Bachelorarbeit, 2012
42 Seiten, Note: 1,3
1 Introduction
1.1 Motivation: The 4th Dimension
1.2 Purpose and Structure
1.3 About TerraSAR-X
1.3.1 TerraSAR-X Scanning Modes
2 Theoretical Basics - SAR Principles
2.1 SAR Geometry
2.2 SAR Phase
2.3 SAR Properties
2.3.1 Layover
2.3.2 Chirp
2.3.3 Speckle
2.4 SAR Image Resolution
2.5 InSAR Basics
2.5.1 Interferometric Phase
2.5.2 InSAR Athmospheric Influences
2.5.3 InSAR Height Retrieval
2.6 Persistent Scatterer Interferometry (PSI)
2.7 SAR Tomography
3 TomoSAR
3.1 TomoSAR Processing Procedures
3.2 Pre-Processing
3.3 Processing
3.3.1 TomoSAR System Model
3.3.2 TomoSAR Algorithm
4 Application on Berlin Data Stack
4.1 Downsampling
4.2 Quality Measurements
4.3 Creating Pixel Pairs
4.4 Integration
4.5 Filter Residual Phase
4.6 Upsamling and Estimation
5 Results and Analysis
5.1 Deformations and Movements
5.1.1 Seasonal Movements
5.1.2 Linear Movements
5.2 Verification and Observation of the processed SAR Data Stack
5.2.1 TomoSAR Result Observation
5.2.2 Virtual Inspection of Observation .
5.3 Inspected Targets
5.3.1 ”Überflieger Brücke”
5.3.2 ”Berlin Hauptbahnhof”
6 Conclusion
First of all, I want to sincerely thank Dr.-Ing. XIAOXIANG ZHU and Prof. Dr. RICHARD BAMLER for giving me the opportunity to carry out my Bachelor thesis at the Department of Remote Sensing Technology at the Technical University Munich in cooperation with the German Aerospace Center / Deutsches Zentrum für Luft- und Raumfahrt e.V. (DLR) and to learn different and new aspects of the interesting field of remote sensing in geodesy.
Furthermore, I would like to thank YUANYUAN WANG for being the second examiner of my Bachelor thesis. Also most thanks to YUANYUAN for giving confidence in my work, for his technical expertise and the friendly atmosphere.
I also want to express my gratitude towards the whole team of the DLR for providing me with such an unsurpassable friendly and productive working atmosphere.
And finally I am very grateful for my parents, my sister, my companion and my friends for their encouragement, support and inspiration.
Driven by military and civilian applications, the demand of very high resolution mapping and accurate monitoring has increased rapidly over the recent years. Nowadays, it is possible to create 4D models involving time variations using multiple synthetic aperture radar (SAR) images, combined with interferometric methods. SAR has evolved to sat- isfy a variety of applications for civilian and military users, for example by supporting catastrophe management, detection of geological changes, monitoring large construction sites or mines. With the help of SAR data obtained from the TerraSAR-X satellite, in- frastructural monitoring is made possible from a distance. The benefit of this is that potential collapse within mines or tunnels could be prevented. Concrete degradation that could lead to building collapse, endangering people’s lives can also be identified before any catastrophe has the chance to occur.
Currently, Tomographic SAR (TomoSAR) is the most advanced and competent interferometric SAR (InSAR) method in the area of urban monitoring. TomoSAR makes monitoring in 4D possible by creating the 3D position with the motion parameters. This thesis applies a new TomoSAR technique and method, developed by ZHU and her colleagues, 20121, on a very high resolution (VHR) spotlight data stack in the area of Berlin. The images were taken by the TerraSAR-X satellite (Germany) over a timeframe of 3 years. The result is a 3D point cloud of the observed area, with the velocity of linear motion and the amplitude of periodic motion.
The result of the work that forms the basis for this thesis, is the realization of high deformation and motion in Berlin’s infrastructure, especially around Berlin’s main station, on bridges (”Überflieger Brücke”) and railways - often up to 10 mm.
Durch fortschreitende zivile oder militärische Anwendungen ist in den letzten Jahren die Nachfrage an hoch aufgelösten Karten und genauen Beobachtungen rapide angestiegen. Heutzutage ist es möglich, mittels interferometrischen Methoden und Bildern von Radar mit synthetischer Aperatur (SAR) ein 4D Modell mit zeitlichen Variationen zu erstellen. SAR hat sich sowohl für militärische als auch für zivile Zwecke für sehr hilfreich er- wiesen: zum Beispiel bei der Unterstützung von Katastrophenmanagements, Ermittlung von Geologischen Bewegungen sowie bei der Beobachtung von großen Baustellen oder Minen. Mit Hilfe der vom TerraSAR-X Satelliten gewonnenen Daten ist infrastrukturelle Überwachung aus der Ferne möglich. Der Vorteil dessen ist dass möglicher Zusammenfall von Minen oder Tunneln vorgebeugt werden kann. Betondegradation, was zum Einstürzen von Gebäuden führen und damit Menschenleben gefährden könnte, kann somit identifiziert werden bevor Katastrophen passieren.
Momentan ist das tomographische Radarverfahren mit synthetischer Aperatur (TomoSAR), das fortschrittlichste und kompetenteste aller interferometrischen Radarverfahren mit synthetischer Aperatur (InSAR) für städtische Beobachtungen. Mit TomoSAR ist die Beobachtung in 4D möglich, das heißt die 3D Koordinaten und Bewegungen eines Boden- punktes können beobachtet werden. In dieser Bachelorarbeit wird eine neue TomoSAR Methode, entwickelt von ZHU und ihren Kollegen, 20121, auf hoch aufgelöste Spotlight Daten angewandt. Die Bilder von der Innenstadt Berlin wurden über einen Zeitraum von 3 Jahren vom deutschen TerraSAR-X Satelliten aufgenommen. Das Resultat zeigt die Stadt Berlin in einer 3D Punktwolke mit jeweils den linearen sowie periodischen Bewegungen. Man kommt zur Erkenntnis, dass die Überflieger Brücke im Stadtteil Wedding, die Eisenbahn Infrastruktur und besonders der Berliner Hauptbahnhof hohe Deformationen aufwiesen, bis zu10mm.
1 Image of the collapsed Cologne city archive in Germany 2009 -[ Source:
www.topnews.de ]
2 Ground deformations causing gaps to open in modern infrastructure [ Source:
http://german.china.org.cn/ ]
3 An illustration of the German radar satellite TerraSAR-X - [ :Source: DLR ]
4 SAR image : Signal intensity from the observed area
5 SAR image : Phase of the signal from observered area (mirrored)
6 SAR Geometry2
7 A sinusoidal function for a transmitted SAR signal3
8 NANNINI’s illustration of a SAR height acquisition: ”The projection of
a 3D scene reflectivity into a two-dimensional plane for a fixed azimuth
coordinate”4
9 ”Residual Topography” RT and ” Deformation Value Map” DV of the
Berlin scene
10 Quality measurement methods
11 Network of pixels. The best pixels have been triangulated (for reasons of
clarity and comprehensibility this figure contains only every 12th pixel pair)
12 Temporal Coherence - multi-temporal coherency properties
13 Full Elevation of the cropped Berlin area, where significant objects can be
seen
14 The area around Berlin main station. Source: Google Earth/ Buildings
2007 3D Realitymaps / DLR taken
15 The area of Berlin main station - processed area overview. [Source: Google
Earth taken 2012]
16 TomoSAR point cloud Berlin - Height measurement in m
17 TomoSAR point cloud Berlin - Change in Height/Deformation in mm. .
18 Visible object/building deformation and moving towards and away from
sensor
19 Locating buildings and remarkable targets using Google Earth
20 Differences between 2009 and 2011 of the Überflieger Railway bridge, cap-
tured by photos
21 Inspecting the Point Cloud from the north
22 Clear change (blue pixels) of height near the buildings can be seen
23 Image from Berlin main station taken from Google Earth and point cloud
The Earth is constantly and continuously moving. In many countries around the world, the fast expansion of urban areas is related to the significant expansion of construction areas and infrastructure. The vast amount of buildings and their degradation by nature could lead to building collapse and affirms the need for dynamic models (4D) in a dynamic world. For instance some latest catastrophes like the collaps of the Cologne city archive in Germany 2009 (Fig. 1), and the ground displacement near Shanghai Tower in China (Fig. 2). With the establishment of very high resolution sensors, the remote sensing technology, for example Synthetic Aperture Radar (SAR), helps to monitor urban infrastructure movements to prevent collapses.
Abbildung in dieser Leseprobe nicht enthalten
Figure 1: Image of the collapsed Cologne city archive in Germany 2009 -[ Source: www.topnews.de ]
There are many ways to monitor the urban area, but in this case we use SAR, because of the numerous benefits. SAR is an active remote sensing technique, providing images by scanning the earth’s surface with electromagnetic waves (microwaves) thereby generating images covering a large area. Unlike optical or laser devices, SAR is almost operational in every weather condition, since atmospheric absorption is very little on microwave bands5 6. However, with the slanted imaging geometry, SAR faces the ”layover” problems
Abbildung in dieser Leseprobe nicht enthalten
(a) A sealed up gap in the ground
Abbildung in dieser Leseprobe nicht enthalten
(b) Gap in the ground caused by ground movements caused by ground movements - near - near the Shanghai Tower in the Pudong disctrict the Shanghai Tower in the Pudong disctrict
Figure 2: Ground deformations causing gaps to open in modern infrastructure [ Source: http://german.china.org.cn/ ]
in urban areas with buildings7. Layover results in a pixel in a SAR image containing both the information from the ground and the building facade. Fortunately, using the technique called SAR tomography (TomoSAR), these ambiguities can be solved8. Using differential Interferometry, even very little building and ground movements can be measured and monitored to prevent collapses or other fatal accidents. High resolution SAR data can be used to create 4D-Models of areas, a 3D-Model including the time- dimension, therefore monitoring areas and building movement with Persistant Scatterer Interferometry (PSI), a method invented in 20019. Since the PSI method expects that every pixel only has one dominant point, TomoSAR can solve multiple scatterers and the layover-effect, which makes it a better observation method for buildings. With TerraSAR-X having a resolution close to 1m x 1m6 and with up to 1 mio PS/ km2, it enables new possibilites in the civilian and military uses of remote sensing. This thesis aims to experiment with and exploit the potential of very high resolution (VHR) SAR in urban areas with TomoSAR using TerraSAR-X high resolution spotlight data.
The main purpose of this bachelor thesis is to gather samples and analyse the ground movements of the urban area using SAR images and TomoSAR techniques. Using satellite images means that a much larger surveillance area can be chosen compared to conventional ground geodesy measurements. The area of interest in this thesis is the area around the Berlin main station and the Berlin zoo as shown in figure 15.
This thesis is divided into six chapters. SAR principles and fundamentals are discussed in chapter two and three. Chapter four covers the procedures of generating a 4D city model from TerraSAR-X data. The application on the Berlin data stack is described in chapter five, while the six chapter inspects and analyses the results. Finally, the work is conclued in chapter seven.
On June 15, 2007, the German observation satellite TerraSAR-X was successfully launched 514 km high into a sun-synchronous orbit with a 97 . 44 ◦ inclination. It circles the earth 15.2 times a day, which makes it possible to reach any position on earth every 11 days10. The satellite works in X-Band at a frequency of 9.65 GHz (3.1 cm wavelength) and a maximum bandwith of 300 MHz.
Abbildung in dieser Leseprobe nicht enthalten
Figure 3: An illustration of the German radar satellite TerraSAR-X - [ :Source: DLR ]
TerraSAR-X has three different scanning modes11 7:
- High Resolution Spotlight Mode
- 5km swath
- 1m resolution
- Stripmap Mode
- 30km swath
- 3m resolution
- ScanSAR Mode
- 100km swath
- 15m resolution
First experimental data acquisitions with new modes have been successfully tested. The unique design of the TerraSAR-X antenna allows measurements as well as fully polarimetric data acquisitions11.
With the new scanning modes and its active phase array antenna, the TerraSAR-X acquires new high-quality radar images of the entire planet. With up to a resolution of 1 m, excellent radiometric accuracy and a unique agility (rapid switching between imaging modes possible), the TerraSAR-X is a real innovation.
Originally, the word ”radar” was an acronym for ”radio detection and ranging”. Now radar has been established as a standard English noun, since the radar-used application spread around the world12. A radar satellite sends, at a known time, a directed microwave signal, which is reflected by the objects on the earth’s surface7. The signals are timed with an interval, so the same antenna can process both sent and reflected signals7. Therefore, the range (line-of-sight direction) resolution depends on how narrow the time interval is. This corresponds to the bandwidth of the transmitted signal. The azimuth (along flight direction) resolution of a radar is inversely proportional to the size of the antenna, the resolution is therefore very limited (as explained in section 2.4), SAR is being applied. The idea behind SAR is: by creating a lot of images with a small moving antenna, a big (synthetic) antenna can be simulated. As long as the direction and position of the smaller real antenna is known and the target is static, a high spatial resolution in azimuth direction can be achieved.
As a result, by transmitting a signal with high bandwidth, and simulating a synthetic aperture, an SAR image with high spatial resolution can be obtained. Figure 4 and 5 show the intensity and phase of a TerraSAR-X spotlight image of central Berlin.
Abbildung in dieser Leseprobe nicht enthalten
Figure 4: SAR image : Signal intensity from the observed area.
SAR is a coherent, microwave imaging technique, which produces a two dimensional reflectivity map of a studied scene13. The 3D coordinate system of a SAR sensor is illustrated in figure 6. Since a single SAR image can only provide cartographic information in azimuth and range (2D) in order to retrieve the 3D position, i.e. elevation, as well as motion information of point scatterers, advanced interferometric SAR techniques, like
Abbildung in dieser Leseprobe nicht enthalten
Figure 5: SAR image : Phase of the signal from observered area (mirrored)
TomoSAR inversion, TomoSAR and differential TomoSAR are required. They provide the most advanced means for 4D SAR imaging to date2. The SAR geometry is defined by (Fig.6):
- azimuth x
is also called the along-track direction and is aligned with the flight direction of the sensor. Although the satellite flies approximately in a circle around the earth, for short lengths of trajectory, it can be handled as a straight line7.
Abbildung in dieser Leseprobe nicht enthalten
is also referred to as slant rage, perpendicular to the azimuth, in line-of-sight (LOS) direction of the antenna and pointing to the earth object objects further away from the sensor have bigger range coordinates.
- elevation
often called the cross range, s, is the third component of the rectangular coordinate system and can be assumed to be a straight line.
The antenna emits short radar pulses with the speed of light c, one after another to the ground scene that reflects part of the energy back to the antenna receiver. The time delay of the reflected signal can be written as:
Abbildung in dieser Leseprobe nicht enthalten
and depends on the distance d between the object and the sensor 13. Different objects and scatterers from the radar (different slant ranges) produce different delays between the transmission and reception of the signal. The next chapter explains the phasemeasurement of a SAR pixel.
Abbildung in dieser Leseprobe nicht enthalten
Figure 6: SAR Geometry2
The signal transmitted from the radar has to reach the targets on the ground and is reflected back to the radar to create the SAR image. Since the transmitted signal is of almost sinusoidal nature3, the delay is equal to the phase change φ between both signals (received and transmitted) as in equation 2.
The phase of a pixel in an SAR image can be written as:
Abbildung in dieser Leseprobe nicht enthalten
with R being the range between the object and the satellite, λ the wavelength, and n being an unknown integer that wraps the phase into the interval of [ −π: π ]. A phaseimage can be seen in figure 5. However, since the signal is periodic, the travel distance is a multiple of the wavelength, and has exactly the same phase change. In other words, the phase of an SAR signal is a measurement of just the last part of the two-way travel distance that is smaller than the transmitted wavelength.
The concept of the SAR phase is illustrated in figure 7.
Since the wavelength of TerrarSAR-X is around 3 . 1cm6, the phase is very sensitive to range. Depending on athmospheric noises, thermal noises or other influences, φ noises and φ APS are added to equation 214.
Layover is a property which results from the slanted radar geometry.
The radar resolution cell (RRC), at a certain azimuth coordinate, contains all scatterers that have the same range to the sensor. If elevated targets are projected orthogonally into the slant range, they appear in the image in exactly the same location as certain ground targets would. This is the phenomena called ”layover”, where the target’s points are being projected into the slant range and appear to ”lay over”.
Abbildung in dieser Leseprobe nicht enthalten
Figure 7: A sinusoidal function for a transmitted SAR signal3
To achieve a high resolution in range, sending a short impulse is better. However, the signal cannot be pulsed arbitrarily short, since high energy pulses have to be emitted to achieve sufficient signal-to-noise ratios (SNRs)13 ; therefore the Chirp signal is being used. As mentioned before, the Chirp signal is a linear-frequency-modulated impulse, which is used as the radar signal, and subsequently focused after receiving. This allows a broader bandwidth under limited signal power15.
An SAR image is often affected by the so called speckle effect, demonstrated by a granular ”salt and pepper” appearance on the images. The speckle phenomenon can be referred to as the random positive/constructive and negative/destructive interference of wave con- tributions from many individual distributed scatterers within one resolution cell16 7. Even though speckle is a scattering effect and not a noise, for SAR data processing pur- poses, it can be modeled as multiplicative noise. Through over homogeneous media, the complex speckled backscattering reflectivity is the product of the original unspeckled re- flection coefficient, and the multiplicative speckle contribution that is generally assumed to be a complex Gaussian distributed random variable13. Due to this fact, it is neces- sary to apply different filtering techniques to reduce the speckle effect16. Speckle can also be slightly reduced by temporal multilooking7.
The high resolution in azimuth is possible due to the synthetic aperture. In a conventional radar system, the azimuth resolution is related to the size of the ground-footprint. With the antenna length L in the azimuth direction, and the wavelength λ, the spread angle of a real antenna in azimuth direction α ra is given by
Abbildung in dieser Leseprobe nicht enthalten
Having a range of r 0, the spatial resolution in azimuth can be written as λ r 0
Abbildung in dieser Leseprobe nicht enthalten
The equation 4 implies that the longer the distances between sensor and target, the lower the spatial resolution in the azimuth direction. That is the reason why the synthetic aperture principle is applied. By the movement of the single monostatic sensor, an artificial antenna is built up. The reflected radar signal from the ground target is contained in radar echoes leading to a phase history over the observed time13. To construct a synthetically enlarged antenna, the pulses are coherently combined.
The angle of the beam, spread by the synthetic antenna can be expressed as in equation 5 with the synthetic antenna length of Lsa.
Abbildung in dieser Leseprobe nicht enthalten
The difference to equation 3 with the factor two originates from the phase shift undergone by the two-way path between antenna and the target.
The maximum length of the synthetic aperture Lsa (eq.6) for a ground target at range distance r 0, is limited by the flight path length during the observation, like the size of the antenna footprint on the ground
Abbildung in dieser Leseprobe nicht enthalten
As a result, the spatial resolution in azimuth using a synthetic aperture, assuming the target is within the beam, is demonstrated by
Abbildung in dieser Leseprobe nicht enthalten
As EINEDER mentioned, the smaller the real antenna, the higher the azimuth resolution7. Depending on the SAR imaging system, specifically the stripmap mode, the synthetic antenna length for the azimuth resolution is equal to the extension of the rectangular baseline Δ b2.
The final slant range resolution δ sr, depends on Wp the chirp bandwidth7 17, and is given by equation 8, where c already appeared in equation 1 and denotes the speed of
light:
Abbildung in dieser Leseprobe nicht enthalten
SAR interferometry (InSAR) is the technique of working with interferograms to detect ground deformations or creating a digital elevation model4 7. An interferogram is the product of one SAR image with the complex conjugate of the other. Depending on the way that the two acquisitions are carried out, one could have:
- single-pass interferometry: two observations within one survey (two antennas measuring the phase difference from different positions at the same time)
[...]
[1] X. X. Zhu, Y.Wang, S. Gernhardt, and R. Bamler, “Tomo-genesis: Dlr’s tomographic sar processing system.” unpublished.
[2] X. X. Zhu, Very High Resolution Tomographic SAR Inversion for Urban Infrastructure Monitoring - A Sparse and Nonlinear Tour. No. 666 in C, Verlag der Bayerischen Akademie der Wissenschaften, ISBN 978-3-7696-5078-5, 2011.
[3] A. Ferretti, A. Monti-Guarnieri, C. Prati, and F. Rocca, InSAR Principles: Guidelines for SAR Interferometry Processing and Interpretation. ESTEC Postbus 229 2200 AG Noordwijk, Netherlands: ESA Publications, 2007.
[4] M. Nannini, “Advanced synthetic aperture radar tomography: Processing algorithms and constellation design,” tech. rep., Deutsches Zentrum f¨ur Luft- und Raumfahrt e.V., DLR-FB–2010-06, 179 S., 2010.
[5] W. Carrara, R. Goodman, and R. M. Majewski, Spotlight Synthetic Aperture Radar. 685 Canton Street: Artech House, ISBN 0-89006-728-7, 1995.
[6] DLR, “Terrasar-x - deutschlands radar-auge im all.” Webpage: http://www.dlr.de/, 03 2010.
[7] D. M. Eineder, Photogrammetrie und Fernerkundung - Kapitel 6 Satellitenfernerkundung - SAR. TUM Institut f¨ur Methodik der Fernerkundung, Deutsches Zentrum f¨ur Luft- und Raumfahrt e.V., 2012.
[8] U. Soergel, ed., Radar Remote Sensing of Urban Areas, vol. 15. Nienburger Str.1 30167 Hannover: Springer, ISBN 978-90-481-3750-3, 2010.
[9] A. Ferretti, C. Prati, and F. Rocca, “Permanent scatterers in sar interferometry,” IEEE Transaction on Geoscience and Remote Sensing, vol. 39, p. 8, January 2001.
[10] DLR, “Terrasar-x - deutschlands radar-auge im all.” Webpage: http://www.dlr.de/.
[11] D. N. Reinke and R. Werninghaus, “Terrasar-x the german radar eye in space.” Mission Brochure, Bonn-Oberkassel K¨onigswinterer Straße 522-524 53227 Bonn Germany, July 2009.
[12] M. A. Richards, Fundamentals of Radar Signal Processing. McGraw-Hill, ISBN 978- 0071444743, 2005.
[13] S. Sauer, Interferometric SAR Remote Sensing of Urban Areas at L-Band Using Multibaseline and Polarimetric Spectral Analysis Techniques. PhD thesis, Universit´e de Rennes 1, DLR-FB-2008-26. 171 S., ISSN 1434-8454, 2008.
[14] R. Scheiber, Hochaufl¨osende Interferometrie f¨ur Radar mit synthetischer Apertur. PhD thesis, Universit¨at Fridericiana Karlsruhe (TH), Adenauerstr. 6, 82178 Pucheim, December 2003.
[15] H. Griffiths and C.J.Baker, Advances in Sensing with Security Applications: Fundamentals of Tomography and Radar. Springer, ISBN 978-1-4020-4284-3, 2006.
[16] W. Hagg, Merkmalbaiserte Klassifikation von SAR-Satellitenbilddaten, vol. 10. D¨usseldorf : VDI-Verl., ISBN 3-18-356810-1, 1998.
[17] C. V. Jakowatz, Spotlight-Mode Synthetic Aperture Radar: A signal processing approach. Kluwer Academic Publishers, ISBN 978-0792396772, 1996.
Der GRIN Verlag hat sich seit 1998 auf die Veröffentlichung akademischer eBooks und Bücher spezialisiert. Der GRIN Verlag steht damit als erstes Unternehmen für User Generated Quality Content. Die Verlagsseiten GRIN.com, Hausarbeiten.de und Diplomarbeiten24 bieten für Hochschullehrer, Absolventen und Studenten die ideale Plattform, wissenschaftliche Texte wie Hausarbeiten, Referate, Bachelorarbeiten, Masterarbeiten, Diplomarbeiten, Dissertationen und wissenschaftliche Aufsätze einem breiten Publikum zu präsentieren.
Kostenfreie Veröffentlichung: Hausarbeit, Bachelorarbeit, Diplomarbeit, Dissertation, Masterarbeit, Interpretation oder Referat jetzt veröffentlichen!
Kommentare