C. De Luca

Ground deformation and source geometry of the 24 August 2016 Amatrice earthquake (Central Italy) investigated through analytical and numerical modeling of DInSAR measurements and structural-geological data

We investigate the ground deformation and source geometry of the 2016 Amatrice earthquake (Central Italy) by exploiting ALOS2 and Sentinel-1 coseismic differential interferometric synthetic aperture radar (DInSAR) measurements. They reveal two NNW-SSE striking surface deformation lobes, which could be the effect of two distinct faults or the rupture propagation of a single fault. We examine both cases through a single and a double dislocation planar source. Subsequently, we extend our analysis by applying a 3-D finite elements approach jointly exploiting DInSAR measurements and an independent, structurally constrained, 3-D fault model. This model is based on a double fault system including the two northern Gorzano and Redentore-Vettoretto faults (NGF and RVF) which merge into a single WSW dipping fault surface at the hypocentral depth (8 km). The retrieved best fit coseismic surface deformation pattern well supports the exploited structural model. The maximum displacements occur at 5–7 km depth, reaching 90 cm on the RVF footwall and 80 cm on the NGF hanging wall. The von Mises stress field confirms the retrieved seismogenic scenario.

Sentinel-1 results: SBAS-DInSAR processing chain developments and land subsidence analysis

This work is aimed at describing the development of an efficient interferometric processing chain, based on the well-known advanced Differential Interferometric Synthetic Aperture Radar (DInSAR) algorithm referred to as Small BAseline Subset (SBAS) technique, for the generation of Sentinel-1A (S1-A) Interferometric Wide Swath (IWS) deformation time-series. Due to the TOPS mode characterizing the IWS acquisitions, the existing SBAS processing chains was properly adapted with new procedures for efficiently handling the S1-A data. The developed SBAS-DInSAR chain has been tested on both S1-A and TOPS RadarSAT-2 interferometric dataset, clearly demonstrating the capability of the developed SBAS-DInSAR processing chain to effectively investigate land subsidence phenomena affecting large areas.

SCALABLE PERFORMANCE ANALYSIS OF THE PARALLEL SBAS-DINSAR ALGORITHM

The effective exploitation of the available huge SAR data archives in reasonable time-frames has motivated the development of P-SBAS, a parallel computing solution for the SBAS (Small BAseline Subset) processing chain. Hence, P-SBAS parallel solution represents a valuable tool for the analysis of the complex phenomena characterizing the surface deformation dynamics of Earth large areas, since it permits to exploit the parallelism offered by the modern computational platforms. In this paper, the performance of the parallel algorithm P-SBAS is investigated. The quantitative evaluation of the computational efficiency of the implemented parallel prototype in terms of achieved speedup is addressed to demonstrate the effectiveness of the proposed approach. An experimental analysis has been carried out on real data by employing a computational platform comprising 32 processors. In particular, the performance analysis has been conducted by exploiting different SAR datasets pertinent to different sensors (Envisat and Cosmo Sky-Med) and the factors limiting the inherent scalability are discussed.

Unsupervised parallel SBAS-DInSAR chain for massive and systematic Sentinel-1 data processing

In this work we present an efficient interferometric processing chain, based on the advanced DInSAR algorithm referred to as Parallel Small BAseline Subset (P-SBAS), for the generation of Sentinel-1A (S1A) Interferometric Wide Swath deformation time-series, which is able to exploit distributed computing architectures. The presented S1A P-SBAS processing chain has been successfully implemented within the ESA Geohazard Exploitation Platform to provide an on-demand automatic service for the unsupervised generation of P-SBAS displacement time-series. To give an idea of the effectiveness of the presented S1A processing chain, as a preliminary result we show a 12-days interferometric analysis at continental scale, carried out by exploiting 150 S1A interferometric pairs acquired over Europe for an overall covered area of about 7,500,000 km2.

New advances in intensive DInSAR processing through cloud computing environments

The current Remote Sensing scenario is characterized by the availability of huge archives of SAR data that are going to increase with the advent of Sentinel-1 satellites. The effective exploitation of this large amount of data requires both adequate computing resources as well as advanced algorithms able to properly exploit such facilities. In this work we discuss the migration of the DInSAR technique referred to as Parallel Small BAseline Subset (P-SBAS), which is used for Earths surface deformation investigation, to the Amazon Web Services (AWS) public Cloud Computing environment. An experimental analysis aimed at evaluating the P-SBAS scalable performances that are achieved within the Cloud environment is presented. The achieved results show very good parallel performances and allow us to identify the major bottlenecks that can hamper such behavior when the amount of data to process highly increases. Accordingly, we present an advanced P-SBAS implementation that is designed to overcome the identified bottlenecks. The experimental analysis is carried out by processing both Envisat and COSMO-SkyMed datasets and by exploiting both a High Performance Computing cluster as well as AWS public Cloud.

THREE-DIMENSIONAL GROUND DISPLACEMENTS RETRIEVED FROM SAR DATA IN A LANDSLIDE EMERGENCY SCENARIO

This work presents the Differential SAR Interferometry and pixel-offset analysis on the event landslide that struck Montescaglioso town (Matera, southern Italy) on December 3rd, 2013. The event occurred after adverse weather conditions that produced a ground displacement of several meters, causing a severe emergency situation. The analysis has shown the presence of two main directions of motion: a major and a minor movement along the South-SouthWest and South-SouthEast directions. The pixel-offset results are well in agreement with both the magnitude and the deformation mechanisms that have been identified and mapped during field observations.

Unsupervised on-demand web service for DInSAR processing: The P-SBAS implementation within the ESA G-POD environment

This paper presents the integration of the advanced Differential SAR Interferometry (DInSAR) algorithm referred to as Parallel Small BAseline Subset (P-SBAS) within the ESAs Grid Processing on Demand (G-POD) environment in the framework of ESA Geohazards Exploitation Platform (GEP). The aim of this activity is to set up a scientific service that allows, in unsupervised manner, the generation of SBAS-DInSAR products, such as surface mean deformation velocity map and the corresponding time series. In particular, such a web tool is aimed at efficiently exploit the huge ESAs SAR data archives (ERS and ENVISAT), giving a support to scientific users, especially those non-expert of SAR data processing, for interferometric analysis in a short time frame.

Big DInSAR data processing through the P-SBAS algorithm

In the present radar remote sensing scenario, the huge availability of SAR data acquired by a number of satellite constellations is a key point for investigating Earths surface at large scale. In particular, techniques as Differential SAR Interferometry (DInSAR) could strongly benefit from such data availability for measuring ground displacement at global scale. However, to efficiently and effectively exploit such amount of Big DInSAR Data, the development of new algorithms and techniques as well as the use of appropriate High Performance Computing infrastructure is becoming mandatory. In this work we present a number of case studies based on the recently proposed Parallel Small BAseline Subset (P-SBAS) DInSAR algorithm, that allows generating surface displacement maps in automatic and unsupervised way, aimed at providing effective solutions to the increased DInSAR data volume processing.

Time series of SAR image fractal maps

In this paper the analysis of time series of fractal dimension maps generated from multi-pass SAR images is dealt with. The objective is twofold: on the one hand such an analysis is addressed to assess the performance of the algorithm for the fractal dimension estimation from a single SAR image, i.e., to verify the fractal estimation efficiency on natural scenes within the multiple sensor passes. On the other hand, by exploiting the statistics evaluated starting from the fractal time series, a final fractal dimension map, including only the areas in which the fractal estimation is efficient and strongly denoised from the speckle effect, is obtained. The analysis has been performed on a Cosmo-SkyMed data-set of 42 stripmap images spanning the time period from October 2009 to December 2012, acquired over the Somma-Vesuvius volcanic complex (South Italy), which is in a quiescent stage since the last eruption occurred in 1944.