The Integrated Wide Area Processor (IWAP): A Processor for Wide Area Persistent Scatterer Interferometry
Rodriguez Gonzalez, Fernando; Adam, Nico; Parizzi, Alessandro; Brcic, Ramon
German Aerospace Center (DLR), GERMANY

Over the last decade, Persistent Scatterer Interferometry (PSI) [1][2] has established itself as a scientific and commercial remote sensing technology. In preparation for the upcoming Sentinel-1 mission, the concept of the PSI Wide Area Product (WAP) was introduced in the framework of ESA's Terrafirma project [3]. The concept emerged from a user requirement for the monitoring of volcanic and seismogenic areas, coastal lowland, landslides in mountain areas and deformation caused by mining and ground water regulation on a small scale basis.

Originally, the PSI technique was developed for urban areas, where the density and quality of persistent scatterers (PSs) is high. The extension to wide area processing presents many technical challenges. On the one hand, the distribution and quality of PSs are lower and highly variable in rural areas and especially in mountain areas. Moreover, shadowing and layover may well limit the areas that can be processed. As a consequence and in order to increase coverage as much as possible, robust algorithms for PS selection, network construction and inversion are required [4][5][6][7]. On the other hand, the atmospheric phase screen contains an additional vertical stratification effect in mountainous areas. Different approaches have been developed in order to mitigate this significant effect on the PSI estimation [6][7][8]. Finally, a WAP is generated by aggregation of PSI processings over multiple frames. A GPS absolute calibration approach has been developed for this purpose.

Further from these technical difficulties, the generation of a WAP constitutes a major effort in terms of processing and planning. It requires a high degree of automation, robustness and quality control of the overall process. On the basis of the in house developed PSI-GENESIS [9] software and following our experience in the development of operational processors, a new processor was designed and implemented: the Integrated Wide Area Processor (IWAP). The IWAP has been conceived as a highly-automatized, efficient and robust multi-sensor processor for the processing of high amounts of data. At the moment, two processing scenarios have been implemented: InSAR stacking and PSI processing.

Previous phases of the Terrafirma project have shown that one of the major costs of PSI processing is operator time. Furthermore, our first experiences showed us that this factor was even more critical for the generation of WAP products. As a consequence, two main design principles were introduced in the processor design: (i) minimization of the number of user interactions (reduced to only some important decisions) and (ii) development of visual interfaces to support the user decision process. The results are illustrated by the InSAR stacking scenario depicted in Figure 1. The user interactions are reduced to: i) preparing an order file (an ASCII file with some administrative and configuration parameters), ii) calling the processor and iii) selecting the master scene (with support of a master selection visual interface).

Figures 1. IWAP InSAR stacking scenario: (left) user workflow, (right) master selection interface. In order to perform InSAR stacking the user must only prepare an ASCII order file for the processor and the input SAR SLCs. The order file specifies the parameters for the processor configuration. After processor invocation, the processor works standalone, except for one interaction: the master selection. The user receives a request for interaction and the processor provides a visual interface which supports the decision making process.

In the last year of the Terrafirma project, two WAPs have been generated with the IWAP: a GPS calibrated WAP of Greece and a non-calibrated WAP of Turkey (see Figure 2). In order to generate the Greece WAP, 671 ERS SLCs acquired between 1992 and 2003 in 10 different frames have been processed. The product covers an area of around 65000 km2 and contains over a million persistent scatterers. The average PS density in each of the frames ranges from 8 to 40 PS/km2, with a total average of 16 PS/km2. Nevertheless, in urban areas the WAP product preserves the high PS density characteristic of PSI products. For instance, in Athens the PS density is around 200 PS/km2.

Figures 2. WAPs processed in the last year of Terrafirma's project: (left) GPS absolute calibrated Greece WAP, (right) non-calibrated Turkey WAP. The Greece WAP has been generated from 671 ERS SLCs in 10 frames. The GPS stations used for the absolute calibration are depicted as purple points. The Turkey WAP has been generated from 312 ERS SLCs in 6 frames. Background images: Google Maps.

Until now, the IWAP fully supports processing of ERS and Envisat/ASAR data. Currently, ALOS/PALSAR-1 and TerraSAR-X are been integrated into the workflow and tested. Moreover, an interface to the Troposphere Effect Mitigation Processor (TEMP) is being built in order to mitigate the effect of the tropospheric delay [7]. This paper will report on the design approach of the IWAP processor, its processing workflow, the current status of its development and some of the updated processing algorithms. Moreover an overview of the WAP products generated in the framework of the Terrafirma project will be given.

Acknowledgements

The WAP has been developed at DLR within the ESA project Terrafirma with the ESRIN/Contract no. C19366/05/I-EC/DLR.

References

[1] Ferretti, A.; Prati, C. and F. Rocca, "Permanent Scatterers in SAR Interferometry", Proc. of IGARSS 1999, Hamburg, 1999.
[2] Ferretti, A.; Prati, C. and F. Rocca, "Permanent Scatterers in SAR Interferometry", IEEE Transactions on Geoscience and Remote Sensing, vol. 38, pp. 2202 2212. 2000.
[3] Terrafirma project's home page: http://www.terrafirma.eu.com/index.htm
[4] Liebhart, W.; Adam, N.; Rodriguez Gonzalez, F.; Parizzi, A. and X-Y. Cong, "Four Level Least Square Adjustments in Permanent Scatterers Interferometry for the Wide Area Product", Proc. of IGARSS 2012, Munich, 2012.
[5] Rodriguez Gonzalez, F.; Bhutani, A. and N. Adam, "L1 network inversion for robust outlier rejection in persistent scatterer interferometry", Proc. of IGARSS 2011, Vancouver, 2011.
[6] Adam, N.; Rodriguez Gonzalez, F.; Parizzi, A. and Liebhart, W., "Wide area persistent scatterer interferometry", Proc. IGARSS, 2011, Vancouver, 2011.
[7] Adam, N.; Rodriguez Gonzalez, F.; Parizzi, A. and Brcic, R., "Wide area persistent scatterer interferometry: current developments, algorithms and examples", submitted to IGARSS, 2013.
[8] Cong, X.; Eineder, M.; Fritz, T.; "Atmospheric delay compensation in differential SAR interferometry for volcanic deformation monitoring - study case: El Hierro," Proc. IGARSS, 2012, Munich, Germany, 2012.
[9] Adam, N.; Kampes, B. M.; Eineder, M.; "The development of a scientific persistent scatterer system: Modifications for mixed ERS/ENVISAT time series", ENVISAT/ERS Symposium, Salzburg,2004