Ecv ice Sheets Technical Proposal

1.2Innovative aspects

The proposed feasibility study addresses both of the innovation-related objectives layed out in section 2.3 of [DA 1]. These are discussed in the following two sub-sections.

Novel products and innovative retrieval methods shall include:

• 
  Novel interseismic deformation product time-series of the areas of interest (see Section 3.1.3). This product shall be novel, since it will be generated by novel SAR processing methods compared to those previously used for the same areas. These include an innovative general-purpose retrieval algorithm for the measurement of the north-south deformation component from SAR; an enhanced SBAS algorithm with the potential of providing an increased spatial coverage; an innovative InSAR-GPS integration method for the generation of the deformation products required by PSHA models.
  
• 
  Deformation-based seismic hazard (ERF) model of the investigated area. Compared to the current official seismic hazard model of Italy (www.zonesismiche.mi.ingv.it, 2004; Stucchi et al., 2011), which follows the standard Cornell’s approach with area source zones to derive the probability of occurrence of earthquakes, the generated product shall be novel, since it will be based on a combination of tectonic and geodynamic information with geodetic data, allowing seismicity rates to be calculated from earthquake-independent data.
  

New earth science results shall be obtained by contributing to the understanding of the marginal contribution of existing and future SAR archives compared to GPS in improving PSHA models in poorly understood plate boundary zones, characterized by a good availability of both data sources.

2Study plan

An overview of the study plan for this project is given in Figure . The workload is divided into four technical Work Packages (WPs), namely WPs 1000, 2000, 3000 and 4000, and one WP for project management, namely WP 5000.

Figure : Project study plan

WP 1000 (Requirement analysis and product specification) fullfills Tasks 1 and 2 of the Statement of Work [DA 2]. It consists of 3 sub-WPs, namely WP 1100, 1200 and 1300. In WP 1100 hazard model generation requirements shall be collected, in particular concerning the algorithms to exploit surface deformation measurements to derive recurrence relations for a given source and hazard model validation approaches. The relevant input deformation parameters shall be identified and their required accuracy as well as spatial and temporal sampling shall be identified. These specifications shall serve as a goal for WP 1200, in which available GPS and InSAR techniques as well as integration methods shall be surveyed. Within WP 1300 two datasets shall be identified: a development and validation dataset (Task 2 of [DA 2]) and an experimental dataset (Task 4 of [DA 2]). The former shall also be procured within the WP timeframe. WP 1000 shall be closed within the first 3 project months (Progress Meeting 1, PM1), as required in Section 5.3 of [DA 2].
WP 2000 (Development and Validation) fulfills Task 3 of the SoW. Four sub-WPs shall run in parallel: the investigation of a multitemporal technique based on Multi Aperture InSAR for the measurement of the north-south deformation component (WP 2100); the investigation of a technique to improve the spatial coverage of the current Small BAseline Subset (SBAS) algorithm (WP 2200); development of SAR and GPS integration methods for the generation of the deformation products required by hazard models (WP 2300); the development of modeling techniques to exploit available surface deformation products (WP 2400). For all sub-WPs, the development dataset procured in WP 1300 shall be used.
WP 3000 (Prototype Demonstration and Impact Assessment) fulfills Task 4 of the SoW. The target deformation products and hazard model products shall be generated and validated in sub-WPs 3100 and 3200 respectively.
WP 4000 (Scientific Roadmap) corresponds to Task 5 of the SoW.

3Detailed Workplan

3.1WP1000 - Requirements analysis and product specification

3.1.1WP1100 - Hazard model generation requirements

In a probabilistic seismic hazard analysis, the aim is to determine probabilities of exceedance of given ground-motion levels over time windows of interest (e.g. depending on the building life expectancies or its significance). Under the Poisson hypothesis of stationarity of seismic energy release, the estimation of the probabilities of exceedance is equivalent to estimating the return periods (or annual rates) of the ground-motion levels.

The main output of a PSHA study at a given site is a hazard curve, displaying probabilities of exceedance (or alternatively annual rates) versus ground-motion levels. Generally, for a larger scale (such as national or global scale) seismic hazard is presented in a map showing the distribution of shaking levels that have a certain probability to be exceeded in 50 years, on hard ground.

PSHA studies are performed mainly in order to contribute to the definition of risk reduction strategies: the most direct applications of PSHA are the zonation of seismic areas (for planning and regulatory applications) and the definition of response spectra to be adopted in designing constructions and utilities that will withstand shaking due to earthquakes.

This is made possible because the most common output of the seismic hazard (a map showing the PGA with 10% of probability of exceedance in 50 year) is a piece of information that is complemented by the seismic hazard elaboration for assessing PGA, calculated for several probabilities of exceedance in 50 years, and spectral accelerations for varied spectral periods and exceedance probabilities.

To follow a conventional PSHA scheme, the seismicity is modelled in several seismic source zones where the earthquake is considered uniformly distributed in space and the recurrence model follows a Poisson distribution. For example, in Italy, the current source zone model divides the national territory into 36 seismic source zones, covering the areas of the country where earthquake with magnitude larger than 5 are expected; the earthquake rate model that gives the long-term rate of all earthquakes throughout the region above a specified magnitude threshold was obtained considering all the earthquake occurred in each zone and distributed according the Gutenberg-Richter empirical relationship. Finally, using a set of ground-motion predictive equations the expected value of ground shaking has been computed (Stucchi et al., 2011).

This is the approach most adopted worldwide; other kinds of seismic source can be used, even if in general they are not available everywhere with the definition required for an application in PSHA. For example Field et al. (2009) in California used three types of sources: known active faults, zones of distributed shear and smoothed seismicity to account for unknown faults or area sources. However, any kind of source has to be accompanied by an earthquake recurrence model that allows to compute the probability that each earthquake will occur during a specified time window.

Within this WP the main tasks shall be to identify a procedure to exploit deformation measurements in the definition of the seismic source zones, reducing the margin for subjective assumptions in their definition. Secondly, methods for slip and strain-rate calculation (fault and off-fault) shall be reviewed. Finally, existing methods to convert long term slip- and strain-rates to earthquake rates shall be analyzed.

3.1.2WP1200 - Deformation measurement requirements

Concerning GPS measurements, currently Italy hosts over 600 permanent GPS stations that provide a temporally continuous and spatially point-like surface deformation dataset of the whole country. The first attempt to build a nation-wide continuous GPS network was undertaken by the Italian Space Agency (ASI) in the late 1990’s. In 2001, the Istituto Nazionale di Oceanografia e Geofisica Sperimentale (OGS) started installing a local GPS network in the Friuli region (Northeast Italy) and in 2004, the Istituto Nazionale di Geofisica e Vulcanologia (INGV) established the national GPS network (RING). More recently an increasing number of permanent GPS sites have been installed by regional administrations and private companies, dedicated mainly to topographic applications and commercial services. These networks, although not conceived to measure long term ground deformations, have proven to be useful in augmenting the network geometry and their data are currently available for scientific studies. All these GPS observations are currently archived and processed at INGV using Bernese v. 5.0 software, following international EUREF guidelines for the processing strategy. Daily GPS time series are routinely generated at INGV for all available stations in Italy and the velocity field is estimated at each GPS station position (e.g. Devoti et al. 2011). At present, more than 500 velocity vectors can be estimated from time series spanning up to 14.7 years (5382 days), the mean temporal length being 5.6 years (2046 days).

GPS velocity fields obtained with different approaches and software procedures have proven to agree at the level of 0.3 mm/yr (Avallone et al. 2010), which corresponds to the order of magnitude of the velocity error. Therefore the GPS velocity fields provided by INGV can be considered as a standard and validated product.

From the SAR point of view, several different datasets shall be available (see following section), partially overlapping in space and time among eachother and with respect to the GPS measurements, so that a processing approach will have to be established. For instance, long data strips which cover several well distributed GPS stations could be processed first and then used for the calibration of intersecting datasets with poorer GPS coverage. To this end a feasibility study to estimate the expected coverage of the areas of interest with state-of-the-art techniques shall be carried out.

3.1.3WP1300 - Dataset collection

The aim of this WP is the definition of the areas of interest for the project and the selection of suitable EO based products, in situ data and relevant ancillary information over these areas. Two datasets shall be identified: a first (smaller) development and validation dataset, to be used for the activities of WP 2000; a second (larger) experimental EO dataset to which the implemented techniques shall be applied in WP 3000.

The main characteristics of the development and validation area should be: the availability of SAR data coverage for most of the study area (see section 4 for a discussion on the sensors which shall be considered for the project); an expected ground deformation rate detectable by multitemporal MAI and DInSAR techniques, i.e. with an expected N-S and E-W deformation of the same order of magnitude; good GPS station coverage and distribution; the possibility of applying a procedure to validate the derived PSHA models. From the seismic hazard point of view it would be of interest to consider a tectonically active well-studied area, prone to moderate-to-large earthquakes. As a proxy for such areas, the approach of Riguzzi et al., (2012) can be followed, seeking for regions exhibiting high strain-rate gradients (Figure ).

A potential test case is the Abruzzo region (Central Italy) and, in particular, the Middle Aterno basin located in the central Apennines. This area falls in the highest seismic hazard zone of Italy (http://mi.ingv.it/mappa_ps_apr04/italia.html) and, on 6 April, 2009 it was affected by a Mw 6.3 earthquake. The latter devastated the downtown area of L'Aquila and many ancient villages located along the Aterno River Valley. Using Envisat and COSMO-SkyMed SAR images and a set of horizontal velocities from 30 GPS stations, Atzori et al. 2009 measured the coseismic ground displacement and defined the seismic source causing the earthquake. Earlier, Hunstad et al. 2009 investigated the surface deformation affecting the Abruzzi region exploting two sets of SAR data from the ERS-1 and ERS-2 satellites, applying the multitemporal Small BAseline Subset (SBAS) technique (Berardino et al. 2002), detecting a NE-SW extension across the Apennines. After the 2009 earthquake, the attention of many scientist operating in different earth observation fields focused on this area, generating a large variety of scientific studies (e.g. D'Agostino et al., 2012; Giaccio et al., 2012 and reference therein). This area is charachterized by a large availability of SAR data acquired by ERS-1/2, ENVISAT ASAR, ALOS PALSAR and COSMO-SkyMed, a dense GPS network, and in situ data from geological surveys, thus making it an appealing candidate area for development and validation activities.

Concerning the experimental dataset, the selection criteria are similar to the development and validation dataset, with the additional requirement that the impact of generating improved PSHA models for this area should be significant, i.e. an area where strong earthquakes could not be excluded during next tens or hundreds of years. We propose to enlarge the canditate area for the development and validation dataset as shown in Figure , extending the study area to the Central-Southern Apennines and their slopes up to the Adriatic and Tyrrhenian sea toward NE and SW, respectively. As shown by the GPS velocity field in Figure , this area is characterized by internal deformation gradients. In fact, in addition to the diffuse northward movement of the region due to the African-Eurasian plate convergence, the GPS vectors highlight some differential velocity trend between different zones within the area. Moreover, this area could be of interest since the GPS velocity field alone does not allow discerning local scale deformation phenomena, that could instead be detected by multi-temporal SAR techniques, thus providing an improved modeling of the seismic hazard. Furthermore, the map of the second invariant of the 2D strain rate tensor estimated by Riguzzi et al., 2012 (Figure , right) shows significant gradients, the boundaries of which could be related to active faults. Additionally, the historical catalogues (e.g. Stucchi et al. 2012) can give further information on seismicity of the area and on the deformation timing. Given the above considerations, the depth of hystorical catalogue, and the huge amount of available measurements (both SAR and GPS), the Central and Southern Apennines provide an excellent candidate for the experimental dataset.

Figure : Development and experimental datasets (small and large rectangles respectively) plotted over (left) the GPS horizontal velocity field from (Devoti et al., 2013, VII Hotine-Marussi Symposium) and (right) the 2nd invariant of the strain rate tensor from Riguzzi et al., (2012)