precise geocoding
achieved.
The C-band PSI performances were assessed for RTE in the Terrafirma Validation
Project. The inter-comparison of results from different teams, computed over large sets
of common PSs, gave a standard deviation of the RTE differences ranging from 1.3 to
2.8 m. Assuming that all inter-compared results have the same precision and are
uncorrelated, we can derive an estimate of each team’s standard RTE deviation:
σ
RTE
= 0.9 - 2.0 m.
This has a direct impact on the PS geocoding: considering the ERS and Envisat
geometries, one gets the following geocoding standard deviation:
σ
GEOC
= 2.1 - 4.7 m.
Note that the above σ
GEOC
only includes the direct effect of σ
RTE
. Furthermore, it
roughly affects the east to west direction, because the impact of an RTE error occurs in
the direction perpendicular to the SAR track, i.e. roughly, the north-south direction.
Even though the above σ
GEOC
values are rather good for the geocoding of satellite-based
imagery, in practice they represent a limitation to the interpretation and exploitation of
PSI results. In fact, with an uncertainty of ± 3⋅σ
GEOC
and especially in urban areas, it is
often difficult to precisely identify what object corresponds to a given PS.
RTE validation results for the X-band PSI are not yet available. However, a first
impression of the achievable RTE quality can be derived from Figure 4, which shows
the colour-coded RTE estimated over the twin skyscrapers of Barcelona’s Olympic
Port. This result was obtained from a set of only 13 TerraSAR-X images, with
perpendicular baselines of up to 491 m. A very dense PS sampling over the two
buildings, each of which are 154 m high, is readily visible. Even though a quantification
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
8
of σ
RTE
is not available, the quality of the estimated RTE and the geocoding appears to
be very high: in the authors’ experience, comparable results had never been achieved
using C-band PSI. This improvement is due to a number of concurring reasons: the
different imaging geometry (the TerraSAR-X data shown in this paper were taken with
an off-nadir angle of 35.5º, while C-band data were typically taken at 23º), the higher
spatial resolution, the shorter wavelength and the improved phase quality.
2.3 Average displacement rates
The average PS displacement rates have been the most important PSI product so far,
mainly for the C-band studies. As in the case of RTE, its quality for the C-band data
was assessed in the Terrafirma Validation Project. From the inter-comparison of results
from different teams, computed over large sets of common PSs, the estimated standard
deviation of the displacement rates or velocities has been estimated:
σ
VELO
= 0.4 - 0.5 mm/yr.
This value provides information on the global behaviour of PSI velocities. However, it
is only representative of areas with characteristics similar to those of the Terrafirma
Validation Project test sites, e.g. urban areas with moderate deformation rates, etc. See
Crosetto et al. (2008b) for details.
No validation results concerning the X-band PSI have been published yet. An example
of average displacement rates is shown in Figure 1. The benefit of getting a high PS
density is clear; it allows us to properly sample the deformation induced by construction
works of an underground parking garage that affects a narrow area of about 200 by 25
m.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
9
Figure 5 shows another interesting feature of the X-band PSI measurements and the
derived displacement rates. The second and third images from the top show two
portions of co-registered interferograms (the phases are 2π-wrapped, colour-coded and
superposed to an optical image), which cover an industrial area. It is obvious that the
two central buildings show two opposite phase gradients: as it goes from left to right of
the top building, the first interferogram measures +6.3 rad, while the second one
measures -3.4 rad. Such phase gradients cannot be caused by any residual topographic
error, because they have no relation with the building geometry. In this case the phase
gradients seem to be due to thermal dilation. We will call the first image of an
interferogram master (M) and the second one slave (S). The two interferograms are
characterized by temperature gradients (i.e. temperature during acquisition of M, minus
temperature during acquisition of S) of opposite signs: +7º C vs. -4º C. This corresponds
to opposing thermal dilations, which are captured by the two interferograms. Note that
this dilation probably affects the buildings at hand horizontally. The relevance of this
result is two-fold. On one hand, it shows the high sensitivity to small displacements of
the X-band phase. On the other, it highlights a potential limitation, which also affects
other precise deformation monitoring techniques: the thermal dilation has to be properly
considered, especially for the analysis of short periods based on reduced SAR datasets.
In fact, for the displacement rates this effect tends to cancel out by considering long
periods of at least one year. However this might not be true for shorter periods, as in the
case of the displacement rates shown in Figure 5 (bottom): over the 9-month period
covered by these data there is a kind of “virtual displacement rate” between the two
sides of the building of about 40 mm/yr in the SAR line-of-sight (i.e. 69 mm/yr
horizontally), which, in fact, is due to the global increase of temperature during this
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
10
period (see the temperature plot in Figure 5, top right). As stated, this kind of systematic
effect should be negligible when analysing longer periods.
2.4 Deformation time series
The PS deformation time series (TS) represent the most interesting and advanced PSI
product. In fact, they provide the whole deformation history over the observed period,
with one estimate per each SAR acquisition, which is fundamental information for
many applications. However, this product has two important limitations: first it is
particularly sensitive to phase noise. In addition to that, its interpretation can be affected
by the (linear) deformation model assumption made in many PSI approaches. This
aspect is further discussed in the following section.
Considering the C-band PSI TSs, interesting results were achieved in the Terrafirma
Validation Project. In particular, by the inter-comparison of TSs from different teams
the following estimated standard deviation was found:
σ
TSERIES
= 1.1 - 4.0 mm.
It is worth noting that these values are mainly coming from PS with zero or very
moderate deformations: they probably degrade with stronger PS deformation
magnitudes. Other TS validation results are described in Crosetto et al. (2008b). In the
same article, interesting results were reported on the degree of similarity of the TS
patterns estimated, over common PSs, by different teams. This was measured by the
correlation coefficient between TS pairs. Considering large PS datasets coming from
ERS and Envisat data, this study reported a generally low degree of similarity, raising
doubts on the actual information contained in the analysed TSs. According to the
authors’ view, the information content of the C-band TSs has yet to be fully analysed.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
11
Even though excellent TS examples have been published in the literature, the global
performances of large TS datasets have not been fully understood.
A comparable comprehensive analysis has yet to be performed for the X-band PSI: the
authors can only report their initial experience with the TerraSAR-X PSI analysis.
Figure 1 shows some examples of TSs: two of them concern a deformation area, while
the other two refer to a stable area. One may clearly follow the entire deformation
history over the 9-month observed period through these TSs. This can be used to study
the causes of the deformation at hand. The two TSs that show no deformation during
this period show a very smooth pattern: considering that they are temporally unfiltered.
This indicates a good quality (low noise) of the observed PSI phases. The authors’ first
impression is that the X-band TSs show a remarkable quality improvement over the C-
band. For instance, in the analysed dataset the X-band TSs of basically stable PSs have
σ
TSERIES
ranging from 1 to 2.5 mm, while for the same type of PSs in C-band the
σ
TSERIES
ranges from 3 to 6 mm. Even though the X-band quality needs to be further
analysed, it opens promising new PSI applications. Finally, it is worth observing that
the above-mentioned thermal dilation affects the TSs, especially in the X-band data.
This has to be properly considered in the TS analyses. Future studies could be devoted
to this issue, proposing methods to properly model thermal dilation effects in the PSI
observations.
3. Discussion of main PSI limits
The most important advantages of PSI are well-understood and documented in the
literature. To the contrary, this can not be said for the main limits of PSI, whose
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
12
description is sometimes omitted in the literature. Some of them are briefly discussed
below.
• PS spatial sampling, as extensively discussed in a previous section, has an
opportunistic character. While in some areas, like cities, PSI gets a reasonably good
sampling, in others, like vegetated and forested areas, PSI tends to systematically
fail. This applies to both C- and X-band data. Provided that sampling capability is a
necessary condition for PSI, this is an important limitation, which has to be properly
considered while assessing the PSI feasibility over an area of interest or a given
deformation phenomenon.
• Observable deformation magnitude and aliasing. Due to the ambiguous nature of
PSI observations, which are 2π-wrapped, PSI suffers limitations in its capability to
measure “fast” deformation phenomena. It is difficult to quantify what “fast” means,
because the aliasing is related to a condition (phase difference < π, which
corresponds to a deformation < λ/4, where λ is the SAR wavelength), which applies
to PS pairs (i.e. it regards spatial phase differences) and to pairs of dates (i.e. it
regards temporal phase differences). That is, it basically depends on the spatial
pattern of the deformation, on its PS sampling density and on the temporal SAR
sampling. Things are further complicated by the fact that many PSI approaches
resolve the unwrapping phase by assuming a linear deformation model for
deformation (see also next bullet point). To the authors’ knowledge, this has
occurred in most of C-band PSI results published in recent years, which usually
experience difficulties in measuring deformation rates above 5-8 cm/yr. In case the
linear deformation model assumption is not made, as is the case of the X-band
results shown in Figure 1, the above condition is more straightforward: in order to
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
13
avoid aliasing, the deformation between neighbouring PS and consecutive SAR
acquisition dates cannot be greater than λ/4. In the best case with X-band this
condition is: up to 0.77 cm of displacement in 11 days between pairs of PSs (at full
resolution we can have PSs on a regular grid of 1.9 by 1.6 m). Any PSI feasibility
study must carefully consider this condition.
• Linear deformation model assumption. As stated in the previous bullet point,
many PSI approaches make use of a linear deformation model in their estimation
procedures, e.g., this occurred in practically all C-band PSI products of the
Terrafirma project (www.terrafirma.eu.com). The linear model assumption can have
a negative impact on the PSI estimates for all deformation phenomena characterized
by non-linear temporal deformation behaviour, i.e. where the assumption is not
valid. In areas where the deformation shows “significantly non-linear motion” the
PSI products may lack PSs, due to the fact that the PSI phases (i.e. the observations)
do not fit the (incorrect) linear model. This may represent a critical limitation
because PSI may be unable to provide deformation measurements precisely over the
most interesting deformation area.
• Deformation tilts or trends. PSI deformation rates may sometimes contain spatial
tilts or trends, which can be due either to uncompensated orbital errors or low
frequency residual atmospheric effects. The critical point is that, in some
applications, similar trends can be due to a real geophysical signal. Theoretically, it
is not possible to separate low-frequency geophysical deformation signals from the
above-mentioned residual effects with standard PSI processing: one of the following
opposite situations usually happens. Either a tilt results in the PSI products that can
be interpreted as geophysical signal, while in fact it is simply due to residual
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
14
processing errors, or a tilt-free product results, which is interpreted by a
geophysicist as no signal, e.g. quiescence of a given phenomenon, while in fact the
site may have undergone significant geophysical low-frequency deformations that
have been removed (together with the other residual effects) during the PSI
processing. Any PSI application focused on spatial low-frequency deformation
signals should assess the potential impact of the above limitation.
• Line-of-sight (LOS) measurement capability. The deformation measurements
coming from PSI refer to the LOS of the SAR sensor used. Given a generic 3D
deformation, PSI provides the estimate of the LOS component of this deformation
(i.e. the projection of the 3D deformation in the LOS direction). By using ascending
and descending SAR data, one can retrieve the vertical and approximately the east-
to-west horizontal components of deformation, while the polar orbiting SAR are
basically blind to south-to-north horizontal deformation components.
Conclusions
In this paper, the main characteristics of PSI deformation measurement and monitoring
have been outlined, and the most important PSI products and their performances have
been discussed. Two main types of PSI data have been considered: the C-band data
from ERS-1/2 and Envisat and the high resolution X-band data from the TerraSAR-X
sensor. Concerning the C-band PSI, some of the most important validation results from
the Terrafirma Validation Project have been discussed. They provide information on the
global behaviour of PSI average displacement rates, TSs, RTEs and PS geocoding. In
addition to this, an initial characterization of X-band PSI results has been provided by
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
15
analysing a TerraSAR-X 13-image dataset covering the city of Barcelona. Its most
relevant conclusions are summarized below.
• A straightforward advantage of high resolution X-band PSI is its capability to
generate dense PS sampling. Using C-band data, PS densities of up to 800-1000
PS/km
2
can be achieved, while the TerraSAR-X results from Figure 1 achieve about
39000 PS/km
2
. This dramatic increase could be partially attributable to the relatively
short observation period considered in this work (9 months): the PS density could
slightly decrease if longer periods were considered. However, the X-band PS
density is high and allows us to achieve a dense sampling of single buildings: this
opens promising application perspectives.
• The increase of the X-band sampling capability with respect to the C-band is
particularly strong over streets, highways and runways, again, opening new PSI
application fields.
• An initial X-band RTE result has been shown, which concerns two Barcelona
skyscrapers. The quality of both RTE and PS geocoding is very high: in the authors’
experience, comparable results have never been achieved using C-band PSI.
• An example of an average displacement rate map has been discussed, which
highlights the PSI capability to measure the deformation induced by construction
works over a narrow 200 by 25 m area.
• The high sensitivity of the X-band observations to small displacements has been
illustrated by showing displacement measurements associated with thermal dilation
of industrial buildings. Dilation effects have to be carefully considered in all PSI
analyses, especially those covering short observation periods.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
16
• The X-band TSs described in this paper show a remarkable quality improvement
with respect to the C-band. Even though further studies are needed, these initial
results suggest promising new PSI deformation monitoring applications.
• The thermal dilation clearly has an impact on the X-band TSs: this has to be
properly considered in the TS analyses. Future works could be devoted to this issue,
e.g. proposing methods to properly model the thermal dilation effects.
Last, the most important PSI limits have been discussed in this paper. They include the
PS spatial sampling capability, which can be very poor in vegetated and forested areas,
and the PSI limitation in measuring “fast” deformation phenomena. Both limitations
call for a careful assessment of PSI feasibility when starting any new deformation
analysis. Other relevant limitations are related to the linear deformation model
assumption made in many PSI approaches; the impossibility to separate low-frequency
geophysical deformation signals from the spurious processing effects (uncompensated
orbital errors and residual atmospheric effects); and the line-of-sight nature of PSI
deformation estimates.
References
Berardino, P., Fornaro, G., Lanari, R., Sansosti, E., 2002. A new algorithm for surface
deformation monitoring based on small baseline differential SAR interferograms,
IEEE Transactions on Geoscience and Remote Sensing, 40 (11):2375-2383.
Colesanti, C., Ferretti, A., Novali, F., Prati, C., Rocca, F., 2003. SAR monitoring of
progressive and seasonal ground deformation using the Permanent Scatterers
Technique, IEEE Transactions on Geoscience and Remote Sensing, 41(7):1685-1701.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
17
Crosetto, M., Crippa, B., Biescas, E., Monserrat, O., Agudo, M., Fernández, P., 2005.
Land deformation monitoring using SAR interferometry, State-of-the-art.
Photogrammetrie Fernerkundung Geoinformation, 6:497-510.
Crosetto, M., Biescas, E., Duro, J., Closa, J., Arnaud, A., 2008a. Quality assessment of
advanced interferometric products based on time series of ERS and Envisat SAR data,
Photogrammetric Engineering and Remote Sensing, 74(4): 443-450.
Crosetto, M., Monserrat, O., Adam, N., Parizzi, A., Bremmer, C., Dortland, S.,
Hanssen, R.F., van Leijen, F.J., 2008b. Final report of the Validation of existing
processing chains in Terrafirma stage 2, Terrafirma project, ESRIN/Contract no.
19366/05/I-E, available on-line at www.terrafirma.eu.com/Terrafirma_validation.htm.
Crosetto, M., Monserrat, O., Jungner, A., 2009. Ground-based synthetic aperture radar
deformation monitoring. 9
th
Conference on Optical 3-D Measurement Techniques,
July 1–3, 2009, Vienna, Austria.
Ferretti, A., Prati, C., Rocca, F., 2000. Nonlinear subsidence rate estimation using
permanent scatterers in differential SAR interferometry, IEEE Transactions on
Geoscience and Remote Sensing, 38(5):2202-2212.
Ferretti, A., Prati C., Rocca, F., 2001. Permanent scatterers in SAR interferometry,
IEEE Transactions on Geoscience and Remote Sensing, 39(1):8-20.
Hanssen, R.F., van Leijen, F.J., 2008. Monitoring water defense structures using radar
interferometry, Radar Conference, 2008. In: Radar '08. IEEE, Rome (Italy), 26-30
May 2008.
Hooper, A., Zebker, H., Segall, P., Kampes, B., 2004. A new method for measuring
deformation on volcanoes and other natural terrains using InSAR Persistent
Scatterers, Geophysical Research Letters, 31, L23611, doi: 10.1029/2004GL021737.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
18
Lanari, R., Mora, O., Manunta, M., Mallorquí, J.J., Berardino, P., Sansosti, E., 2004. A
small-baseline approach for investigating deformations on full-resolution differential
SAR interferograms, IEEE Transactions on Geosciences and Remote Sensing, 42(7):
1377-1386.
Luzi, G., Pieraccini, M., Mecatti, D., Noferini, L., Guidi, G., Moia, F., Atzeni, C., 2004.
Ground-based radar interferometry for landslides monitoring: Atmospheric and
instrumental decorrelation sources on experimental data, IEEE Transactions
Geoscience and Remote Sensing, 42(11): 2454–2466.
Kampes, B.M., Hanssen, R.F., 2004. Ambiguity Resolution for Permanent Scatterer
Interferometry, IEEE Transactions on Geoscience and Remote Sensing, 42(11): 2446-
2453.
Mora, O., Mallorquí, J.J., Broquetas, A., 2003. Linear and nonlinear terrain deformation
maps from a reduced set of interferometric SAR images, IEEE Transactions on
Geoscience and Remote Sensing, 41(10):2243-2253.
Pepe, A., Sansosti, E., Berardino, P., Lanari, R., 2005. On the Generation of
ERS/ENVISAT DInSAR Time-Series via the SBAS technique, IEEE Transactions on
Geoscience and Remote Sensing Letters, 2(3):265–269.
Rosen, P.A., Hensley, S., Joughin, I.R., 2000. Synthetic Aperture Radar Interferometry,
Proc. of the IEEE, 88 (3): 333-382.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
19
Figure 1: Average displacement rates over an urban area: Envisat C-band, from November 2007 to
January 2009 (top) and TerraSAR-X, from December 2007 to September 2008 (middle). The plot on the
bottom shows four TerraSAR-X TSs, whose PS location is indicated by two squares in the middle image.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
20
Figure 2: Example of the X-band PSI sampling capability over the Barcelona airport: multi-mean SAR
amplitude (left) and, in black, plot of the available PSs considering the period from December 2007 to
September 2008 (right).
Figure 3: Example of the X-band PSI sampling capability over Barcelona’s Montjuïc area, which includes
the Olympic Stadium, the Palau Sant Jordi and other forested and vegetated areas located in the bottom-
right part of the two images: multi-mean SAR amplitude (left) and, in black, plot of the available PSs
considering the period from December 2007 to September 2008 (right).
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
21
Figure 4: Colour-coded RTE estimated over the twin skyscrapers of the Barcelona’ Olympic Port. The RTE
show the estimated PS elevations with respect to a Digital Terrain Model from the Cartographic Institute of
Catalonia, which was used in the PSI processing. An accurate PS geocoding is fundamental to fully interpret
and exploit PSI results. This 3D view is based on Google Earth and its 3D building collection, @2009 Google.
Original manuscript submitted to PERS. The final published version is:
Crosetto, M., Monserrat, O., Iglesias, R. and Crippa, B., 2010. “Persistent Scatterer Interferometry: Potential, Limits and Initial C-
and X-band Comparison”. Photogrammetric Engineering and Remote Sensing. Vol. 76, N. 9, pp. 1061-1069.
22
Figure 5: Thermal dilation effects measured by TerraSAR-X data over an industrial area (see overview on
the top-left image). Mean temperature profile in correspondence to the 13 SAR acquisitions considered
over the December 2007 to September 2008 period (top-right). Two wrapped and colour-coded
interferograms (superposed to an optical image), which show opposite phase gradients (2
nd
and 3
rd
figures
from the top). The first interferogram corresponds to the 4
th
and 5
th
images of the SAR stack (∆T: +7 ºC),
while the other one to the 3
rd
and 4
th
images (∆T: -4 ºC). Estimated LOS displacement rates over the same
above period, caused by the global increase of temperature, see the top-right temperature profile (bottom
image).
