You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

International cooperation and technology transfer
Mussio, Luigi

M. Crosetto ( \ B. Crippa (t)
( ] DIIAR - Sez. Rilevamento - Politecnico di Milano, 20133 Milano, Italy
(+) DRP - Università degli Studi di Messina, 80100 Messina, Italy
Commission VI, Working Group 3
KEYWORD: Interferometric SAR, DEM generation, InSAR geometry, calibration.
The generation of digital elevation models (DEM) with interferometric techniques has become one of the major potential
applications of SAR imagery. This paper is focused on the geometric aspects of the DEM generation with interferometric
SAR (InSAR). The different parameters that describe the satellite orbits, the image acquisition geometry and the SAR
processing are analysed. The accurate geolocation of the InSAR DEMs can be only obtained through the refinement of
such parameters. A new procedure for the calibration of the InSAR geometry based on the measure of ground control
points (GCP) is introduced. The results obtained on a test site covering South Catalonia (Spain) are discussed.
The complex SAR images contain both the amplitude and
the phase of the radar signal. When used in the
interferometric way, the phase brings a valued information
that greatly expands the potentialities of the SAR. The
main products of InSAR include the DEMs; the Coherence
Maps, employed in land-use classification; and the
Displacement Maps used to monitor landslides,
subsidence areas, etc. The SAR images can be acquired
either by airborne or spaceborne sensors. A general
review of the technique is given in (Gens and Van
Genderen, 1996). In this paper only the generation of
DEMs using data coming from spaceborne sensors
(repeat-pass SAR interferometry) is considered.
The InSAR DEM generation is based on the processing of
at least two complex SAR images. In order to generate a
DEM, the SAR data have to undergo several processing
stages among which there are the image registration, the
interferogram calculation and filtering, the phase
unwrapping (i.e. the reconstruction of the full
interferometric phase value from its principal one, which is
only known module n) and the phase-to-height
In the InSAR procedure we implemented (see Crippa et
al., (1998) for more details) the first processing stages
(image registration, interferogram generation and the
coherence calculation) are based on the ISAR-
Interferogram Generator software (distributed, free of
charges, by ESA-ESRIN), an effective tool to obtain good
filtered interferograms and the related coherence images
(Koskinen, 1995). The filtered interferograms are
unwrapped using the “branch cuts” approach (Goldstein et
al., 1988). The phase-to-height transformation (calibration
of the InSAR geometry and the generation of the DEM) is
described in detail in the following sections.
The generation of the InSAR DEM is performed using the
unwrapped phases, the orbital parameters of the two
images and few other sensor parameters (available in the
image auxiliary data) that describe the imaging geometry.
The transformation from unwrapped phases to heights is
realised with a very simple procedure that works point
wise and generates an irregular grid of 3D points (i.e.
already geocoded). This procedure differs from those
usually employed in InSAR, based on approximate
transformations from phases to heights, followed by a
geocoding (transformation from image space to object
space). Very often this geocoding is similar to those used
for the amplitude SAR images, i.e. it requires a known
DEM. The proposed procedure does not require any a
priori known DEM and it is quite flexible to allow the fusion
of data coming from different sources (e.g. ascending and
descending SAR, SAR and SPOT).
For each pixel of the interferogram (i.e. for each
unwrapped phase), starting from the azimuth coordinates
of the two images (the master one has the same
geometry of the interferogram, the slave azimuth can be
easily derived from the image registration transformation),
the relative acquisition times T are derived using:
T =T 0 + AT (lin-1)
where To is the acquisition time of the first image line, AT
is the pixel size in azimuth direction and lin is the azimuth
coordinate. The acquisition times known, positions and
velocities of the master M and slave S satellites are
calculated (see Figure 1).
Assuming the master M, the slave S and the unknown
point P to lie in the same plane (the Doppler centroid
plane or antenna mid-plane that goes through M), the
position of S that fulfil the equation (1) is found. Then,
using the range equation (2), the interferometric equation
(3) and the Doppler centroid equation (4) the position
(xp,yp,zp) of the pixel footprint P is estimated.
Repeating the procedure for all the pixels of the
unwrapped interferogram, an irregular grid of 3D points is
generated. The points are known in a geocentric
Cartesian system (the same used for the orbits), thus a
transformation to a cartographic projection and to
orthometric heights is performed. Finally does the
resampling to get the final regular geocoded grid follow.