Christian Bignami

Satellite radar and optical remote sensing for earthquake damage detection: results from different case studies

In case of a seismic event, a fast and draft damage map of the hit urban areas can be very useful, in particular when the epicentre of the earthquake is located in remote regions, or the main communication systems are damaged. Our aim is to analyse the capability of remote sensing techniques for damage detection in urban areas and to explore the combined use of radar (SAR) and optical satellite data. Two case studies have been proposed: Izmit (1999; Turkey) and Bam (2003; Iran). Both areas have been affected by strong earthquakes causing heavy and extended damage in the urban settlements close to the epicentre. Different procedures for damage assessment have been successfully tested, either to perform a pixel by pixel classification or to assess damage within homogeneous extended areas. We have compared change detection capabilities of different features extracted from optical and radar data, and analysed the potential of combining measurements at different frequency ranges. Regarding the Izmit case, SAR features alone have reached 70% of correct classification of damaged areas and 5 m panchromatic optical images have given 82%; the fusion of SAR and optical data raised up to 89% of correct pixel‐to‐pixel classification. The same procedures applied to the Bam test case achieved about 61% of correct classification from SAR alone, 70% from optical data, while data fusion reached 76%. The results of the correlation between satellite remote sensing and ground surveys data have been presented by comparing remotely change detection features averaged within homogeneous blocks of buildings with ground survey data.

Uplift and subsidence due to the 26 December 2004 Indonesian earthquake detected by SAR data

The Indonesian earthquake took place on 26 December 2004 at 00:58 GMT (moment magnitude 9.3) in the Indian Ocean, offshore the west coast of Sumatra, at a depth of about 30 km. This earthquake is one of the largest of the past 100 years, comparable only with those in Chile (1960) and Alaska (1964). The earthquake originated in the subduction zone of the Indian and Burma plates, moving at a relative velocity of 6 cm/year. The aftershocks were distributed along a plate boundary of about 1000–1300 km between Sumatra and the Andaman Islands. Some hours after the earthquake a destructive tsunami followed and hit the coastlines of the surrounding regions, causing widespread destruction in Indonesia, India, Thailand and Sri Lanka. The European Space Agency (ESA) made available a data package composed of European Remote Sensing Satellite Synthetic Aperture Radar (ERS‐SAR) and Environment Satellite Advanced SAR (ENVISAT‐ASAR) data covering the affected area, acquired before (four acquisition dates) and after (five acquisition dates) the earthquake. A total of 26 frames were analysed. We used this dataset to evaluate the effects of the earthquake and tsunami on the human settlements and the physiographic conditions along the coast. The proposed method is based on a visual comparison between pre‐ and post‐seismic SAR intensity images, and on an analysis of their correlation coefficients. No complex data were made available by the ESA to exploit phase coherence. Analysis of pre‐ and post‐earthquake SAR backscattering showed wide uplift areas between the Andaman Islands and Simeulue Island, and large modifications of the coastline of Sumatra. Subsiding areas were detected along the southeast coast of Andaman up to the west coast of Nicobar Island. Tidal effects were filtered out of the SAR images to identify the consequences of the earthquake. Global Positioning System (GPS) stations in the Andaman provided results confirming the surface displacement pattern detected by SAR. The analysis enabled us to draw a boundary line separating the uplift and subsidence.

The May 12, 2008, (Mw 7.9) Sichuan Earthquake (China): Multiframe ALOS-PALSAR DInSAR Analysis of Coseismic Deformation

A destructive (Mw 7.9) earthquake affected the Sichuan province (China) on May 12, 2008. The seismic event ruptured approximately 270 km of the Yingxiu-Beichuan fault and about 70 km of the Guanxian-Anxian fault. Surface effects were suffered over a wide epicentral area (about 300 km E-W and 250 km N-S). We apply the differential synthetic aperture radar interferometry (DInSAR) technique to detect and measure the surface displacement field, using a set of ALOS-PALSAR L-band SAR images. We combine an unprecedented high number of data (25 frames from six adjacent tracks) to encompass the entire area which has coseismically displaced. The resulting mosaic of differential interferograms covers an overall area of about 340 km E-W and 240 km N-S. We investigate the source of the Sichuan earthquake by modeling the DInSAR data. The geometry and position of the fault parameters are inferred by a nonlinear inversion, followed by a linear inversion to retrieve the relative slip distribution. Our results show two different source mechanisms for the 145-long Yingxiu-Beichuan fault and for the 105-long Beichuan-Qingchuan fault. Both faults are characterized by slip concentrations of up to 8 m.

Monitoring Soil Moisture in an Agricultural Test Site Using SAR Data: Design and Test of a Pre-Operational Procedure

An algorithm for pre-operational high resolution soil moisture mapping using Synthetic Aperture Radar (SAR) data is presented. It has been conceived to be inserted in the operational weather alert system of the Italian Department of Civil Protection. The Maximum A Posteriori (MAP) probability criterion is applied to retrieve soil moisture by inverting a forward backscattering model, and ancillary data such as optical images and land cover maps are also used to identify areas in which the retrieval can be carried out. The well-established semiempirical water cloud model is adopted to correct for the effect of vegetation on SAR data. In anticipation of the use of the algorithm in an operational system, in which the SAR-derived high resolution soil moisture product can be assimilated within weather prediction models or hydrological ones, an uncertainty index is associated to each estimate. The algorithm has been tested on a dataset consisting of ground data gathered for seven years (2003-2010) on an agricultural test site in Northern Italy and radar data provided by the C-band ENVISAT/ASAR instrument. A comparison, performed at field scale, between estimated and in situ soil moisture data has shown that, by discarding the estimates with the largest uncertainty, the correlation coefficient can exceed 0.80 and the root mean square estimation error is less than 0.05 m3/m3. Moreover, the uncertainty index has turned out to be fairly correlated to the actual estimation error.

X-, C-, and L-Band DInSAR Investigation of the April 6, 2009, Abruzzi Earthquake

This letter compares the coseismic deformation maps obtained from different synthetic aperture radar (SAR) sensors using the well-known differential SAR interferometry technique. In particular, four deformation maps have been obtained from X-, C-, and L-band SAR sensors onboard COSMO-SkyMed, Envisat, and ALOS satellite missions correspondingly. The test case is the April 6,2009, earthquake (Mw = 6.3). This seismic event struck a densely populated region of the Apennines and was felt all over Central Italy. The SAR data set is rather inhomogeneous, since it includes interferograms with three different wavelengths, four acquisition geometries, different spatial resolutions, variable temporal and spatial baselines, and differently emphasized signal noise. However, we find that the detected displacements are highly comparable. The outcome of this work is that, even though such differences have an impact on the properties of the interferograms, the displacements can be measured with an overall discrepancy of about half the value of the shortest wavelength (COSMO-SkyMed) data set.

Did the September 2010 (Darfield) earthquake trigger the February 2011 (Christchurch) event

We have investigated the possible cause-and-effect relationship due to stress transfer between two earthquakes that occurred near Christchurch, New Zealand, in September 2010 and in February 2011. The Mw 7.1 Darfield (Canterbury) event took place along a previously unrecognized fault. The Mw 6.3 Christchurch earthquake, generated by a thrust fault, occurred approximately five months later, 6 km south-east of Christchurchs city center. We have first measured the surface displacement field to retrieve the geometries of the two seismic sources and the slip distribution. In order to assess whether the first earthquake increased the likelihood of occurrence of a second earthquake, we compute the Coulomb Failure Function (CFF). We find that the maximum CFF increase over the second fault plane is reached exactly around the hypocenter of the second earthquake. In this respect, we may conclude that the Darfield earthquake contributed to promote the rupture of the Christchurch fault.

Multisensor Satellite Monitoring of the 2011 Puyehue-Cordon Caulle Eruption

This paper shows the main outcomes of the Puyehue volcano (Chile) eruption monitoring by means of multisensor remote sensing instruments working from thermal infrared (TIR) to microwave (MW) spectral range. Thanks to the use of Synthetic Aperture Radar (SAR) and the Moderate Resolution Imaging Spectroradiometer (MODIS), the eruption evolution was observed, capturing the deformations of volcano edifice, the lava extension, as well as the information on ash and gas emitted. On the one hand, SAR Interferometry applied to ENVISAT-ASAR data allowed the estimation of the deformation occurred just before the beginning of the eruption and the subsequent deflation, with monthly sampling. On the other hand, with the combined use of the very high-resolution (VHR) images taken by COSMO-SkyMed X-band SAR, and ENVISAT-ASAR ones, we were able to follow the lava deposition during the most intense phase of the eruption. Additionally, the joined exploitation of SAR and optical MODIS images allowed ash detection, also in cloudy sky conditions. Finally, the information gathered by both types of sensors allowed to highlight some volcanological features of the eruption and the relationship between surface deformation and the amount of ash and gases emitted by the volcano.

X- and C-Band SAR Surface Displacement for the 2013 Lunigiana Earthquake (Northern Italy): A Breached Relay Ramp?

In this paper, we applied the differential interferometric synthetic aperture radar (DInSAR) technique to investigate and measure surface displacements due to the Mw 5.3 (M1 5.2), June 21, 2013 earthquake, occurred north of the Apuan Alps (NW Italy), in the discontinuity zone between the Lunigiana and Garfagnana area. Two differential interferograms showing the coseismic displacement have been generated using X-band and C-band data, taken from COSMO-SkyMed and RADARSAT-2 satellites, respectively. Both interferograms highlighted a clear pattern of subsidence of few cm located between the Lunigiana and Garfagnana basins. We then modeled the observed SAR deformation fields using the Okada analytical formulation and found them to be consistent with an extensional fault plane dipping toward NW at about 50 . The integrated analysis of DInSAR, geological data, modeling, and historical seismicity suggest that the fault responsible for the June 2013 earthquake corresponds to a breached relay ramp connecting the Lunigiana and Garfagnana seismogenic sources.

The Calibration of the Envisat Radar Altimeter Receiver by a Passive Technique

The passive calibration of the Radar Altimeter (RA) consists of characterizing the receiver gain by observing natural surfaces with known emission in the so-called noise-listen mode. It is based on the comparison between the simulated values of the brightness temperature impinging on the altimeter antenna, and the digital counts at the output of the altimeter receiver in the absence of echo. The proposed method aims to calibrate measurements of the backscattering coefficient performed by a spaceborne altimeter based on the assumption that the receiver gain is the main source of uncertainty. This paper focuses on the general approach undertaken to characterize the receiver and to simulate the brightness temperature at the top of the atmosphere observed by the Envisat RA-2. The simulations rely on emissivity models for land and sea, as well as on atmospheric radiation models supported by a continuous flow of online data used as model inputs. To assess the accuracy, the model outputs are compared with observations from calibrated radiometers, namely the Special Sensor Microwave/Imager and Tropical Rainfall Measuring Mission Microwave Imager, with particular attention to the low-frequency channels (10 and 19 GHz). The new method has been first tested on European Remote Sensing satellite data and has been subsequently adopted for Envisat RA-2 in the framework of the Envisat Calibration and Validation activities managed by European Space Agency. The evaluation of the receiver gain at both Ku-band and S-band is presented and compared to the preflight values, as well as to transponder calibration done for Ku-band. An error budget for the final estimates is also presented and discussed

The October 23, 2011, Van (Turkey) earthquake and its relationship with neighbouring structures

The present work reports the analysis of a possible relationship due to stress transfer between the two earthquakes that hit the province of Van, Eastern Turkey, on October 23, 2011 (Mw = 7.2) and on November 9, 2011 (Mw = 5.6). The surface displacement field of the mainshock has been obtained through a combined data set made up of differential interferograms from COSMO-SkyMed and ENVISAT satellites, integrated with continuous GPS recordings from the Turkish TUSAGA-AKTIF network. This allowed us to retrieve the geometry and the slip distribution of the seismic source and to compute the Coulomb Failure Function (CFF) variation on the aftershock plane, in order to assess a possible causal relationship between the two events. Our results show that the November 9 earthquake could have been triggered by the October 23 shock, with transferred stress values largely exceeding 1 bar.