Sunteți pe pagina 1din 12

RADARGRAMMETRY AND INTERFEROMETRY SAR FOR DEM

GENERATION
Jung Hum Yu1, Xiaojing Li, Linlin Ge, and Hsing-Chung Chang
School of Surveying and Spatial Information Systems
University of New South Wales, Australia
+61-2-9385-4201, +61-2-9313-7493
1
Jung.yu@student.unsw.edu.au

Abstract
Digital elevation models (DEMs) can be generate by interferometric SAR
(InSAR) and radargrammetry techniques from different positions of Synthetic
Aperture Radar (SAR) images. Radar imaging systems record both the phase
(or time) and intensity information of the backscattered signals. InSAR utilises
the phase information of the images to extract useful geodetic information, such
as the height of terrain, ground deformation. However, InSAR technique is
constrained by the temporal and spatial separations between the images used,
as well as the various atmospheric conditions at the time of acquisitions. In
comparison, radargrammetry technique utilises the intensity information in a
stereo-pair of radar images. That is similar to stereogrammetry or
photogrammetry which is a classic method for relief reconstruction using
airborne/spaceborne optical images.
In this paper, the advantages and disadvantages of both InSAR and
radargrammetry for DEM generation are demonstrated using the
ALOS/PALSAR and Envisat/ASAR data.

Introduction
There are two major techniques for DEM generation using SAR data. One is
based on interferometric SAR (InSAR) and the other is based on
radargrammetry. In general, higher accuracy DEM can be generated through
using InSAR technique; therefore, InSAR is used when application.
InSAR method is relatively cost efficient and effective, also wide-coverage on
DEM generation. This technique involves interferometric phase comparison
from two SAR images acquired at different positions with a separation of
perpendicular baselines. It can generate DEMs at metre-level accuracy. InSAR
extracts information of the terrain from the phase difference “interferogram” of
two SAR data images (Ferretti et al. 2001). However, InSAR DEM generation is
subject to decorrelation, atmospheric disturbance and the conditions on
incidence angle and Doppler similarity are stringent (Massonnet and Souyris
2008). InSAR need the expectations of a certain range baseline because the
interferometry is sensitive to the direction of sensor movement and some other
factors. To overcome those limitations, the radargrammetry technique is then an
important alternative for DEM generation (Chen and Dowman 2001). The main
difference of two techniques is that radargrammetry calculates the image range
offset using the position matching of the same ground targets in two images
while InSAR calculate the phase difference of two images (Kyaruzi 2005;
Sansosti 2004).
Radargrammetry is based on stereogrammetry which is a classic method for
relief reconstruction using optical remote sensing images. This technique can
be applied to radar images for generating good quality DEMs (Paillou and
Gelautz 1999). One of the advantages of radargrammetry is less affected by
atmospheric influence compare with interferometry. Basically, atmospheric
effect on the SAR imagery is same in the radargrammetry or in the InSAR.
However, radargrammetry uses the magnitude (intensity) value while InSAR
uses the phase difference in SAR imagery. Considerably, magnitude is less
affected than phase by atmospheric heterogeneous. The atmospheric
disturbance is undesirable for the InSAR processing but not much of a concern
for the radargrammetry processing (Massonnet and Souyris 2008). But the
radargrammetry uses stereoscopic pairs acquired from different incidence
angles. Also, the radargrammetry is different with photogrammetry mainly in
three aspects (Schanda 1985): (1) the interaction effects with the surface at
radar wavelengths are different from those at optical wavelengths, (2) radar
measures distance between sensor to target, therefore the parallax appears
reversed when compared with optical image, (3) the long wavelengths cause
poor angular resolution at a given size of ‘optics’ therefore a useful stereo base
cannot be established simultaneously from one simple platform.
Furthermore, stereoscopic pairs for radargrammetry should be considered
that the geometry and parallax produced from a particular system configuration
and the awareness of the image pairs by the interpreter. Obviously, the
radargrammetry technique for terrain elevation extraction requires the
conditions of have approximately 10~20 degrees incidence angle difference and
overlapping of two images between input image pairs (Mercer 1995; kaupp et al.
1983).
In this paper, the authors discuss and demonstrate the advantages and
disadvantages of both InSAR and radargrammetry using the real satellite data.
Radargrammetric DEMs generated using the stereo-image pairs with various
look-angles, baselines, ascending and descending orbits are examined closely.

Methodology

DEM generation using Radargrammetry


Radargrammetry requires image acquisitions with varying incidence angles.
The quality of radargrammetry DEM depends on the base to height ratio or
intersection angle of the radargrammetry pair demonstrated in Figure 1. To
acquire good geometry for radargrammetry pairs, the intersection angle
between the two SAR images should have enough angles for the observed
parallax which is used to determine the terrain elevation. However, in order to
have good stereo-viewing, the nearly identical images (small intersection angle)
are necessary in processing. Approximately 10-20 degree intersection angles
between the two images and shallow look angle (i.e. angle between vertical and
the bean direction>20˚) are usually considered on optimal configurations for
medium to high relief areas (d’Ozouville et al. 2008). Thus, a compromise has
to be reached between a better stereo-viewing and more accurate elevation
determination (Toutin and Gray 2000).

z
S1
S2
θ1 Bx θ2
H Bz

R2
R1
y

x2 P
P2
x1 h
dp
P1
x
Figure 1. The different observation positions and geometry for radargrammetry (Maitre,
2008)
Where in Figure 1, S1, S2 are the satellites, Bx, Bz is the horizontal and vertical
baseline, R1,R2 are the distances between the sensors and ground target P.
The target P is seen as P1 and P2 in both SAR images from S1and S2. Then, dp
is called ‘disparity’ distance of P1 and P2. If the ground elevation is zero (h=0),
disparity will be zero and it increments for increasing heights h. It is expressed
by:
dp= x2 + (H − h)2 − H2 − (x − Bx )2 + (H + Bz − h)2 − (H + Bz )2 − Bx (1)
More importantly, the elevation calculation has the reverse relationship which
gives elevation h for the disparity point dp.
2HBx + 2Hdp − 4H 2 Bx2 + dpΛ
h= (2)
dp + Bx
there
Λ = 8Bx ( H 2 − x 2 + xBx ) + dp(4Bx2 + dp2 + 4dpBx ) + 4dp( H 2 − x 2 + xBx )
(3)
These equations are simplified further as the images are acquired from
satellites’ sensors (sensor elevation, H, is significant compared to target
altitudes, h). Therefore, the equation can be a simple expression in accordance
with look-angles of θ1 and θ 2 :
dp
h= (4)
cotan θ 2 − cotan θ1
Figure 2 shows the processing steps of radargrammetry for DEM generation.
The radargrammetry DEM processing steps can be described in terms: (1)
acquiring stereoscopic images; (2) subset in the areas of interest; (3) despeckle
to remove the noise; (4) co-registration of two subset images; (5) matching
between co-registered images; (6) height calculation; (7) geocoding and DEM
generation.
Reference image Match image

Subset image Subset image

Despeckle Despeckle

Co-registration

Matching

Height calculation

DEM generation
Figure 2. The flow-chart of radargrammetry processing

DEM generation using InSAR

The phase differences of radar wavelengths recorded by SAR images are able
to provide precise measurement, to the sub-wavelength level, of the range or
distance between the location of the observing antennas and the points of
reference (‘pixel centres’). With two observing antennas, or the same antennas
at different satellites overpass times, differential phases can provide information
on terrain elevation or on terrain displacement (Toutin & Gray 2000).
In the InSAR processing, it is crucial that the imaging geometry of the first path
is repeated as closely as possible in the second path. In other words, the
location difference (or ‘perpendicular baseline’) must be kept as short as
possible - typically less than a kilometre. The reason for this condition is due to
the geometric relationships utilised in the interferometric processing of images.
However, in the DEM generation, the perpendicular baseline prefers long
baseline images for increase the terrain sensitivity. SAR images have a spatial
resolution defined by the pulse length, altitude of the satellite and look-angle.
The slant range resolution is half the pulse length. The ground range resolution
is defined as the slant range resolution divided by the cosine of the look-angle.
The set of phase differences produced for all pixels of the two SAR images is
used to generate an interferogram. The SAR interferometry method allows to
generation of DEM, as well as the detection and measurement of ground
deformation.
In producing an InSAR DEM, the difference in the phase of at least two SAR
images of the same area identifies the interferometric phase contribution due to
the terrain:

φ = ( R1 − R 2 ) (5)
λ
where λ is the radar wavelength (Crosetto 2002; Zhou et al. 2005).
Interferometry
A2

Bs
A1

R2
R1
H
ΔθI

P
Azimuth h

Range

Figure 3. Geometry of Interferometric SAR (InSAR)


In Figure 3, H is the altitude of the imaging satellite, R1, R2 are the distance
between the target P on the ground and satellite antennas A1 and A2
respectively. Bs is the distance between the two antennas A1 and A2 at the
satellites’ imaging locations.
The elevation information is obtained from the interferogram by “unwrapping”
the interferometric phases (e.g. Ferretti et al. 1999).
The interferogram has three folds of information: 1) topographic information, 2)
surface displacement that has taken place between the two SAR image
acquisitions, and 3) atmospheric delay and noise. The atmospheric effect is
primarily due to the water vapour content in the atmosphere between the
satellite radar sensor and the ground target.
The relation in phases therefore is:
φInSAR=φflat +φtopo +φdefo +φatm +φnoi (6)
where φ InSAR is the interferometric phase, φ flat is the so-call “flat earth” phase,
φ topo is the topographic phase, φ defo is the deformation phase, φ atm is the
atmospheric delay phase and φ noi is the noise. The atmospheric delay
component φatm can be identified using the fact that its phase difference patterns
are independent over several interferograms, or, alternatively, it can be
modelled by using a GPS ground network to independently determine the
atmospheric delay, or by a technique known as ‘interferogram stacking’.
In equation (6), the “flat earth” phase φ flat and noise φ noi can be removed by
using the orbit information correction and applying an interferogram filtering
method.
When the imaging interval is sufficiently short it may be assumed that there is
no deformation phase φ defo . InSAR is the process to extract the topographic
phases while to eliminate other undesired phase components. If the
atmospheric delay phase φatm can be ignored (or determined from other sources),
then, the equation (6) can be reduced to:
φInSAR = φtopo (7)
The DEM is obtained by “unwrapping” the phase φ , then converting the phase
to a height for that pixel, and then geo-coding each pixel in turn. The attention
also is, the pixel coordinates and parameters of the InSAR DEMs are
dependent on the parameters of the master image (Ferretti et al. 2001).

Master SLC Slave SLC

Co-registration

Interferogram

Coherence

Denoise Interferogram

Phase Unwrapping

Phase to Height

Geocoded Products

Figure 4. The flow chart of InSAR DEM generation


The unwrapped phases have to be converted in terrain heights. Phase to
height conversion is the procedure which relates the unwrapped phase to
topographic height.

Experimental Result

The test site has been set at the Appin area in the state of New South Wales,
Australia. This research used the Advanced Land Observing Satellite
(ALOS/PALSAR) SAR image (L-band) and European Remote Sensing Satellite
(ERS-1) SAR image (C-band) to InSAR generated DEMs. Table 1 presents the
information of SAR data. In order to improve the coherence of image pairs,
short temporal baseline is selected for InSAR DEM generation. The Shuttle
Radar Topography Mission (SRTM) DEM is used as the external DEM.
Table 1. The information of InSAR processing images

Sensor Master date Slave date Bperp (m) Btemp (days)


ALOS 27/12/06 11/02/07 549 46
ALOS 14/02/08 31/03/08 659 46
ERS 29/10/95 30/10/95 49 1
ERS 03/12/95 04/12/95 145 1
In radargrammetry DEM processing, four pairs of ALOS/PALSAR and four
pairs of ENVISAT/ASAR images were used. The characters of radargrammetry
allow larger coverage processing than interferometry method. Images were
acquired from different path orbits which have different incidence angle. Table 2
is the image information used in radargrammetry. The same-side stereo method
was used in radargrammetric DEM generation.
Table 2. The information of radargrammetry images
Incidence Incidence
Sensor Reference image Match image angle_Reference angle_Match

ALOS 31/03/08 (370) 05/04/08 (373) 38.7˚ 47.3˚


ALOS 07/04/08 (365) 05/04/08 (373) 24.0˚ 47.3˚
ALOS 23/05/08 (365) 21/05/08 (373) 24.0˚ 47.3˚
ALOS 01/07/08 (370) 06/07/08 (373) 38.7˚ 47.3˚
ENVISAT 18/12/09 (152) 26/09/09 (467) 33.6˚ 43.8˚
ENVISAT 02/04/10 (152) 31/10/09 (467) 33.6˚ 43.8˚
ENVISAT 15/03/10 (402) 12/03/10 (359) 22.9˚ 33.6˚
ENVISAT 08/02/10 (402) 05/02/10 (359) 22.9˚ 33.6˚

Figure 5. The radargrammetry intensity images of 31/03/2008-05/04/2008


ALOS/PALSAR pair (31/03/2008: reference image-left, 05/04/2008: matching image-
right).

Table 3. The information of intensity average processing images


Track (angle) Date
338 (18.9˚) 02/10/2008 06/11/2008 11/12/2008
381 (28.7˚) 05/10/2008 09/11/2008 14/12/2008
152 (33.6˚) 24/10/2008 28/11/2008 19/09/2008

Table 3 lists the data information which was used to re-generated intensity
SAR images using averaging method. Figure 5 shows the intensity images for
radargrammetry technique. Reference image has a incidence angle of 38.7˚
and matching image has incidence angle of 47.3˚ from satellite sensor. The
terrain shapes are appeared different appearances such as distance and width
difference between rivers and size of glass area. These phenomenons generate
the stereoscopy of radargrammetry. The reference and match images were
registered for pixel offset calculation.
Automatic matching method calculates the correlation between two images
for generate a disparity map. Matching can be performed using initial gray-level
images, edge images, or some other image features such as linked-edge
elements or regions. The main problems encountered when matching radar
images for radargrammetry are speckle noise and that the difference of two
stereo partners from one another, as the backscattered radar signal mainly
depends on the local incidence angle (Paillou and Gelautz 1999; Tountin and
Gray 2000).

Figure 6. The correlation image between reference image and match image after
matching process

Figure 6 shows the correlations between reference and match. The


correlation number gives a measure of how strong a match exists between the
reference and match points. The correlation scale ranges from 0 to 1, where the
closer the result is to 1, the better the match. The higher correlations are
appeared in city area like the Appin and Wollongong and surround the rivers.
Figure 7 shows the InSAR generated DEM results using ALOS/PALSAR
pairs in test area. Same parts of DEM have elevation errors due to the ground
deformation between different acquisitions and due to the low coherence of two
images. Figure 8 shows the InSAR generated DEMs result using ERS pairs in
test area. ERS images provide the larger coverage and shorter data processing
time than PALSAR images.
Figure 9 and 10 show the radargrammetry DEM results using Envisat /ASAR
pairs in test area. In the DEM results, the elevation errors were appeared in
near shoreline and ocean area. Figure 11 illustrates the radargrammetry DEM
results using ALOS/PALSAR pairs in test area. Figure 12 presents the
radargrammetry DEMs using Envisat intensity average image which was
processed by three different acquisition time.
Figure 7. The InSAR DEM generated from ALOS/PALSAR (left:
14/02/2008~31/03/2008, right: 27/12/2009~11/02/2007)

Figure 8. The InSAR DEM generated from ERS (left: 03/12/1995~04/12/1995, right:
29/10/1995~30/10/1995)

Figure 9. Radargrammetry DEM generated from Envisat /ASAR (left:


08/02/2010~05/02/2010, right: 15/03/2010~12/03/2010)
Figure 10. Radargrammetry DEM generated from Envisat /ASAR (left:
02/04/2010~31/10/2009, right: 18/12/2009~26/09/2009)

Figure 11. Radargrammetry DEM generated from ALOS/PALSAR (left:


31/03/2008~05/04/2008, right: 01/07/2008~06/07/2008)

Figure 12. Re-generated radargrammetry DEM from Envisa intensity average


images (left: track 338 and 152, right: track 338 and 381)

Concluding remarks
,Two types of wavelength SAR imageries and two DEM generation methods
were used for DEM generation in this paper.
The DEM generation using radar imageries is operated by InSAR technique
for high accuracy products generation and the radargrammetry is the
superseded technique of InSAR for DEM generation method. Radargrammetry
is comparable to photogrammetry in that a stereo parallax found in a stereo
image pair corresponds to terrain elevation.
The advantage of the radargrammetry technique for DEM generation is that it
can provide larger coverage per processing time and software capability than
InSAR processing. However, the main problem of radargrammetry is that the
DEM products have low level quality of DEM due to the spatial resolution of
SAR imageries and the terrain slope. The disparity and convergence of objects
are the two cues when viewing stereo imagery. Disparity is predominating when
viewing radar image, but the shade and shadow cues also have a strong and
cumulative effect on accuracy. Especially, the range direction errors in
radargrammetry are higher than azimuth direction errors. Furthermore, smaller
intersection angle provides the more accurate DEM products.

References

Chen, P.H., and Dowman, I.J., 2001, A weighted least squares solution for
space intersection of spaceborne stereo SAR data. IEEE transactions on
geoscience and remote sensing, 39(2): 233-240.

Crosetto, M., 2002, Calibration and validation of SAR interferometry for DEM
generation. ISPRS Journal of Photogrammetry & Remote Sensing, 57(3): 213-
227.

d’Ozouville, N., Deffontaines, B., Benveniste, J., Wegmuller, U., Violette, S.,
and Marsily, G., 2008, DEM generation using ASAR (ENVISAT) for addressing
the lack of freshwater ecosystems management, Santa Cruz Island, Galapagos.
Remote Sensing of Environment, 112(11): 4131-4147.

Ferretti, A., Prati, C., and Rocca, F., 1999, Multibaseline InSAR DEM
reconstruction: The wavelet approach. IEEE Transactions on Geoscience &
Remote Sensing, 37(2): 705-715.

Ferretti, A., Prati, C., and Rocca, F., 2001, Permanent scatterers in SAR
interferometry. IEEE Transactions on Geoscience & Remote Sensing, 39(1): 8-
20.

Kaupp, V. H., Bridges, L. C., Pisaruck, M. A., MacDonald, H. C., and Waite, W.
P., 1983, Simulation of spaceborne stereo radar imagery: Experimental result.
IEEE transaction of geoscience and remote sensing, GE-21(3): 400-405.

Kyaruzi, J. K, 2005, Quality assessment of DEM from radargrammetry data.


International institute for geo-information science and Earth observation
enschede, The Netherlands, master of science.

Maitre, H., 2008, Processing of synthetic aperture radar images. ISTE Ltd and
John Wiley & Sons Inc.
Massonnet, D., and Souyris, J. C., 2008, Image with synthetic aperture radar.
EPFL Press, 189-191.

Mercer, J. B., 1995, SAR technologies for topographic mapping.


Photogrammetric Week ‘95’, 117-126.

Paillou, P., and Gelautz, M., 1999, Relief reconstruction from SAR stereo pairs:
the “Optimal Gradient” matching method. IEEE transactions on Geoscience and
Remote Sensing. 37(4): 2099-2107.

Sansosti, E., 2004, A simple and exact solution for the interferometric and
stereo SAR geolocation problem. IEEE transactions on Geoscience and
Remote Sensing, 42(8): 1625-1634.

Schanda, E., 1985, A radargrammetry experiment in a mountain region.


International Journal of Remote Sensing, 6(7): 1113-1124.

Toutin, T., and Gray, L., 2000, State-of-the-art of elevation extraction from
satellite SAR data. ISPRS Journal of Photogrammetry & Remote Sensing, 55:
13-33.

Zhou, C., Ge, L., E, D., and Chang, H.C., 2005, A case study of using external
DEM in InSAR DEM generation. Geo-Spatial Information Science, 8(1): 14-18.

S-ar putea să vă placă și