Methods InSAR 2

Monitoring long-term deformation processes require continuous observations. With the gap between Envisat and Sentinel-1, C-band observations by ESA satellites were interrupted by 5 years. The Netherlands choose for nation-wide coverage of RADARSAT-2 to bridge this gap. Recent developments within the MDA team, resulted in, amongst-others, the high resolution Extra Fine (XF) acquisition mode and the newly released EDOT state vectors, in which the spatial resolution and orbital positioning accuracy increased significantly.

In this work we demonstrate the impact of the improved accuracy of the orbital state vectors on the disentangling of deformation signal and atmospheric and/or orbital signal contributions. With residual orbital trends of less than 1 fringe/100 km we expect less deformation signal to leak into orbital trend estimates and vice versa.

Furthermore we demonstrate switching to a higher resolution within an existing medium-resolution stack. We show that switching to new beam modes does not necessarily mean discontinuity of existing time series analysis. On the contrary, the 5 times higher resolution of the XF beam mode compared to the Standard mode allows for high resolution analyses and hence more detailled monitoring of urban infrastructure. Tests on simulated data have shown that no more 10% of the original PS will be lost due to the higher compression rate of the XF data, but this is more than compensated by the higher resolution and thus significantly higher density of PS.

We can now use long time series at medium resolution for e.g. atmosphere estimation while smoothly switching to higher resolution deformation updates when required.

The availability of long time series of interferometric data acquired all over the world from several synthetic aperture radar (SAR) satellite missions makes possible to perform a worldwide assessment of the terrain and infrastructure stability by SAR interferometry techniques. This technology is computationally demanding, in particular because it requires a 3D type of processing. When applied to large areas, several problems have to be faced to handle huge amounts of data.

In this work, we present the first example in the world (to our knowledge) of PS SAR interferometry analysis at national scale (the whole Italian territory), performed with ERS, Envisat and COSMO-SkyMed data.

Moreover, we discuss possible worldwide extension based on the new Sentinel-1 SAR satellite.

As part of a program called “Special Plan of Remote Sensing,” with the objective of mapping and preventing geo-hazards, the Italian Ministry of the Environment awarded to an industrial team lead by e-GEOS a series of contracts aimed at realizing a database of ground surface deformation measurements by PS SAR interferometry processing of all the available ERS and Envisat SAR data over Italy, and then at updating the PS measurement database based on COSMO-SkyMed data stacks.

This ambitious task has required developing the most advanced PS interferometry algorithms, and using high performance computing (HPC) systems, in order to process and make accessible hundreds of interferometric data-stacks, thousands of SAR images and billions of PS measurements, and represents a pioneering service for mapping and preventing geo-hazards.

The PS processing of the whole ERS and Envisat SAR archive (over 10,000 images) acquired over Italy from 1992 up to 2010 was completed, and the obtained products are made available through the Italian national cartographic portal via WebGIS technology. Currently, about one hundred high-resolution interferometric data stacks acquired in 2011–2014 from the COSMO-SkyMed constellation (for a total of about 5,000 images) have been processed.

The PS analysis brought to the identification of about 14 million PS points with ERS data, 28 million PS points with Envisat data and 1 hundred million PS points with COSMO-SkyMed data. For each PS point, a set of deformation measurements corresponding to every acquisition date is obtained, for a total of more than 6 billion ground deformation measurements covering the period 1992–2014 and the whole Italian territory.

The density and the accuracy of the PS measurements obtained with the high-resolution Cosmo-SkyMed SAR data make possible the study of the stability of each single building and infrastructure and the analysis of landslides phenomena that were not observable with the old ERS and Envisat low-resolution SAR systems.

After the recent launch of the Sentinel-1 SAR satellite, new interesting scenario has become possible. The capability of Sentinel-1 to cover regularly the globe and with a short revisit time, even though with a lower spatial resolution, opens the possibility of extending the area of interest to continental or worldwide scale, and providing routinely SAR interferometry deformation measurements. In this context, high resolution SAR systems like COSMO-SkyMed or TerraSAR-X can be complementarily exploited for detailed analysis of a limited number of restricted areas, chosen based on a particular interest and the knowledge of critical phenomena happening, or also based on the lower resolution analysis performed with Sentinel-1 data.

This scenario will pose new challenges may call for cloud based paradigms that can be necessary both to process such an amount of data and to make accessible to the users this product in a convenient way. 

Since the launch of the ERS-1 satellite in 1991, an increasing number of SAR satellite missions have become operational. Based on a stack of acquired SAR images, a Persistent Scatterer Interferometry (PSI) analysis can be applied to estimate the surface motion of a certain area. However, due to different acquisition characteristics (e.g., wavelength, orbit), the images acquired by different SAR missions can typically not be combined within a single PSI analysis. As a result, for each mission, a separate PSI result is obtained, covering a limited period and each relative to a unique reference point. This situation remains in the future, since typically the orbit configuration will be different, for example in case of the Sentinel-1 and RadarSAT Constellation mission. Nevertheless, the creation of a single surface motion product for a certain area is desirable. Hence, an integration of the different PSI datasets is required.

Apart from PSI datasets, additional geodetic data may be available, such as time series of levelling and GNSS data. Levelling data has the advantage of a physical reference datum, and the possible availability of long time series (potentially >100 years). GNSS provide 3D observations in a global reference frame and a high temporal sampling. The complementary characteristics of PSI, Levelling and GNSS make it desirable to estimate an integrated product.

We present our methodology to integrate PSI, levelling and GNSS data based on the time series of the individual techniques. So far, the integration was based on linear deformation rates, see e.g., Caro Cuenca et al. (2011), and Fuhrmann et al. (2015). However, this approach is not optimal, since non-linear motions are not taken into account. Our approach considers the full time series of the various techniques to create integrated time series. The method is based on geodetic estimation and testing theory, taking the correlations between the different measurements into account. The results are longer and denser time series compared to those obtained by the individual techniques, which are referenced to physical datums.

We demonstrate our methodology using data of a mining area in Zuid-Limburg, the Netherlands. Here, mining occurred until 1975. To keep the mines dry, continuous pumping of ground water was required. Since 1975, the pumping has stopped gradually, resulting in non-linear uplift of the surface in time. Various faults, mining concessions, and nearby mining activities in Belgium and Germany, cause a complex surface deformation pattern. For the region, PSI datasets based on ERS-1/2, ENVISAT and RadarSAT-2 data are available (together 200 SAR images), as well as levelling and GNSS data from the Netherlands, Belgium and Germany.

References

Caro Cuenca, M., R. Hanssen, A. Hooper, M. Arıkan, 2011. Surface deformation of the whole Netherlands after PSI analysis, Fringe 2011 workshop, ESA SP-697, 1-8.

Fuhrmann, T., M. Caro Cuenca, A. Knöpfler, F.J. van Leijen, M. Mayer, M. Westerhaus, R.F. Hanssen, B. Heck, 2015. Estimation of small surface displacements in the Upper Rhine Graben area from a combined analysis of PS-InSAR, levelling and GNSS data, Geophysical Journal International, 10/2015, 203(1). DOI:10.1093/gji/ggv328.

SAR interferometric technique is widely used for analysing the geodynamics of earth surface phenomena. Basically, such imaging technique is accomplished by considering two coherent SAR images from two passes of a single SAR antenna (repeat pass interferometry) or with the single pass of two-antenna system (single pass interferometry) [1], [2]. However, the efficacy of interferometric technique is mostly affected by multiple decorrelation effects, which collectively results as SAR interferogram phase noise. Besides the disturbances such as orbital errors, typical sources of noise is the decorrelation associated with thermal noise, image co-registration error and temporal and geometrical decorrelation . To mitigate decorrelation effects, in this paper we introduce a two steps approach, which combines a minimum mean square error space varying filtering [3] for filtering of the spatial decorrelation and phase denoising based of sparse representation [4] of the interferometric signal.

In this context, we start with mitigation of geometrical decorrelation effects as proposed by [3]. Geometrical decorrelation depends mainly on the angular diversity between the two focused images, which causes baseline decorrelation as well as Doppler centeroid decorrelation due to different viewpoint of antenna beam. But presence of nonplanar topography can even increase the limit of geometric decorrelation effect. To overcome the limitation of such critical aspect, the SV-MMSE technique [3] has been successfully used as preliminary step in our procedure. To further reduce the decorrelation effects due to other noise sources, we followed an approach that intend to deal with the residual noise reduction problem by using sparse and redundant representation over “trained dictionaries”[5], strategy that we found to be highly effective and promising. Consequently, we choose K-SVD algorithm [4] as signal representation technique, which efficiently separates signal from noise, through a suitable choose of elementary signals, named “atoms” organized in a matrix form, named “dictionary”.

In this framework, we choose a data pair of ERS-1/ERS-2 (Tandem) data acquired over the area of Mt. Vesuviusvolcano near Naples, Italy, with a spatial baseline of about 250 m between the acquisitions, on December 29 and 30, 1995: the corresponding interferogram is shown in Figure 1(a). The preliminary estimation of noise free interferogram has been performed by considering SV-MMSE filtering technique, whose outcome has been displayed on Figure 1(b). Further we applied K-SVD technique on the SV-MMSE filtered interferogram, whose output is shown in Figure 1(c). From the experimental result, it is evident that the combination of SV-MMSE and KSVD filtering procedure allows a significant improvement of noise reduction in interferogram among all. To better analysing the experimental results of our approach, we displayed two zoomed views (see Figure 2), marked as rectangular white box, of the local area on the interferogram. All the zoomed views of filtered interferograms show effectiveness of our approach for noise reduction in interferometric phase. It is important to highlight that the proposed strategy can be effectively applied on all the interferometric data pairs, irrespective of decorrelation effects, and also with image pairs having large spatial baseline.

References:

[1] G. Franceschetti and G. Fornaro, “SAR interferometry,” in Synthetic Aperture Radar Processing, G. Franceschetti and R. Lanari, Eds. New York: CRC, 1999, ch. 4.

[2] R. Bamler and P. Hartl, “Synthetic aperture radar interferometry,” Inv. Probl., vol. 14, pp. R1–R54, 1998.

[3] G. Fornaro, and A. M. Guarnieri, “Minimum Mean Square Error Space-Varying Filtering of Interferometric SAR Data,” IEEE Transactions On Geoscience And Remote Sensing, VOL. 40, NO. 1, January 2002.

[4]M. Aharon, M. Elad, and A. M. Bruckstein, “The K-SVD: An algorithm for designing of overcomplete dictionaries for sparse representation,” IEEE Trans. Signal Process., vol. 54, no. 11, november 2006.

[5] C. Ojha, A. Fusco, and M. Manunta, ”Denoising of Full Resolution Differential SAR Interferogram based on K-SVD Technique,” IGARSS 2015.

Sentinel-1A (S1A) is the first of a family of satellites designed to provide a satellite data stream for the European environmental monitoring program Copernicus, previously known as GMES (Global Monitoring for Environment and Security). The S1A satellite, which has been launched on April 3, 2014, provides scientific community with C-Band SAR data collected in continuity with the first generation ERS-1/2 and ENVISAT missions, guaranteeing further enhancements in terms of revisit time, coverage, timeliness and reliability of service. In particular, the SAR Instrument is designed to operate over land with the innovative acquisition mode referred to as Terrain Observation with Progressive Scans (TOPS), by means of which S1A Interferometric Wide Swath (IWS) scenes are collected. TOPS mode is quite similar to the ScanSAR one, since during the acquisition time the antenna beam is switched cyclically among different sub-swaths, allowing a significant improvement of the range coverage at expenses of the azimuth resolution.

The IWS acquisitions are specifically collected to carry out interferometric analyses, through Differential SAR Interferometry (DInSAR) technique, to analyze and investigate Earth’s surface displacements.

This work is aimed at describing the development of an advanced and efficient interferometric processing chain, based on the well-known DInSAR algorithm referred to as Small BAseline Subset (SBAS), for the generation of S1A IWS deformation time-series. In particular, the high data stream expected by S1A and the upcoming twin system Sentinel-1B, together with the big size of the data (around 10 times greater than ERS and ENVISAT scenes), make increasingly important the computing efficiency of the DInSAR processing chains.

In this framework, the pursued SBAS-DInSAR processing strategy strongly takes into account the data acquisition characteristics of the TOPS mode. Indeed, IWS scenes consist of series of bursts that can be considered as separate acquisitions. This makes a large part of the processing to be easily performed in parallel, at a burst granularity level, thus implying a significant processing time reduction when large computing resources are available. The last aspect is of high importance/relevance in the use of SBAS-DInSAR processing chain in operational contexts, where dealing with large amounts of data represents a challenging task. However, one of the main issues occurring when processing large DInSAR datasets, which are characterized by several hundreds of SAR acquisitions and interferograms, is related to the very high network and Input/Output (I/O) capabilities required to overcome the limitation due to such a big data transfer and sharing among different computing resources. Such a bottleneck becomes particularly critical when a high number of processing nodes is exploited for the computation. Indeed, in this case, tens or hundreds of parallel jobs that work concurrently have to access the shared storage resource, thus causing a saturation. The presented SBAS-DInSAR processing chain, properly designed to process large IWS Sentinel-1 datasets, allows us to minimize the above-mentioned limitations and achieve good scalable performances without the mandatory need of high performance computing resources. It, indeed, benefits from both a proper design of the computing architecture and an ad-hoc job scheduling aimed at distributing the data storage among different nodes, thus optimizing the parallel performances.

The proposed SBAS-DInSAR processing chain has been properly designed to massively process IWS Sentinel-1 data on a continuous basis, and can represent the starting point of an advanced Sentinel-1 service able to provide users with continuously and systematically updated deformation time-series of very large areas on the Earth’s surface.