INTERFEROMETRY SESSION Afternoon, 2 April 2001 Chair : Fabio Rocca Co-chair : Jean-Claude Souyris Abstract Number : CEOS-SAR01-002 Preferences : Oral Session : Interferometry Inland Flexure Due to Ocean Tide Loading : A Possible Detection by Satellite Radar Interferometry at Korean Peninsula Masato Furuya and Shuhei Okubo Earthquake Research Institute , The University of Tokyo Address : Yayoi 1-1-1, Bunkyo, Tokyo 113-0032, Japan, Email : furuya@eri.u-tokyo.ac.jp After the launch of Topex/POSEIDON satellite altimeter, a number of high-quality oceanic tidal models are available, and thus we can precisely evaluate the effect of ocean tides upon several geodetic observables such as displacements, strain, gravity and so on. Meanwhile, synthetic aperture radar (SAR) interferometry has emerged over the last several years as a new promising tool for detecting crustal deformation. In contrast to the conventional point-wise measurement technique, SAR interferomety has enabled us to image the crustal deformation over a wide area with high spatial resolution. Those two developments motivated us to observe the oceanic tidal loading deformation by SAR interferometry. Comparison of the interferogram with the prediction may test the precision and accuracy of ocean tide model, and may provide a newer insight into the load Green function, because load Love numbers have greater sensitivity to a shallower structure of the crust than do the body tide Love numbers. The analysed area is the west coast of south Korea facing the Yellow Sea, which is famous for its large amplitude of ocean tide. The observed deformation is the difference of two snapshots between January 6 and March 16, 1996. The difference of the ocean tidal amplitude of these two periods at Inch'on reaches about 3 meters. We employed ERS1 and ERS2 satellite SAR data. In order to get rid of topographic fringe, we had to prepare a digital elevation map with high spatial resolution. It was derived from the ERS1/2 tandem mode data between March 16 and 17, 1996. The theoretical tidal loading deformation is computed by convolving the elastic load Green function with the east Asian regional ocean tide model, NAO99, by Matsumoto et al (2000). We could observe a concentric circular pattern around Asan Bay in the derived interferogram, which is expected from theoretical computation. However, at the time of writing this abstract, the observed amplitude is somewhat larger than model computation. Abstract Number : CEOS-SAR01-004 Preferences : Oral Session : Interferometry A Study on Interferometric SAR Image Restoration Using Complex-Valued Neural Networks and Its Application to Phase Unwrapping Problem Andriyan Bayu Suksmono (1) and Akira Hirose (2) (1) Research Center for Advanced Science & Technology (RCAST), The University of Tokyo 4-6-1, Komaba, Meguro, Tokyo, 153-8904 Japan, Email : suksmono@okabe.rcast.u-tokyo.ac.jp (2) Department of Frontier Inforamtics , The University of Tokyo 4-6-1, Komaba, Meguro, Tokyo, 153-8904 Japan, Email : ahirose@ee.t.u-tokyo.ac.jp We propose a pre-processing before unwrapping of interferometric synthetic aperture radar (InSAR) image. The objective is to reduce the number of residues (singular points). It reduces the computational complexity in branch-cut-type unwrapping and, hence yields a better unwrapping result. We combine intensity and phase information that is available in InSAR image to produce complex image. We use complex-valued Markov random field (CMRF) model in the restoration process. This is an extension of a well-known MRF model that has been used widely in the area of image classification, segmentation and restoration [1]. In our previous work, a causal 2nd order CMRF model has been applied in InSAR image classification [2]. The system is implemented as a (lattice) complex valued neural network. We put marks to the neurons corresponding to corrupted part (including residues) in the complex image. We choose the normal area to estimate 5th order non-causal CMRF parameters and use the result to restore the corrupted part. That is to say, we define an energy function as a squared amplitude value of the difference between raw value and estimated value. By reducing the energy, the corrupted part is restored. Energy minimization is achieved by using least mean square (LMS) steepest descent algorithm. The energy is decreased exponentially during the iteration and the residues number is eventually decreased. A closer observation on the experiment result shows that the reduction is due to smoothing process and appropriate fringe-line cutting as we expected. Because the restored image has fewer residues than the raw image, it is an advantage in branch-cut-type phase unwrapping process. We apply our method to interferometric image of Mt. Fuji which has a dense residues in the peak area. As a result, we obtain a faster process and a better result of unwrapped phase by the present method in combination with the minimum spanning tree (MST) branch-cut phase unwrapping algorithm in comparison with a direct unwrapping by the MST algorithm. [1] S. Geman & D. Geman, "Stochastic relaxation, Gibbs distributions and Bayesian restoration of images," IEEE. Trans. Pattern Anal. Machine Intel., Vol. PAMI-6, pp. 721-741, Nov. 1984. [2] A.B. Suksmono & A. Hirose, "Adaptive complex-amplitude texture classifier that deals with both height and reflectance for interferometric SAR images", IEICE Trans. Electron., Vol. E83-C, No.12 December 2000, to appear. Abstract Number : CEOS-SAR01-020 Preferences : Oral Session : Interferometry Constraints on the Design of DINSAR Satellites Andrea Monti Guarnieri, Claudio Prati, and Fabio Rocca Elettronica ed Informazione , Politecnico di Milano (POLIMI) Piazza Leonardo da Vinci 32, 20133, Milano, Italy, Email : rocca@elet.polimi.it, monti@elet.polimi.it In the new generation of satellites for earth observation, an important role could be covered by satellite based SAR systems designed for differential interferometry (DINSAR. The necessity of long time series would advocate the continuation of the existing observation series, namely that of ERS 1 - 2 and ENVISAT. In this last case, the frequency change from 5300 to 5331 MHz will make classical interferometry impossible, unless this frequency shift will be compensated by an equivalent baseline shift (about 2.3km on flat terrain). However, this will induce the rather small altitude of ambiguity of about 4.5 meters. This could be useful for high precision DEM's, but very taxing for differential interferometry. If distributed scatterers are systematically discarded, as it happens with the Permanent Scatterers technique, the situation would improve. If new satellites will be designed for interferometrical applications, several consequences derive: i) the need of coherent observations (unique central wavenumber); ii) the opportunity of an intermediate ground spatial resolution (say 10x10m); iii) the acceptability of relatively high Noise Equivalent sigma 0; iv) the need of short revisiting times. In fact, measurements show that the average radar cross section of stable scatterers in urban areas is about 450m2, and their dimension is rather limited as it can be seen from the scarce sensitivity of their reflectivity to the baselines. Thus, their average brightness allows for an increase of the usually accepted NE sigma0. On the other hand, a higher resolution could be useful for a more refined analysis of closely positioned targets moving at different subsidence rates. Too high resolution however, would impose shorter swaths and therefore would increase revisiting times. These considerations lead to the study of satellites without steerable antennas but operating in tandem pairs; this would also ensure reliability of operations to be able to propose to the users a DINSAR service. An alternate solution would lead to single more complex satellites with steerable antennas and shorter orbit repeat times. The need of simple artificial scatterers, with high endurance and such to guarantee long term millimetric stability will also be discussed. The results of preliminary tests will be presented. Abstract Number : CEOS-SAR01-025 Preferences : Oral Session : Interferometry Permanent Scatterers and Conventional DInSAR Techniques: A Comparison Alessandro Ferretti, Claudio Prati, and Fabio Rocca Elettronica ed Informazione , Politecnico di Milano (POLIMI) Piazza Leonardo da Vinci 32, 20133, Milano, Italy, Email : prati@elet.polimi.it, aferre@elet.polimi.it The Permanent Scatterers (PS) Technique has been recently introduced as a new approach to surface deformation monitoring using SAR data from satellite. In urban areas and where exposed rocks are visible, it is possible to identify numerous back scatterers that do not change their signature with time (the Permanent Scatterers) and therefore they can be used as natural monuments to estimate the progressive motion of the terrain. The precision of the measurement is a small fraction of a wavelength (5.6 cm) and millimetric motions are appreciable with good reliability. The atmospheric contribution is rather smooth spatially and independent from take to take, so that it can be identified and removed from the data using a proper processing, provided that the density of the PS is high enough (>5 per km2). Then, it is possible to obtain maps of subsidence with very high spatial sampling rate (more than 400 PS/km2, in urban areas) and high quality. After a short description of the possibilities and limits of this technique, the paper will present a comparison of the results that can be achieved with this new methodology with those that can be achieved by means of the conventional Differential SAR Interferometry. In particular the following advantages PS technique will be discussed both from the theoretical and experimental point of view: 1. Isolated coherent scatterers can be identified among non-coherent ones. Terrain displacement can be tracked in time on a pixel-by-pixel basis. 2. The critical (decorrelation) baseline as well as the Doppler centroid variations are no longer limitations for PS and all ERS images available in the ESA archives can be exploited. 3. The elevation of the PS can be estimated with high accuracy due to the very large baseline. 4. Atmospheric effects can be estimated and compensated. Abstract Number : CEOS-SAR01-029 Preferences : Oral Session : Interferometry RADARSAT-1 : Canadian Interferometric Mission Greta Reynolds, Ahmed Mahmood, and Marc Forbes RADARSAT Mission Planning , Canadian Space Agency 6767 route de l’Aeroport, St-Hubert, Quebec, J3Y 8Y9 Canada, Eamil : greta.reynolds@space.gc.ca In the fall and winter of 2000, the Canadian Space Agency undertook a mission to acquire interferometric data in conjunction with the Modified Antarctic Mapping Mission. This activity will provide the first high-resolution (fine beam) interferometric coverage of the northern Polar Regions The Canadian landmass north of 60 degrees was the primary area of interest for the mission. This would provide a complementary data set to the recent SRTM mission. Greenland was identified as a critical area for coverage due to the interest in polar ice caps. Iceland, and some of the remote islands provide targets of high geodynamic activity offering good interferometric case studies. Glaciated regions of North and South America, as well as Europe, are important for watershed management studies as well as global warming. And finally, regions impacted by mining and exploration provide case studies for land subsidence. During this period the orbit was more tightly controlled in an attempt to achieve a desired baseline of 250 meters (over Antarctica) for RADARSAT-1 standard beams used in acquisition. The implementation and results of orbit maintenance as related to the successful control to achieve desired baselines; mission planning activities to achieve successful acquisition results; as well as coverage results will be discussed. Abstract Number : CEOS-SAR01-039 Preferences : Oral Session : Interferometry Observations of the Earthquake Cycle from ERS and ALOS InSAR David Sandwell, Yehuda Bock, Suzanne Lyons, and Bridgett Smith Institute for Geophysics and Planetary Physics, Scripps Inst. of Oceanography 9500 Gilman Drive, 0233, La Jolla, CA 92093-0225 USA, Email : dsandwell@ucsd.edu The 9-year archive of synthetic aperture radar data (SAR), and precise orbital tracking data, from the ERS-1 and ERS-2 spacecraft (European Space Agency) is comparable in scientific value to the radar altimeter measurements of the marine gravity field from the Geosat spacecraft (US Navy, 1985-1989). Interferograms formed from matching SAR scenes reveal detailed topography (10-m vertical accuracy and 20-m horizontal resolution) as well as mm-level crustal deformation on 100-m horizontal length scales. Several groups in the US are working to make the ERS-archive more widely available for crustal dynamics research by forming a consortium (WInSAR) for purchasing, archiving, and distribution of ERS SAR data. In addition, the WInSAR consortium was selected NASDA to obtain large quantities of L-band SAR data from its upcoming ALOS mission (2003 launch). We propose to use these data to monitor co-seismic, post-seismic, and inter-seismic crustal deformation. Published models of the earthquake cycle predict that much of the diagnostic crustal strain signal occurs within 10 km of the fault zone and thus it will not be observed by the relatively coarse distribution of GPS monuments (> 10 km spacing). We plan to: * Develop high-resolution topographic models in the ERS/ALOS radar co-ordinates for accurate removal of topographic phase from change interferograms having baselines up to 150 m. This may depend on topography measured from SRTM. * Estimate the depth distribution of slip on the Hector Mine rupture to test the hypothesis that surface slip is less than slip at depth because the surface creeped during the inter-seismic period. * Examine post-seismic slip at the Landers and Hector Mine ruptures in relation to the earthquake cycle models. * Establish the creep rate and creep depth along major faults in Southern California. Preliminary results suggest that creep is triggered by strong nearby earthquakes. * Develop a boundary-element model of the Southern California fault system incorporating 3-D fault slip (i.e. deep and shallow slip). * Prepare for L-band SAR data to be collected by the Japanese ALOS mission (2003). An important issue for any SAR mission is the ability to view deformation from at least 2 look directions. We provide an example from the Hector Mine Earthquake where slip estimates derived from a single look direction are incorrect. Assuming zero displacement normal to the fault plane, the correct slip distribution is obtained from two look directions. Abstract Number : CEOS-SAR01-059 Preferences : Oral Session : Interferometry SRTM/X-SAR Calibration Michael Eineder, Bernhard Rabus, Helko Breit, Nico Adam, Juergen Holzner, Steffen Suchandt, and Richard Bamler Remote Sensing Technology Institute, German Aerospace Center (DLR) Oberpfaffenhofen D-82230 Wessling, GERMANY Email : Richard.Bamler@dlr.de, Michael.Eineder@dlr.de The Shuttle Radar Topography Mission (SRTM) flown in February, 2000 used two single-pass SAR interferometers (C- and X-band) to produce the most complete 30m digital elevation model (DEM) of the world. Phase calibration of the entire system as well as orbit and attitude modelling is of utmost importance for the quality of the final DEMs. This status report summarizes the calibration and DEM quality issues involved in the processing of the SRTM X-band data. The following points will be addressed: * orbit and timing calibration * motion compensation and dynamic behavior of the 60m boom * long term and temperature dependent phase drift of electronics and cables * phase noise and coherence * processor error sources Error budgets and first validation results will be presented. Abstract Number : CEOS-SAR01-064 Preferences : Oral Session : Interferometry Calibration of P-band Interferometric Radar Data from the GeoSAR System Scott Hensley and Elaine Chapin Radar Science and Engineering, NASA/JPL MS 300-235, 4800 Oak Grove Drive, 91109 CA, Pasadena, USA, Email : sh@kaitak.jpl.nasa.gov GeoSAR is new airborne radar inteferometric mapping radar that operates at two frequencies X and P-band. The system is designed to map beneath the foliage using a combination of the multi-frequency interferometric mapping data to obtain a map of the true ground surface. Data can be collected in either 80 or 160 MHz modes covering the band from 270-430 MHz. Since there are many other emitters operating in this band causing radio frequency interference coupled with a novel processing algorithm required a number of innovations to calibrate the P-band radar. In this paper we will present these ideas and show first examples of P-band data over vegetated and unvegetated regions. Abstract Number : CEOS-SAR01-076 Preferences : Oral Session : Interferometry Correction of Atmospheric Excess Path Delay appeared in Repeat-pass SAR Interferometry using Objective Analysis Data Masanobu Shimada (1), Mai Minamisawa (2), Shigeki Kobayashi (1), and Naoyuki Fujii (3) Earth Observation Research Center, National Space Development Agency of Japan (NASDA/EORC) Roppongi First-Bldg. 13F, 1-9-9, Roppongi, Minato-ku, Tokyo 106-0032 Japan Email : shimada@eorc.nasda.go.jp, shigeki@eorc.nasda.go.jp (2) Remote Sensing Technology Center of Japan (RESTEC) Roppongi First-Bldg. 12F, 1-9-9, Roppongi, Minato-ku, Tokyo 106-0032 Japan, Email : mai@restec.or.jp (3) Research Center for Seismology and Volcanology , Nagoya University Furou-chou, Chigusa-ku, Nagoya, Aichi 464-8602 Japan, Email : fujii@seis.nagoya-u.ac.jp In repeat-pass SAR interferometry, errors in the satellite's state vector and the atmospheric excess path delay degrade the measurement accuracy of geodetic information (topography and surface deformation). This information requires centimeter order precision for phase measurement, but the current state vector of the satellite does not meet this accuracy requirement. The atmospheric excess path delay caused by water vapor may differ by several tens of centimeters from day to day and it must be corrected. In this paper, we investigate a method to correct the atmospheric excess path delay using global objective analysis data (GANAL) and to estimate the state vector using ground control points. We also have evaluated this method by using Mt. Fuji, observed by JERS-1, and found it to be very effective. We have also conducted the comparative study for correcting the atmospheric excess path delay using the several objective analysis datasets, i.e., Global Analysis data (GANAL), Regional Analysis Data (RANAL), European Center for medium-Range Weather Forecasts (ECMWF), and National Centers for Environmental Prediction (NCEP). Abstract Number : CEOS-SAR01-080 Preferences : Oral Session : Interferometry Neotectonics and SAR Interferogram in Taiwan : Uplift of Tainan Anticline (SW Taiwan) and 921 Chichi Earthquake B. Deffontaines (1), E. Pathier (1), C. T. Lee (2), B. Fruneau (3), D. Raymond (1), J. Angelier (1) and J. P. Rudant (3) (1) ESA-CNRS, Universite Pierre et Marie Curie T.26-25, E1, Case 129, 4, place Jussieu F-75252 Paris Cedex 05, France, Email : benoit.deffontaines@lgs.jussieu.fr (2) Institute of Applied Geology , National Central University, Taiwan, ROC (3) : Institut Francilien des Geosciences , Universite de Marne-La-Vallee Cite Descartes, FRANCE This talk will highlight the interest of interferometric studies linked to neotectonic activity in Taiwan. Two case examples will be detailed herein. 1. Tainan anticline: Interferometric processing of five SAR-ERS images reveals uplift of the Tainan relief (SW of Taiwan) during the period 1996-1998. The maximum measured ground motion is 2.8 cm along the radar line of sight towards the satellite, indicating for the displacement vector a vertical component of 3.2 cm, and an horizontal component of 0.8 cm towards the WSW considering the additional information from GPS data. The reconstructed displacement field is consistent with the geological interpretation of the relief as an actively growing anticline. The existence of this tectonic uplift implies that the deformation front of the Taiwan orogen is located farther west than usually assumed. The city of Tainan is thus located in an active deformation zone. However, some ambiguity remains concerning the mechanism and kinematics, which makes the natural hazard evaluation difficult. 2. Chichi earthquake : Taiwan island displays active tectonics expressed by high seismicity. The 921 Chichi earthquake (Mw= 7.3 Foothills, Western Taiwan), occured last 21st September 1999 at 1 :47am. By SAR-ERS interferometry, we show herein the deformation field in the western side of the active Chelungpu thrust fault (Taichung area) induced by the Chichi earthquake. It shows a slight subsidence, with a maximum displacement of 30cm. To conclude, this new methodology combined to precise GPS measurements, field work (trenching, fault and syndeposits analyses), geophysics, sismologist and morphotectonic works, lead to a better knowledge of active faults in Taiwan. Abstract Number : CEOS-SAR01-081 Preferences : Oral Session : Interferometry InSAR Analysis of the Volcanic Crustal Deformation of Miyake Island with RADARSAT Images Hiroshi Ohkura (1) and Masanobu Shimada (2) (1) National Research Institute for Earth Science and Disaster Precvention 3-1 Tennodai, Tsukuba, Ibaraki, 305-0006 Japan, Email : ohkura@ess.bosai.go.jp (2) Earth Observation Research Center, National Space Development Agency of Japan (NASDA/EORC) Roppongi First-Bldg. 13F, 1-9-9, Roppongi, Minato-ku, Tokyo 106-0032 Japan Email : shimada@eorc.nasda.go.jp Miyake-jima, which is a volcanic island located in about 200 km south of Tokyo, has erupted since the end of June 2000. All people of the island evacua ted on September 4, 2000, because of severe active condition of the volcano. Eleven GPS stations were set before the eruption in the island. However all of them stopped working mainly because of the failure of the electric power supply. Differential SAR inteferomety is the only way to monitor crustal deformation caused by the volcanic activity. From September 3 to November 7, 2000 orbit of RADARSAT-1 was carefully controlled in order to detect glacial movement in the Antarctic by differential interferometric SAR analysis for Modified Antarctic Mapping Mission-2 (MAMM-2.) We analyzed twelve RADARSAT images of the island acquired over the period of MAMM-2. Eight fine interferograms were produced although the island is covered with vegetation. The interferograms gave us valuable information of the activity. We also tried to analyze ERS-2 images but could not get a fine interferogram. This shows that space-borne C-band SAR with HH polarization is available to differential interferometric SAR analysis over the vegetated area if its orbit is carefully controlled.