Logo 2016 Dragon 3 Final Results Symposium

Conference Agenda

Overview and details of the sessions and sub-session of this conference. Please select a date or session to show only sub-sessions at that day or location. Please select a single sub-session for detailed view (with abstracts and downloads if available).

Session Overview
D2: ID.10569 Terrain Measurement
Tuesday, 05/Jul/2016:
9:00am - 10:00am

Session Chair: Cécile Lasserre
Session Chair: Qiming Zeng
Workshop: Terrain & Geoid Measurement
Location: A-Li Mountain Room, 5.5 Floor, Junyi Dynasty Hotel

Show help for 'Increase or decrease the abstract text size'
Oral presentation

TerrainMeasurement with SAR/InSAR

Deren Li1,2, Mingsheng Liao1,2,4, Lu Zhang1, Timo Balz1, Tianliang Yang3,4, Hanmei Wang3,4

1LIESMARS, Wuhan University, China, Wuhan 430079, P.R. China; 2Collaborative Innovation Center for Geospatial Technology, Wuhan 430079, P.R. China; 3Shanghai Institute of Geological Survey, Shanghai 200072, P.R.China; 4Key Lab. of Land Subsidence Monitoring and Prevention,Ministry of Land and Resources, Shanghai 200072, P.R.China;

Topographic mapping and surface motion estimation with spaceborne SAR sensors are the main topics of the Dragon-3 project "Terrain Measurement (id 10569)”. With the increasing number of SAR satellites operational, working in different modes, with different resolutions and wavelengths, the combined analysis of this data is getting more important. In Dragon-3, we work on different test sites investigating the following four topics:

1. Topographic mapping with SAR. Interferometric SAR (InSAR) is the main method for the generation of digital elevation models (DEM) with SAR. With the TanDEM-X mission, the DLR supported a complete satellite mission for generating a worldwide DEM using interferometric SAR. However, in repeat-pass mode InSAR is suffering from temporal decorrelation and therefore cannot be used in many situations. In Dragon-3 we worked on solving these problems. To mitigate the geometric decorrelation problem, we developed a method to fuse height measurements from both ascending and descending InSAR observations to reduce data voids as much as possible. To cope with the temporal decorrelation problem, we explored using radargrammetry for repeat-pass configurations especially over forest areas. Furthermore, the processing of bi-static and multi-static pursuit mode configurations of recent and future SAR systems are investigated.

2. Landslide movement analysis. Landslide surface motions analysis with SAR data has been one of our main research topics since Dragon-2. Traditional differential InSAR (D-InSAR) and advanced time-series D-InSAR methods have been tested in Dragon-3. However, we found these phase-based methods can only retrieve line-of-sight (LOS) displacements at slow-moving landslides. In combination with the temporal decorrelation, this makes D-InSAR based methods seldom suitable. Therefore, in Dragon-3 we used the point-like target offset tracking (PTOT) method to measure large displacements in both azimuth and range dimensions at the Shuping landslide. Time series displacements measured at corner reflectors (CRs) were further analyzed to evaluate the impacts of reservoir water level variation and rainfall events on the landslide stability.

3. Urban subsidence analysis. SAR systems can measure distances and movements with high precision. Using for example PS-InSAR, deformations can be estimated with a very high precision. The long-term surveillance of urban subsidence in Shanghai is our major research goal since Dragon-1. With data starting from ERS-1, over ENVISAT ASAR, ALOS PALSAR, up to modern systems like TerraSAR-X, COSMO SkyMed, PALSAR-2, and Sentinel-1, we continuously monitor the subsidence over Shanghai for far over a decade now. The combination of this data and the analysis of the continuous deformation is still on-going. Remarkably, the PS-InSAR precision stays stable over time even using different sensors.

4. Tomographic SAR analysis. Tomographic SAR is another approach of multi-baseline SAR for 3D/4D analysis. In Dragon-3 we are focusing on the coherent approach of TomoSAR used mainly in urban areas. This technique is strongly related to PS-InSAR analysis and even uses PS-InSAR for atmosphere estimation as a pre-processing step. The spatial resolution in elevation is typically rather though, especially using approaches, like SVD (Singular Value Decomposition) or TSVD (Truncated SVD). On the other hand, compressive sensing based approaches like BP reach super resolution in elevation, but require large amounts of computation. With TWIST (Two-step Iterative Shrinkage / Thresholding) for TomoSAR we are working on an approach for super resolution TomoSAR with only slightly higher computational requirements than TSVD.

Li-TerrainMeasurement with SARInSAR_Cn_version.pdf

Oral presentation

Assessment of radargrammetric DSMs accuracy depending on the acquisition geometry

Simone Mancon1, Stefano Tebaldini1, Timo Balz2

1Politecnico di Milano, Italy; 2Wuhan University, China;

This study is intended to assess the capability of radargrammetric processing of SAR data for retrieving a digital surface model (DSM) of forested areas.

In particular, we focus on the experimental verification of the theoretical performances of the technique depending on the look angle separation.

Historically radargrammetric processing had been exploited to measure DSM by inverting the coregistration shifts as measured by matching two non-interferometric SAR acquisitions.

This is because the residual coregistration shifts are a measure of the errors in the geometric information knowledge (DSM and sensor trajectory).

In the interferometric processing the dependence between the mis-registration shift and the DSM errors is well known and depends on the baseline separation.

The same theoretical dependence from geometric acquisition can be defined in the field of the radargrammetric processing in terms of the look angle separation between the two non-interferometric acquisitions.

In this work we intend to use the interferometric formulation of the theoretical mis-registration shifts sensitivity to the DSM errors and verify its applicability to the radargrammetric processing.

The results of a performance analysis (in terms of DSM accuracy) will be provided for different values of look angle separation and for different examples of real RCS measured over forest areas.

In particular, the analyses will be performed on a test site near Mount Song, located in Henan province on the south bank of the Yellow River (China). Concerning the SAR systems, we will exploit both TerraSAR-X and COSMO-SkyMed data, and in particular: High Resolution SpotLight (HS), SpotLight (SL) and StripMap (SM) for TerraSAR-X and StripMap (HIMAGE) for COSMO-SkyMed.

Mancon-Assessment of radargrammetric DSMs accuracy depending_Cn_version.pdf


Monitoring Structural Health of Different Types of Bridges Using Advanced Multi-temporal InSAR

Xiaoqiong Qin1, Mingsheng Liao1,2, Mengshi Yang1, Lu Zhang1, Timo Balz1

1Wuhan University, China, People's Republic of; 2Collaborative Innovation Center for Geospatial Technology, Wuhan, China;

Tianjin Binhai New District, the center of Bohai economic circle, is established on waterways with bridges as a connector with urban center. However, the materials degradation, traffic increase and environmental stresses may destroy the structure and function of bridges and further pose significant risks to public safety. The widespread deterioration and some recent collapses of bridges due to limited maintenance and repair have highlighted the importance of developing effective bridge monitoring strategies that can help identify structural problems before they become critical and endanger public safety[1].

To ensure the longevity and safety of bridges, as well as to optimize their management, 3D Laser Scanning, Fiber Optics and Leveling have been implemented. Unfortunately, the in-situ sensors are appropriate for targeted monitoring and cannot be readily deployed on a large scale. Spaceborne SAR interferometry has been shown to be a viable method to offer a source of independent information for structural health of bridges cost effectively and is capable of eliminating the need for bridge lane closure and traffic disruption. Although successful applications have been achieved, applying the conventional InSAR technology for bridge-specific analysis still presented significant challenges such as virtual point targets and seasonal motions[2-4]. To overcome these problems, improvements have been made to enhance the performance of multi-temporal InSAR for the special characteristics of the bridge deformation monitoring based on the Persistent Points(PSs).

Usually, the complex structures of Bridges over water resulting confusion of single bounce, double bounce and triple bounce points. Since our purpose is to estimate the deformation parameters of the bridge, we focus on identifying and analyzing the single bounce points which exactly on the bridge structure. In a spatial context, the points falling outside of a buffer zone from the bridge or a reasonable elevation range are excluded. Meanwhile, in a temporal context, only points with multi-temporal amplitude stability throughout the observation period are considered. However, it is difficult to give a proper statistical description of PSs just with the limited images. Therefore, maximum likelihood of the coherence is used to evaluate the phase accuracy for each pixel, which is independent from the number of images. Only points with high coherence are chosen as the final PS points.

Additionally, the “virtual displacement” correlated with temperature is removed from the total displacement. The residual phases, calculated by subtracting the estimated linear deformation and DEM error, still include two essential terms: the atmospheric turbulence and nonlinear deformation. Taking into account their different frequency characteristics in space and time, the atmospheric artifacts can be estimated with a filtering process in both spatial and time domains. As the steel core bridges may be very sensible to thermal dilation effects, we can infer that temperature is a main triggering factor of the nonlinear deformation. Therefore, temperature is introduced for the investigation of the nonlinear deformation. The deformation seems highly related to the seasonal change in temperature and a linear relationship is fitted between the deformation and temperature. The distribution of the estimated thermal coefficient along the bridges is carried out. After removing the deformation fitted by the temperature from the total deformation, the deformation signal of interest is then obtained.

Two high-resolution SAR image stacks, TerraSAR spotlight and Cosmo SkyMed, are used to extract the deformation along Nanjiang Bridge and Caihong Bridge in Tianjin Binhai New District by the aforementioned method. The high resolution and short wave radar data can help to improve the reliability and coverage of monitoring by increasing the density of suitable point targets. As Nanjiang Bridge is a continuous rigid frame with pre-stressed steel and concrete integrally linked, the PSs evenly distributed in the overall structure. On the other hand, Caihong Bridge is a tied arch bridge with PSs intensive in piers and sparse over the span. From the distribution characteristics of PS points, we can infer that the Nanjiang Bridge is a coherent bridge with high coherence in the overall structure and Caihong Bridge is a partial coherent bridge with high coherence in some structures and decorrelation in other structures. From the distribution characteristics of thermal coefficient along the bridges, we can find that the piers lock the structure causes the accumulation of the thermal dilation along the spans, reaching its maximum values in their central part. In addition, the steel rigid bridge is affected more by the thermal dilation than the tied arch bridge. From the perspective of settlement pattern along the main bridge, the deformation located mainly between the bridge piers and stable areas occurred near the bridge piers. This is highly consistent with the actual principles of bridge mechanics, and confirms the reliability of our results. In addition, a comparison between measurement from InSAR and Leveling in Caihong Bridge also shows good agreement.

In summary, our results indicated that high-resolution multi-temporal InSAR has unique advantages in identifying the type of bridges, analyzing the spatial-temporal deformation characteristics of PSs and measuring mm-scale deformation of the overall bridge structure. It is not only beneficial to predictively analyzing the requirements of field inspection for a given bridge, but also to improving the usability and value of deformation interpretation for bridge engineers and decision makers.

Qin-Monitoring Structural Health of Different Types of Bridges Using Advanced Multi-temporal InSAR_Cn_version.pdf


Deformation Detection of Potential Loess Landslide with InSAR Observation

Yuzhou Liu1, Mingsheng Liao1,2, Xuguo Shi1, Lu Zhang1,2

1State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University; 2Collaborative Innovation Center for Geospatial Technology, Wuhan University;

Loess landslides are a common geological phenomena occurring in distinct environments. In loess region of China, the micro structure of loess along with external triggering factors have an adverse effect on the slope stability. In consideration of the fact that loess is widely distributed in the mountainous area of Qinghai Province, traditional landslide monitoring methods such as geodetic surveying, GPS monitoring networks and telescopic gauges have practical limitations of cost and coverage in this area. In recent years, multi-temporal synthetic aperture radar interferometry (InSAR) has been regarded as an effective tool for deformation detection with wide range and high precision. Therefore, it is essential to employ multi-temporal InSAR technique to evaluate the stability of both known landslides and potential unstable slopes, which may be the basis for landslide recognizing, precursory information extracting and early warning.

As a representative method of multi-temporal InSAR, Small Baseline Subset (SBAS) technique obtains a series of small-baseline differential interferograms by combining acquisitions appropriately, which can overcome a spatial decorrelation phenomenon. In this study, one descending Environmental Satellite C-band Advanced Synthetic Aperture Radar (ASAR) data stack and one ascending Advanced Land Observation Satellite Phased Array type L-band Synthetic Aperture Radar (PALSAR) data stack are used as test data. Considering the characteristics of the acquisitions, SBAS technique is used to implement multi-temporal InSAR analysis.

In this paper, the same two anomalous deformation regions in Guide County have been detected from PALSAR and ASAR data stacks. Qualitative and quantitative evaluations of the point-like targets are made for understanding the distribution regularity of deformation. Compared with C-band ASAR acquisitions, the L-band PALSAR data have smaller temporal decorrelation effects and higher density of measuring points, which might be attributed to the good penetrating capability of vegetation canopy with its long wavelength. Meanwhile, preliminary correlation between the time series displacement and triggering factors such as rainfall and freeze-thaw processes is analyzed to explore the driving mechanism for landslide movement.

Liu-Deformation Detection of Potential Loess Landslide with InSAR Observation_Cn_version.pdf


Landslide Deformation Mechanism Analysis by coupling SAR Offset Tracking Results with Hydrological Factors: a Case Study of Shuping Landslide

Yanan Jiang, Mingsheng Liao, Lu Zhang

State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, China, People's Republic of;

Monitoring the surface deformation can help us to understand the characteristics and regularities of landslide evolution, which is fundamental for forecasting and early warning [1]. In recent years, time-series SAR techniques have been widely and successfully used in detection of earth's surface deformation, including landslide monitoring. Nevertheless, there is few work on in-depth interpretation of these results with external trigging factors. Thus, it is necessary to investigate the deformation mechanism of landslide in reservoir coupling with the impact of hydrological factors.

In this study, data assimilation (DA) is introduced to study the deformation response mechanism with the external inducing factors for understanding the development process of landslide. DA technique has been successfully applied in atmospheric, oceanic and land surface process [3]. Herewith, the DA is considered as essentially an estimation problem, i.e., acquiring the optimal estimate of the state variables of the landslide displacement given a set of comprehensive observations, a model operator and other available prior information.

The Shuping landslide in Three Gorges Reservoir (TGR) is chosen as the study area. Previous studies revealed that the Shuping slope stability was affected by the fluctuation of water level and rainfall [2]. The deformation used for response analysis are obtained by a modified offset tracking technique. In order to better reflect the relationship, time series decomposition is used to split the displacement from offset tracking into trend term and periodic term. The trend term displacement is controlled by the geological conditions of the slop and periodic term displacement is affected by periodic dynamic functioning of inducing factors, such as water level and rainfall. Then the periodic term can be regarded as a function of the water level and rainfall.

To determine the corresponding relationship of periodic term displacement to external inducing factors, a sequential data assimilation method, ensemble Kalman filter (EnKF) is used since it can handle multi-source data. At the beginning of assimilation, the estimation error of EnKF may be high. Whereas after a few steps forward, it reduces dramatically. On this basis, a short-term forecasting of the periodic term displacement is done. To make cross-comparison, two dataset of SM-mode and HS-mode TerraSAR-X images acquired from Feb 2009 to Apr 2010 are collected. The preliminary experiment shows that EnKF can effectively combine offset tracking results with slope displacement mechanism and finally reduce the predictive uncertainties.

Jiang-Landslide Deformation Mechanism Analysis by coupling SAR Offset Tracking Results with Hydrological_Cn_version.pdf


Monitoring Landslides with Advanced Multi-temporal InSAR Techniques in Danba of Sichuan Province, China

Jie Dong1, Mingsheng Liao1,2, Lu Zhang1, Jianya Gong1,2

1State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, China; 2Collaborative Innovation Center for Geospatial Technology, Wuhan University, China;

Danba County, located at the northwestern Sichuan, is one of the areas prone to severe landslides in China because of its vulnerable geological environment and complicated meteorological condition. Landslides often occurs in the area close to human settlements and main roads, posing great threaten upon local public safety and social economy. There are also many famous heritage architectures located on some active landslides. Therefore, monitoring landslide movements is of great importance for the sustainable developments in this area.

With its wide coverage and sub-centimeter accuracy, radar remote sensing has already proven its potential for remotely measuring unstable slopes. In this presentation, several classical InSAR techniques will be jointly employed to investigate typical landslide activities in Danba County, based on the multi-temporal SAR images acquired from ENVISAT ASAR and ALOS-1/2 PALSAR1/2.

Firstly, differential InSAR (D-InSAR) is used to recognize known landslides and find potential unstable slopes in a region scale. And a landslide inventory containing detailed information of each landslide such as type, location, coverage, deformation grade and so on, is established.

Then, for a specific landslide with huge threats, advanced multi-temporal InSAR (MT-InSAR) method is exploited to characterize its surface deformation by obtaining time-series displacement on coherent targets. In particular, for rapid-sliding Jiaju landslide, phase-based MT-InSAR is no longer applicable, hence the amplitude-based point-like targets offset tracking (PTOT) method is adopted instead to measure both azimuth and range displacements.

Lastly, above obtained deformation results are compared with these from external GPS in the same monitoring period to validate the reliability and rationality of our InSAR techniques. At the same time, correlation analysis between landslide deformation and triggering factors, such as human activities, rainfall, earthquakes, etc., is carried out to explore the driving mechanisms for landslide motion.

Dong-Monitoring Landslides with Advanced Multi-temporal InSAR Techniques_Cn_version.pdf


Atmospheric Correction in Interferometric Synthetic Aperture Radar Surface Deformation-A Case Study of Taiyuan City, China

Tang Wei1, Liao Mingsheng1,2, Yuan Peng3, Zhang Lu1, Timo Balz1

1State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, China; 2Collaborative Innovation Center for Geospatial Technology, Wuhan University, China; 3Research Centre of GNSS, Wuhan University, China;

Interferometric synthetic aperture radar (InSAR) has been widely used as a geodetic technique to monitor the Earth’s surface deformation produced by several phenomena like earthquakes, landslides, land subsidence and volcanic eruptions. Advanced techniques, like Persistent Scatterer Interferometry (PSI), exploit a temporal stack of SAR images to overcome the limitations of classical InSAR such as atmospheric delay, topographic errors and temporal decorrelation.

The main systematic error on the accuracy of InSAR comes from phase propagation delays through the troposphere. The delay is typically decomposed into a stratified component, which correlates with the topography and often dominates the tropospheric signal, and turbulent components. In this paper, we used ERA-Interim reanalysis products from European Centre for Medium-Range Weather Forecasts (ECMWF) to estimate the stratified delay at epochs of SAR acquisitions. Our comparative results with radiosonde data demonstrated that atmospheric conditions from ERA-Interim could produce reasonable patterns of vertical profiles of atmospheric states. The stratified atmosphere shows seasonal changes which sometimes are significantly correlated with time. These properties indicate that the stratified delay cannot be properly estimated and extracted by the temporal high-pass (HP) filtering, which has been usually applied in the conventional PSI. Thus, we proposed the atmosphere-corrected PSI method that tried to correct the stratified delay on each interferogram by using ERA-Interim reanalysis products. The (residual) turbulent components were considered as random signal in space and time that can be effectively eliminated by the temporal high-pass (HP) filtering in PSI.

We applied the proposed method to ENVISAT ASAR images covering Taiyuan basin, China, to study the ground deformation associated with groundwater extraction. Experimental results show that the proposed method significantly mitigate the topography-correlated delay and the estimated ground displacements agree more closely with GPS measurements than the conventional PSI.

Wei-Atmospheric Correction in Interferometric Synthetic Aperture Radar Surface Deformation-A Case Study_Cn_version.pdf


High-Precision DEM Generation Using InSAR Technology

Tao Li1, Xinming Tang1, Xiaoming Gao1, Weinan Chen1,2, Qianfu Chen1

1Satellite Surveying and Mapping Application Center, NASG, Beijing, 100041, China; 2College of Geomatics, Xi’an University of Science and Technology, Xi'an, Shanxi, 710054, China;

DEM elevation accuracy is limited by using InSAR technology except in bistatistic interferometric mode such as SRTM and TanDEM-X. The main error sources in InSAR application in terrain modeling are temporal decorrelation consist of atmospheric phase screen, volume decorrelation, land cover change, etc. However, those temporal decorrelation factors in bistatistic mode can be eliminated by sending and receiving radar pulses simultaneously. Therefore, in this paper we conduct satellite InSAR capacity assessment in DEM generation by using TanDEM-X bistatistic images which cover the four typical terrains such as flat ground, hill, mountain land, and alpine land.

To fulfill the DEM quality of Chinese 1:50000 DEM standards, we propose to use the look factor iteration, short-and-long baseline combination, ascending and descending data compensation to increase the phase quality. First of all, three different multi-look factors, i.e. 8×8, 4×4 and 2×2, are iteratively used to increase the phase quality successively. 8×8 look factor is able to provide interferometric phase with the highest signal-noise-ratio (SNR). The interferometric phase continuity is high enough to suppress the phase unwrapping errors. However, it maintains the least texture information such as artificial buildings, forests, rocks, etc. 2×2 look factor is risky if directly used in InSAR processing. Its phase continuity is the lowest, meaning lowest coherence and highest phase unwrapping error. We subtract the coarse DEM generated using 8×8 from the 4×4 interferogram, the residual phase is used to calculate DEM differences between 8×8 and 4×4. Then the DEM differences are added back to the coarse DEM, thus a refined DEM is obtained. The 4×4 DEM is then used in 2×2 interferogram in the same way. Through the iteration, the resolution size of DEM could be increased without losing the high accuracy inherited from large look factors. Secondly, we make further DEM quality improvement by using the short-and-long baselines. Interferometric phase of image pairs with short baseline is not sensitive to height changes, meaning the phase continuity is high but the geometric accuracy is low. The elevation phase is calculated in short baseline image pair and then used as reference for the long baseline image pair. This decreases phase increments among pixels and make phase unwrapping results more reliable. Finally, DEMs generated using ascending and descending image pairs are compensated to maintain the valid information in shaded and foreshortening areas induced by the side-looking effect.

Experiments in Tianjin show that DEM height accuracy (DHA) in flat ground is 0.8 m. Yanliang is a hilly region near Xi’an city, the related DHA is 2.3 m. The northern part of Beijing is regarded as mountain land with DHA equal to 2.6 m. While DEM in an alpine region, i.e., Shuanghu of Tibet, is 5.6 m. Chinese 1:50000 DEM standard show that the corresponding elevation errors of the aforementioned four different terrains are 3 m, 5 m, 8 m and 14 m, respectively, indicating that TanDEM-X images are able to be used as data sources for 1:50000 DEM generation.

Li-High-Precision DEM Generation Using InSAR Technology_Cn_version.pdf


Multi-Resolution SAR Data Analysis Of Urban Structures In Wuhan

Ziyun Wang, Timo Balz, Mingshen Liao, Lu Zhang

LIESMARS, Wuhan University, No.129 Luoyu Road, Wuchang District, Wuhan 430072, Hubei Province, China;

In 2014, the TerraSAR-X missions, which consist of a variety of products ranging from 1m resolution Sliding Spotlight to 18m resolution ScanSAR images, implemented a new modes: the new Staring Spotlight (ST) mode, considerably increases the resolution of SAR acquisitions by fully exploiting the potential of SAR instrument (Fritz et al., 2013; Mittermayer, et al., 2014; Breit, et al., 2014), with 300 MHz bandwidth and combined with azimuth steering angles of , in contrast to the in sliding spotlight modes (Mittermayer, et al., 2012), achieving a azimuth resolution below 24cm.

With the new sub-meter resolution SAR data, more opportunities are rendered for a deeper research in to urban structures and a deeper understanding of backscattering behavior of typical urban structures. Earlier experiments of urban structure detection has been implemented by researchers using air-born SAR data with resolution less than half a meter, typical urban structures like buildings facades (Soergel, et al., 2005) and traffic roads (Amberg, et al., 2004) are identified.

Nowadays with 0.24m resolution space-born TerraSAR-X ST data, as well as 2.5m resolution Kompsat-5 (Yoon, et al., 2011) strip map data, and additionally 20m resolution Sentinel-1 (Torres, et al., 2012) IW data, the topic can be discussed in different levels with the help of multi-resolution SAR analysis in urban areas. Thus, multi-resolution SAR analysis will be conducted on urban area in our experiments. With multi-scale description of the urban areas, structures of typical artificial targets can be identified and possibly extracted in different level.

In this paper, experiments of multi-resolution SAR analysis will be conducted on Wuhan test site with both amplitude images and interferograms in order to exploit a deeper understanding of the urban structures. Interferometric analysis will be conducted on TerraSAR-X ST data, including 15 ascending images and 16 descending images, as well as Kompsat-5 strip map and Sentinel-1 data. Nevertheless, in order to handle the new imaging mode of ST data, focusing process of the traditional procedure needs to be updated. Specific processing procedures and experiment descriptions will be included in the full paper.

Wang-Multi-Resolution SAR Data Analysis Of Urban Structures_Cn_version.pdf


Research on 1:5000 scale DEM generation technology using airborne Ka-band InSAR data

QianFu Chen, Tao Li, XiaoMing Gao, WeiNan Chen, DanQin Wu

R&D Dept., Satellite Surveying and Mapping Application Center, China, People's Republic of;

Airborne Ka-band InSAR is able to provide images with high resolution,
high signal-noise-ratio, high-precision attitude, and high-precision
positioning data. It is able to provide highly precise phase accuracy with
cross-track interferometric mode, thus DEM with high precision can be obtained using InSAR. In this paper, we have used the orthogonal decomposition of look vector model to analyze the three dimensional positioning model based on the airborne Ka-band imaginggeometry. Firstly we get the orthogonal decomposition of look vector results of target points under the geocentric coordinate system, and then calculate the look vector, and then get the target points’ analytic coordinates which can straightly describe the three-dimensional position of ground targets. We thoroughly analyze the error propagation in the model and then establish the
interferometric calibration model to correct the main error sources. Finally we
conducte multi-baseline fusion given that the airborne instrument consists of
three antennas, which make it possible to construct three interferograms
through one flight. At first the interferometric calibration of multi-baseline InSAR
data will be made respectively. Then we get the fused DEM using coherence as weight. We conducte experiments using airborne Ka-band InSAR data covering Xi’an, China. And we use the Global Positioning System (GPS) to obtain ground control points (GCPs) with D-level accuracy, then using GCPs for the interferometric calibration model optimization and generate related DEM. Optimization results show that the precision of baseline length increased 1.39 mm, optimized the baseline angle 3.21, initial slant distance correction is 0.75 m, Doppler center frequency correction is -11.08 Hz, the interferometric phase correction is -11.12 radian. The DEM height accuracy obtained using aforementioned algorithms is 0.3 m in flat ground, 1.4 m in mountain region. At the same time we use multi-baseline fusion algorithm to obtain a new DEM, experiments show that multi-baseline fusion after DEM accuracy can be further improved, the corresponding errors in flat ground and mountain region are 0.25 m and 1.13 m, respectively. Finally, we use the fused DEM to generate corresponding digital orthophoto map (DOM), the results show that the error in the X direction is 0.75 m, Y direction is 0.91 m, the plane positioning error is 1.39 m, and it meets the requirements of DEM quality at 1:5000 measurement scale.

Chen-Research on 1_Cn_version.pdf


Forest Biomass Retrieval Using Multibaseline PolInSAR Data

Xinwei Yang1, Mingsheng Liao1, Stefano Tebaldini2, Lu Zhang1

1State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, China, People's Republic of; 2Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, 20133 Milano,Italy;

There is an urgent need for human beings to measure forest parameters and mapping biomass at global scale as they are critical factors for understanding the underlying mechanisms of climate change, carbon emissions and biodiversity. Forest height plays an important role in characterizing forest stand parameters, which can be related to above ground biomass (AGB) through allometric relations. Dealing with volumetric objects, remote sensing techniques which are suitable for information retrieval about forest structure are those who have under foliage penetration capabilities, such as the synthetic aperture radar (SAR). The signals emitted by SAR can propagate down to the vegetation layer, information can then be obtained by exploiting the received signals generated from the interactions between transmitted signals and natural media.

Since had been proposed by Cloude and Papathanassiou, polarimetric SAR interferometry (PolInSAR) technique has shown great potential in retrieving vegetation parameters such as forest height due to its sensitivity to the vertical distribution of the natural scattering media. Many sensors such as the JAXA’s ALOS-1 and ALOS-2, the CSA’s RADARSAT-2 and the ESA’s proposed BIOMASS mission all have the ability to retrieve canopy height and forest biomass. In PolInSAR both polarimetry and interferometry techniques are coherently combined to overcome the limitations when taken alone. For forest scenarios observed at L-band, the two-layer RVoG model has been proven to be sufficient for height retrieval in terms of robustness and performance. The derived tree height can then be taken as the input parameter to establish theoretical relationship with biomass for a given species.

In this work, the forest height and biomass parameters estimation performance using PolInSAR technique is evaluated on real airborne L-band SAR data acquired by E-SAR sensor in multibaseline configuration over Krycklan test site. The inversion method will be enhanced by considering the impact of decorrelation sources such as slope variation and temporal decorrelation, etc. The main purposes of our work are to propose improved multibaseline PolInSAR height inversion approach and establish revised model when evaluating AGB over boreal forest by using PolInSAR height results. Finally, the validity of the proposed methods will be analyzed by comparing the obtained results against LIDAR reference measurements.

Yang-Forest Biomass Retrieval Using Multibaseline PolInSAR Data_Cn_version.pdf


High Voltage Power Line Scattering Time Series Analysis in Space borne High Resolution SAR Images

Chen Zhiguo, Li Tao, Zhang Xiaozhi, Zhang Wei, Wu Wenhao, Xu Kan, Zhou Zhiwei

GNSS Research Center of Wuhan University, China, People's Republic of;

In high resolution SAR images such as TerraSAR and COSMO-Skymed, the metal power line may scatter as an ellipse spot similar with corner reflector. The location, pixel size and maximum amplitude of the power line spots vary according to satellite azimuth, line length, incidence angle, temperature and so on. How to find the relationship between scattering spots characters and the physical conditions need more analysis.

This paper chose 1000kV high voltage power lines in Hubei Province, the biggest power line in china, within 3 strip mode TerraSAR images more than 150 groups of power line scattering spots are specified for the testing. Those line scattering spots’ reflection geometry between satellite azimuth and line orientation have been analyzed. These spots’ intersection angles of lines and satellite azimuth directions are between -15 to 15 degree, and the location of spots change accord to the angle. In descending strip mode, while the intersection angle closed to -15 degree, the spots’ positions in SAR images approached to the upper tower, vice versa. Once the intersection angle is fixed, the scattering position will close to the line center with line length increasing. It shows that the longer line length and deeper sag maneuver the scattering position to the line center.

20 strip mode TerraSAR images in Hongkong area are selected for time series analysis on power line spots’ location shifting and phase changes. It shows that the spots’ magnitude are stable, but spots’ center location may shifting in range and azimuth direction. Comparing with the local average temperature trend, the line scattering spot center shifted with the same tendency. Fixing one of the top line spots phase as the reference, the time series differential phase between the other spots have been analyzed. However, the differential phase between two power line spots is not consistent with the height between two lines. Results show that the difference of elevation calculated by the phase of spots is far away from the real height.

In this paper, power lines’ reflection geometry has been studied and the change of location, amplitude, phase of the spots in time series SAR images have been analyzed. The power line scattering spots’ position changes according to line length, sag and incidence angle has been testified. The power line scattering spots location center shift with temperature. The differential phase of two spots are not consistent with the height difference. In the future, longer time series analysis will show more seasonal changes.

Contact and Legal Notice · Contact Address:
Conference: 2016 Dragon 3 Final Results Symposium
Conference Software - ConfTool Pro 2.6.94
© 2001 - 2016 by H. Weinreich, Hamburg, Germany