Ph.D. in EARTH SCIENCES

Ph.D. IN EARTH SCIENCES

Tuscan Earth Science Ph.D. Program Earth Science Department, Pisa University

CYCLE XXVIII Coordinator Prof. Luca Pandolfi

TECTONICS AND MUD VOLCANISM IN THE NORTHERN APENNINES FOOTHILLS (ITALY) AND IN THE GREATER CAUCASUS (AZERBAIJAN): A SATELLITE INTERFEROMETRY (INSAR) ANALYSIS

Tettonica e vulcanismo di fango lungo il margine Pede-Appenninico emiliano e nel Gran Caucaso (Azerbaijan): un’analisi d’interferometria satellitare (InSAR)

Ph.D. student Benedetta Antonielli

Tutor Prof. Federico Sani

Co-tutors Dr. Marco Bonini

Dr. Gaia Righini

Dr. Oriol Monserrat

Academic years 2012/2015

Acknowledgments

During the three years of Ph.D. many people have contributed, in different ways, to the development of this study. The work presented in this thesis would have not been possible without the cooperation of specialists of various disciplines, from structural geology, to physical telecommunications, to geophysics. First of all, I would like to thank my tutor, Prof. Federico Sani, for giving me the opportunity to carry out this work and for representing a constant reference point, always by my side during these years. I am very grateful to Dr. Marco Bonini, co-tutor of this Ph.D., for his constant support and guidance. I have learned a lot working with them, for instance the working method and the accuracy required for the scientific research. I thank also the other two co-tutors of this Ph.D.: Dr. Gaia Righini for following my path, and for giving me helpful recommendations and opinions, and Dr. Oriol Monserrat for showing me the field of Synthetic Aperture Radar processing, and for co-operating with me with patience, optimism and cheerfulness. I also would like to thank Dr. Giacomo Corti and Dr. Giovanna Moratti for the interesting and stimulating discussions and advices, as well as for the kindness and the friendship they gave me. I am also grateful to Dr. Nicola Cenni for his help in the field of the use of GPS data. Moreover, I want to thank all the members of the Remote Sensing department of the Geomatics division of CTTC (Castelldefels, Spain) for the valuable technical advices, the kindness, and for a thousand smiles they gave me during the overall eight-months stay at the foreign institution: Dr. Guido Luzi, Dr. Núria Devanthéry Arasa, Dr. Maria Cuevas González, Dr. Michele Crosetto, Dr. Marta Agudo. Finally, I thank my parents and brother, my grandparents and friends for their unconditional support. In particular, I would like to thank my partner Marco Jan Martinelli.

Index

ABSTRACT………………………………………………………………………..….……p. 7 RIASSUNTO………………………………………………………………………..…...... p. 8

Chapter 1: General introduction and aims…………………………………….………p. 11 1.1 Aims and methodologies……………………………………………..…...p. 11 1.2 Thesis Structure………………………………………………………...... p. 14

Chapter 2: Methodologies…………………………………………………………...... p. 16 2.1 Remote sensing: introduction…………………………………………...... p. 16 2.2 Synthetic Aperture Radar………………………………………………....p. 19 2.2.1 Orbit geometry, LOS, vertical and horizontal motion……….p. 22 2.2.2 Satellite-based SAR sensors ……………………………...... p. 24 2.3 The DInSAR technique………………………………………………...... p. 25 2.3.1 Atmospheric effects and noise……………………………….p. 27 2.3.2 Phase unwrapping………………………………….……...... p. 28 2.4 The PSI technique………………………………………………….…...... p. 29 2.4.1 Key issues and processing……………………………….……p. 31 2.4.2 Comparison with existing survey techniques…………………p. 39

Chapter 3: Study of the superficial deformation of Azerbaijan mud volcanoes through the DInSAR technique………………………………………………………………………...... p. 41 3.1 Introduction………………………………………………………….…...... p. 41 3.2 Mud volcanism……………………………………………………………..p. 42 3.3 Geologic setting of the Greater Caucasus…………………………….……p. 49 3.4 Geologic setting of the Azerbaijan mud volcanoes……………………...... p. 51 3.5 Previous studies on mud volcanoes through satellite interferometry………p. 55 3.6 The used dataset and DInSAR technique…………………………………..p. 56 3.7 Results……………………………………………………………………...p. 59 3.7.1 Ayaz-Akhtarma mud volcano………………………………….....p. 60 3.7.2 Khara-Zira Island…………………………………………………p. 64 5

3.7.3 Akhtrarma-Pashaly mud volcano………………………………p. 68 3.7.4 Byandovan mud volcano……………………………………….p. 70 3.7.5 Akhtarma-Karadag mud volcano……………………………….p. 72 3.8 Discussion………………………………………………………………...p. 74 3.9 Concluding remarks………………………………………………………p. 78

Chapter 4: Study of the ground deformation of the Apennines foothills and of the Emilia thrust folds through the PSI technique……………………………………………………...p. 80 4.1 Introduction………………………………………………………………..p. 80 4.2 Geological setting…………………………………………………….……p. 82 4.2.1 The Emilia-Romagna Apennines and buried thrust and folds…….p. 78 4.2.2 Seismicity and main active structures of the outermost Northern Apennines……………………………………………….p. 92 4.2.3 The Pede-Apennine mud volcanoes………………………….……p. 95 4.3 Previous studies using satellite interferometry in the study area…...... p. 97 4.4 The used dataset and PSI technique……………………………………....p. 100 4.5 Data analysis……………………………………………….……………...p. 102 4.5.1 PSI and GPS data comparison………………………………..…..p. 104 4.5.2 PSI results overview and interpretation……………………….….p. 106 4.6 Concluding remarks……………………………………………………….p. 114 4.7 Appendix: Envisat ascending dataset……………………………………..p. 115

Chapter 5: General Conclusions……………………………………………………...p. 118 5.1 DInSAR analysis applied to mud volcanoes of Azerbaijan………………p. 118 5.2 PSI analysis of active fold-and-thrust belts: The Po Plain case study……p. 119 5.3 General remarks…………………………………………………………..p. 120

Bibliography…………………………………………………………………………p. 122

Abstract

Satellite radar interferometry provides some unique capabilities for assessing geological rates of deformation and therefore for studying active geological processes. This Ph.D. thesis employs the interferometric techniques for studying the activity of mud volcanoes and the ongoing tectonics of the related compressive structures. Mud volcanoes indeed usually develop at convergent plate margins, and occur in fold-and-thrust belts. The study areas are two orogenic fronts, namely: the Northern Apennines margin (Emilia-Romagna region; Northern Italy) and the southeastern Great Caucasus margin (Azerbaijan), both characterized by the presence of active thrust folds and mud volcanism. The latter is typically linked to hydrocarbon traps and leads to the extrusion of subsurface mud breccias that build up a variety of conical edifices. Interferometry has been applied, using Envisat images, through both (1) the Differential Satellite based Synthetic Aperture Radar Interferometry (DInSAR) approach for studying the ground deformations related to mud volcano activity, and (2) the Persistent Scatterers Interferometry (PSI) in order to investigate the ongoing tectonics along fold-and- thrust belt margins. The first goal has been developed for the Azerbaijan mud volcanoes. The results are encouraging, since it was possible to observe the mud volcanoes activity from a dynamic point of view, and to infer the processes that drive the deformation. The detected ground deformation events at mud volcanoes edifices are in general due to the fluid pressure and volume variations, and they have been observed both (i) in connection with main eruptive events, in the form of pre-eruptive uplift (~20 cm in about two years of cumulative uplift at the Ayaz-Akhtarma mud volcano), and (ii) in the form of short-lived deformation pulses that interrupt a period of quiescence. Important similarities with the deformation pattern of magmatic volcanoes have been proposed. The second goal has been carried out for the outermost sector of the Northern Apennines including the Po Plain between Piacenza and Reggio Emilia. This study attempts to correlate the superficial deformation signals measured by radar satellite-based sensor with the known geological features. The PSs velocity pattern shows some ground uplift above active thrust- related anticlines (with mean velocities ranging from 1 to 2.8 mm/yr) of the Emilia and Ferrara folds, and part of the Pede-Apennine margin. On this basis, a correlation between the observed ground uplift and the ongoing activity of the tectonic structures is proposed. The results of the current analysis would thus be taken into account when evaluating the seismic hazard of the study region. 7

Riassunto

Questo progetto di ricerca ha avuto l’obbiettivo di indagare sull’attività dei vulcani di fango presenti in contesti compressivi (catene a pieghe e thrust) e sulla tettonica delle strutture attive ad essi collegate, tramite la tecnica dell’interferometria radar satellitare. Le aree di studio sono due fronti orogenici: il margine Pede-appenninico dell’Appennino Settentrionale ed il margine sud-orientale del Grande Caucaso, entrambi caratterizzati dalla presenza di thrust attivi e del fenomeno del vulcanismo di fango. I vulcani di fango che caratterizzano il margine Emiliano-Romagnolo dell’Appennino Settentrionale, consistono generalmente in gryphons che possono raggiungere 3 o 4 m di altezza. Di dimensioni notevolmente maggiori sono invece i vulcani di fango presenti nel Gran Caucaso orientale (Azerbaijan), alcuni dei quali possono essere alti fino a 400 m, con diametro fino a 5 km, pertanto con dimensioni e caratteristiche morfologiche simili a quelle dei vulcani magmatici. I dati utilizzati per entrambe le zone di studio consistono in immagini radar Envisat, che sono state fornite dall'Agenzia Spaziale Europea (ESA) nel contesto di un progetto CAT1, creato appositamente per questo progetto di dottorato (CAT1 13866, dal titolo: Assessing the relationships between tectonics and mud volcanism by integrating DInSAR analysis and seismic data in active tectonic areas). La tecnica utilizzata in questo dottorato di ricerca è stata l’interferometria radar satellitare, che è stata impiegata attraverso due distinte metodologie: (1) l’interferometria differenziale (DInSAR) e (2) la Persistent Scatterers Interferometry (PSI). La scelta del tipo di metodologia è stata dettata da diversi fattori, in particolare (i) l’obiettivo specifico da raggiungere (studiare le deformazioni del suolo legate ai vulcani di fango oppure alle anticlinali sepolte), (ii) il numero di immagini Envisat disponibili nell’archivio dell’ESA e (iii) le caratteristiche fisiche delle due diverse aree di studio, (grado di urbanizzazione, uso del suolo, aridità) che incidono fortemente sulla risposta del dato radar. La tecnica DInSAR si è rivelata un ottimo strumento di analisi della deformazione superficiale presso i vulcani di fango, per quanto riguarda l’area di studio dell’Azerbaijan. Il periodo coperto dalle immagini Envisat utilizzate va da Ottobre 2003 a Novembre 2007. L’analisi di un ristretto set di interferogrammi selezionati in base ai migliori valori di coerenza, ha permesso di identificare importanti deformazioni superficiali in corrispondenza di 5 vulcani di fango, sia durante fasi prossime all’eruzione, sia durante fasi di normale attività di background. È stato possibile quindi osservare l’attività di questi oggetti da un punto di vista dinamico e di monitorare le deformazioni superficiali indotte da episodi di inflazione e deflazione. Tali pattern 8 deformativi mostrano chiare analogie con l’evoluzione spazio-temporale delle deformazioni del suolo di vulcani magmatici descritti in letteratura, in relazione alle varie fasi di attività. Questo risultato rafforza pertanto l’idea che i vulcani di fango e quelli magmatici siano governati da processi simili, quali le variazioni di pressione e volume dei fluidi. La tecnica PSI ha prodotto risultati di grande interesse per quanto riguarda lo studio delle strutture tettoniche attive del settore più esterno dell’Appennino Settentrionale. Il periodo coperto dalle immagini utilizzate va da settembre 2004 a settembre 2010. I risultati permettono di individuare estese deformazioni del suolo in Pianura Padana, in particolare un (i) segnale di subsidenza (allontanamento rispetto al satellite lungo la linea di vista del satellite, LOS) nella zona a nord-est di Reggio Emilia ed un (ii) segnale di sollevamento (avvicinamento rispetto al satellite lungo la LOS) in alcune zone dell’area di studio tra Reggio Emilia e Piacenza. Purtroppo la bassa risoluzione delle immagini Envisat non ha permesso l’analisi dei piccoli apparati dei vulcani di fango presenti sul margine Pede-appenninico. I dati GPS presenti nell’area di studio sono stati confrontati con i risultati della tecnica PSI. Per permetterne il confronto con le misurazioni da satellite, i dati GPS sono stati proietatti lungo la LOS (line of sight) del satellite. La subsidenza in Pianura Padana è un fenomeno molto ben conosciuto, in quanto si tratta in primo luogo di subsidenza indotta da attività antropiche, quali l’estrazione di acqua. Le aree in sollevamento, che sono l’oggetto di primario interesse di questo lavoro, mostrano deformazioni che vanno da 1 fino a picchi di 2,8 mm/anno. È interessante notare che quasi tutte le aree in sollevamento si trovano in corrispondenza di thrust attivi delle pieghe Emiliane e Ferraresi e sono caratterizzate dalla presenza di sismicità soprattutto storica. Data questa corrispondenza spaziale, avanziamo l’ipotesi (documentata dall’analisi delle serie temporali e dal confronto con le sezioni sismiche con i profili di velocità di deformazione) che esista una correlazione fra l’attività delle anticlinali sepolte sotto la Pianura Padana ed il sollevamento delle zone soprastanti queste strutture e che quest’ultime si sollevino con un movimento di creep asismico. Le faglie inverse attivate in occasione della sequenza sismica del Maggio 2012 (Finale Emilia e Mirandola) potrebbero aver avuto una simile evoluzione pre-sismica, e cioè sollevamento del suolo e scarsa sismicità. Tuttavia il fatto che alcune delle anticlinali di crescita studiate siano in sollevamento non implica che si arrivi necessariamente ad un sisma, anche se i risultati di questo lavoro dovrebbero essere presi in considerazione nella valutazione del rischio sismico dell’area di studio. Per concludere, le tecniche di interferometria satellitare hanno dimostrato di essere efficaci per studiare diversi tipi di processi geologici attivi e costituiscono un mezzo molto vantaggioso per misurare i tassi di deformazione del suolo e quindi per valutare i rischi geologici.

Chapter 1

General introduction

1.1 Aims and methodologies

This Ph.D. thesis aims to study the activity of mud volcanoes and the ongoing tectonics of the related compressive structures. Mud volcanoes indeed usually develop at convergent plate margins and occur in fold-and-thrust belts, although may be also found in extensional and strike- slip regimes. The study areas (Fig.1.1) are two orogenic fronts: the Northern Apennines margin (Emilia-Romagna region; Northern Italy) and the southern Great Caucasus margin (Azerbaijan), both characterized by the presence of active thrusts-folds and mud volcanism. Mud volcanoes that characterize the Northern Apennines foothills are represented by mud pools, gryphons and mud cones, and can be up to 3 - 4 m tall, with few tens of meters of diameter. The mud volcanoes of the Eastern Great Caucasus in Azerbaijan are considerably larger, some of which can be up to 400 m high, with diameter up to 5 km, and with size and morphological characteristics similar to those of magmatic volcanoes.

Figure 1.1. Location of the two study areas.

In this Ph.D research project, the principal adopted technique has been the Satellite Radar Interferometry (InSAR), that can detect small (millimetric-scale) superficial terrain displacement and provides detailed information about the nature of the deformation. InSAR has been extensively used in various sectors of Earth Sciences, and it is a particularly powerful tool for studying volcanic activity and active tectonic deformation. In this study, the Satellite Radar Interferometry was applied through both the Differential Satellite based Synthetic Aperture Radar Interferometry (DInSAR) basic approach, and the Persistent Scatterers Interferometry (PSI). DInSAR and PSI techniques have allowed to measure ground deformation of the order of some centimeters in 35 days (revisit time of Envisat) and of the order of few mm/year, respectively. The used data are Envisat-ASAR images provided by the European Space Agency in the framework of an ESA Category 1 Project, specifically created for this Ph.D work (CAT1 13866, entitled: "Assessing the relationships between tectonics and mud volcanism by integrating DInSAR analysis and seismic data in active tectonic areas"). The use of the two different techniques (DInSAR or PSI) was dictated by several factors, in particular: (i) the number of Envisat images available in the ESA archive, (ii) the physical characteristics of the two different areas of study (degree of urbanization, land use, aridity) that strongly affect the coherence of the radar data, and finally (iii) the achievement of the two specific objectives - i.e., (1) studying the ground deformation related to mud volcano activity, and (2) studying the ongoing tectonics along folds and fold-and-thrust belt margins (Tab. 1.1).

Constraints Using DInSAR Using PSI considerable amount of number of available small amount of available available images (a minimum Envisat images images of ~15 images is generally required for PSI) physical characteristics of the sparse vegetation, aridity urbanized area study areas ground deformations related ground deformations related to objective to study to large mud volcano edifices thrust-folds activity

Table 1.1. Factors that constrain the use of DInSAR and PSI techniques.

In particular: (1) The first objective was achieved with the DInSAR technique, using the same methodological approach implemented for studying the eruptive phases of magmatic volcanoes. This has been one of the first attempts to apply interferometry to the study of mud volcano dynamics. Detecting ground deformation at the mud volcano edifices has been possible only in the cases where the size of the mud volcanoes was large enough to be analyzed by the Envisat spatial resolution. Moreover, a good level of coherence should characterize the interferograms in the area where the mud volcanoes occur. For this reason, this area should be arid or sparsely vegetated. The gryphons and mud cones that occurs in the Northern Apennines margin are too small to be analyzed by this sensor. Furthermore, the whole Pede-Apennine foothills is a densely vegetated area, affected by decorrelation problems in the Envisat interferograms. Very interesting results have been instead achieved in the case of the Great Caucasus study area, where the large volcanic edifices of Azerbaijan have been suitable for our goals. The lack of vegetation that characterizes the eastern part of Azerbaijan allowed to obtain good level of coherence of the interferograms. Five mud volcano edifices undergoing deformation have been analyzed in detail, indicating that satellite radar interferometry constitutes a suitable tool for studying mud volcano activity. Finally, the results contribute to a wider understanding of the processes driving ground deformation at mud volcanoes, and show that mud and magmatic volcanoes display some similarities in the behavior of ground deformation during both quiescence and pre-eruptive stages.

(2) The second objective was achieved through the use of the PSI technique (Ferretti et al., 2000, 2001), which is effective in detecting surface deformation of wide regions affected by low tectonic movements. This methodology is based on the detection of a considerable amount of radar targets which maintain the same electromagnetic signature in all radar images (i.e. permanent scatterers, PS). They may correspond to rock exposures, but normally, they are represented by buildings, metal objects, infrastructure and therefore, for the proper use of this technique, a densely urbanized zone is preferable. Finally, the PSI technique requires the availability of several radar images. These conditions occur along the Pede-Apennine foothills and particularly in the Po Plain, where it has been possible to identify a dense network of permanent reflectors. The study area of the Eastern Great Caucasus, on the contrary, is devoid of settlements and urbanized areas (with the exception of the Baku urban area) and therefore unsuitable for using the PSI technique. This second part of the Ph.D. project has attempted to correlate the superficial deformation signals measured by radar satellite-based sensor with the known geological features. The PSI analysis of ongoing tectonic activity in the Pede- Apennine margin and Po Plain, led to the detection of ground uplift related to the growth of thrust anticlines, with important seismo-tectonic implications, leading to new insights into the present-day deformation pattern of the frontal area of the Northern Apennine.

1.2 Thesis Structure

This thesis is divided into 3 principal chapters and is organized as follows: - Chapter 2 is devoted to a description of the main technique used for this research project, namely satellite radar interferometry. The two distinct approaches, DInSAR and PSI, are briefly summarized, highlighting the advantages and draw-back of both. - Chapter 3 is devoted to the use of the DInSAR technique and to the study of the ground deformations connected to the activity of the mud volcanoes. - Chapter 4 deals with the use of the PSI technique and with the measurement of superficial ground movements linked to tectonic features. Both Chapter 3 and Chapter 4 contain an introduction and a summary of the geological setting of the related study areas (the Great Caucasus in Azerbaijan and the Northern Apennines margin, in northern Italy, respectively). In both chapters a description of the studied geological phenomenon (the mud volcanoes of Azerbaijan and the main active structures of the outermost

Northern Apennine front, respectively) is provided, then the results and the individual case studies are examined, and the interpretations and the new hypotheses derived from data analysis are argued. Finally both chapters end with the conclusions paragraphs, where the principal findings of the study are summarized. Chapter 5 reports the General Conclusions.

Chapter 2

Methodologies

2.1 Remote sensing: introduction

The idea and practice of looking down at the Earth's surface emerged in the 1840s when pictures were taken from cameras secured to tethered balloons for purposes of topographic mapping. From the first World War until the early 1960s, the aerial photograph remained the single standard tool for depicting the surface. Satellite remote sensing can be traced to the early days of the space age (both Russian and American programs) and began to imaging surfaces using several types of sensors from spacecraft. Remote sensing reaches a substantial maturity in the seventies, with operating systems for capturing images on the Earth with a certain periodicity: the Landsat, the first satellite implemented for non-military goals by NASA, was dedicated to monitoring of land and oceans in order to map the natural resources. The term "remote sensing", first used in the fifties, is now commonly used to describe the science of identifying, observing, and measuring an object without coming into direct contact with it. This process involves the detection and measurement of radiation of different wavelengths reflected or emitted from distant objects or materials, by which they may be identified and classified. Unless it has a temperature of absolute zero (-273°C) an object reflects, absorbs, and emits an electromagnetic radiation in a unique way, and at all times, depending primarily on its temperature. The areas of the electromagnetic (EM) spectrum (Fig. 2.1) that are absorbed by

16 atmospheric gases such as water vapour, carbon dioxide, and ozone are known as absorption bands. Most remote sensing instruments on aircraft or space-based platforms operate in the areas where the atmosphere is transparent to specific wavelengths bands. Some sensors measure the radiant energy of an object reflecting sunlight with detectors tuned to specific frequencies that pass through the atmosphere. Other sensors, especially those on meteorological satellites, directly measure absorption phenomena, such as those associated with carbon dioxide and other gases. The atmosphere is nearly opaque to EM radiation in part of the mid-IR and all of the far-IR regions. In the microwave region, by contrast, most of this radiation moves through unimpeded, reaching the surface. Remote sensing data provide unique identification characteristics leading to a quantitative assessment of the Earth's features. A primary use of remote sensing data is in classifying the myriad features in a scene (usually presented as an image) into meaningful categories or classes. The image then becomes a thematic map (the theme is selectable e.g., land use, geology, vegetation types, rainfall). When more than two wavelengths are used, the resulting images tend to show more separation among the objects. Optical satellite sensors can record reflected energy in the red, green, blue, or infrared bands of the spectrum, providing a basic remote sensing data resource for quantitative thematic information, such as the type of land cover and natural resources, including mineral ores, vegetated areas, and wetlands. The use of optical data has proved to be effective in geological mapping due to their synoptic coverage, high spatial resolution and multispectral capability (Drury, 1987; Sultan et al., 1987; Alberti et al., 1993; Gupta, 2003). Advantages also include the availability of these data at reasonable cost and the more effective targeting of areas of particular lithological significance for fieldwork. In fact lithological investigation of a poorly known area can draw substantial help from remotely sensed data, particularly in arid environments. The remote sensing systems can be basically classified as active and passive. The former are equipped with a transmitting system and receive the signal backscattered from the illuminated surface, the latter make use of the radiation naturally emitted or reflected by the Earth’s surface. Reflected sunlight is the most common external source of radiation sensed by passive instruments, like the optical sensors. The active sensors avoid some of the limitations of the passive ones: in particular, they are independent on sunlight. Synthetic Aperture Radar (SAR) is an active sensor that works in the microwave region of the EM spectrum, transmitting radar waves. Their frequency bands drastically reduce the impact of clouds, fog and rain, therefore they can be considered time and weather independent. The measurements of time of arrival of

17 the transmitted signal and of the return time of the backscattered signal from the ground surface, allow the detection of the distances between the target and the satellite. Therefore the comparison between the travel-time information allows to know if the target has moved between different acquisitions of the sensor. The launching of radar satellites with SAR sensors onboard, purposely built for interferometric applications, and able to detect small (millimetric-scale) terrain displacements, opened new opportunities for measuring deformations of the Earth’s surface. InSAR has exploited this basic principle since the late 1980s (Goldstein et al., 1988; Gabriel et al., 1989) to measure topography and surface displacement (Bürgmann et al., 2000), benefiting from the data provided by civilian or dual-use space-borne SAR sensors launched worldwide since the early 1990s. From 1992 (Landers Earthquake, Massonnet et al., 1993) this technique has been frequently applied to study the surface displacement field, and from that moment on, this technique has been applied to detect and monitor terrain motions such as landslides (Carnec et al., 1996; Kimura and Yamaguchi, 2000; Catani et al., 2005; Crosetto et al., 2005), volcanoes (Wicks et al., 1998; Amelung et al., 2000; Lundgren et al., 2001; Pagli et al., 2006; Chang et al., 2007; Biggs et al., 2011), earthquakes and release of seismic energy (Massonnet, and Feigl, 1998; Price and Sandwell, 1998; Wright et al., 2004; Lyons and Sandwell, 2003; Lanari et al. 2010) other types of deformations, such as subsidence and slow slope movements (Gabriel et al., 1989; Massonnet et al., 1997; Rott et al., 1999; Crosetto et al., 2011; Herrera et al., 2011), and building stability (Dong et al., 1997; Stramondo et al., 2006; Brenner and Roessing, 2008; Chini et al., 2009).

Figure 2.1. The electromagnetic spectrum.

2.2 Synthetic Aperture Radar

SAR uses RADAR (RAdio Detection And Ranging) technology, and thus uses a transmitter operating at either radio or microwave frequencies (Fig. 2.1) to emit electromagnetic radiation and a directional antenna or receiver to measure the time of arrival of reflected or backscattered pulses of radiation from distant objects. In other words, the principle of operation of a radar system is quite simple: a transmitting device illuminates the surrounding space with an electromagnetic wave that affects the objects (targets). The latter reflects the energy (diffusion, scattering) and a part of the scattered electromagnetic wave back toward the station, also equipped for reception. The sensor is able to detect the target and evaluate his distance localizing it in a precise way, by measuring the time delay between the instant of transmission and the instant of reception. Distance to the object can be determined since electromagnetic radiation propagates at the speed of light. The SAR images are usually acquired on board an airplane or a satellite. Let’s consider a spaceborne sensor that travels at the velocity of a satellite at constant altitude in a straight line, i.e. the azimuth direction, and emits the electromagnetic beam perpendicular to the azimuth direction, i.e. the range direction (see Fig. 2.2). The illuminated area on the ground is the antenna footprint. Figure 2.2 illustrates the Envisat-ASAR imaging geometry, in StripMap mode, which is the acquisition mode of the SAR images used in this thesis. The StripMap mode scans the ground track with constant incident angle θ. Radar images are developed along the range and azimuth directions, usually called SAR coordinates. The spatial range resolution depends on the pulse duration τ, which must be short enough to allow a good resolution, but large enough to ensure a sufficient backscattered energy. The spatial azimuth resolution is inversely proportional to the antenna dimension in the azimuth direction and to the operational wavelength: the shorter the wavelength and the larger the antenna dimension, the better the resolution. The spatial resolution, is an important characteristic of the sensor i.e. their capability to distinguish two points at a certain distance. However, the antenna dimension has restrictions in terms of size. Therefore, to overcome this limitation, SAR creates a synthetic antenna exploiting the movement of the antenna: starting from a short physical antenna this allows obtaining a much larger synthetic antenna.

Figure 2.2. StripMap Envisat-ASAR imaging geometry.

The latter is obtained by exploiting the orbital movement of the antenna along the orbital path and then processing the received signal. The Synthetic Aperture Radar (SAR) is an active sensor based on this approach, overcoming the main limitation of Real Aperture Radar (RAR) i.e. the poor spatial resolution they achieve with the working wavelengths. The SAR images used in this thesis are a two dimensional complex images, called single look complex (SLC) images, whose rows and columns are directly related to the azimuth and range positions of the scatterers. Each pixel corresponds to just one signal, which is the vectorial sum of the multiple signals of the scatterers contained within the ground resolution cell. While the amplitude (Fig. 2.3a) of the backscattered signal can be related to the surface scattering properties (mainly on the roughness of the terrain), its phase (Fig. 2.3b) contains travel-time information. More precisely, the phase difference between two acquisitions can be related to the changes in the distance of the radar from the illuminated object. In fact, the phase is the measure of the travel distance (going and coming back, i.e. 2R) covered by the signal. The relation between phase and distance is:

2휋 4휋 휙 = 2푅 + 휙 = 푅 + 휙 (1) 휆 0 λ 0

20 where ϕ0 is the phase shift generated during the interaction between the signal and the measured object, and λ is the wavelength of the sensor. It is important to note that the radar only measures fractions of the whole phase, measured modulo 2π. This aspect is called phase ambiguity.

Figure 2.3. (a) Mean amplitude image of 20 Single Look Complex (SLC) images over Parma city. White pixels correspond to high reflectivity areas, whereas dark pixels correspond to low reflectivity points, as the sea (bottom right corner). (b) Phase image of a SLC over the same area. The phase ranges from –π to π.

A distinctive feature of radar systems is that they have a side-looking geometry. The look angle varies depending on the satellite but generally ranges from 20˚-50˚ off the nadir (straight down; about 23° for Envisat, Fig. 2.2). This fact makes it possible to view the same area from two different geometries. In fact, the terrain elevation results in geometric distortions in the SAR images (Hanssen, 2001), and this affects the ground resolution cell. Figure 2.4 illustrates the three distortion effects that can be found in the SAR images: foreshortening, layover and shadow. Foreshortening appears in the slopes that are facing the radar sensor: the measured terrain is compressed, i.e. it appears shorter in the SAR image (see points A-B in Fig. 2.4). On the other hand, the opposite effect occurs for the slopes that are inclined away from the sensor: the terrain is expanded in the SAR image. Layover occurs in slopes that are higher than the incidence angle, i.e. the sensor firstly receives the signal reflected by the top and then the signal coming from the bottom of the slope (see points C-D in Fig. 2.4). Finally, shadow occurs when an object of the ground prevents the radar signal to reach other ground scatterers (see points D-E in Fig. 2.4). It occurs in steep slopes opposite to the sensor. No measurements are possible on those areas.

Figure 2.4. Foreshortening, layover and shadow effects. Scatterers between A-B suffer foreshortening effects, points C-D are reverted due to the layover, and scatterers between D-E are not illuminated by the radar due to the shadow effect.

2.2.1 Orbit geometry, LOS, vertical and horizontal motion

The satellites currently used for acquiring SAR images follow a fixed trajectory around the Earth (inclined by a few degrees off the N-S axis), and circumnavigate the globe at an altitude of approximately 800 km. As the satellite moves along its orbit, the Earth is also spinning on its axis, so that at each orbit of the satellite scans a different portion of the Earth’s surface. As already mentioned in the previous paragraph, is also important to consider that satellite based radar systems have a side-looking geometry. When the satellite travels in the descending part of its orbit, meaning that it is travelling from N to S, it views a target area looking westward, while during the ascending part of its orbit, that is, when it moves from S back to the N, it views the same target area looking eastward (Fig. 2.5).

Figure 2.5. Illustration of ascending and descending satellite orbits.

Radar satellites can only measure motion along their Line-of- Sight (LOS). This is defined as the line along which the sensor views the Earth surface target and, due to the side-viewing geometry, it is not vertical but inclined. In other words, what is actually measured by the sensor is the projection of a target’s motion onto the LOS. If the motion direction is close to the angle of the LOS then the measured and actual motions will be similar. However, the LOS motion can often differ noticeably from the real value of motion, especially in cases where the ground motion is not vertical (Fig. 2.6).

Figure 2.6. Motion measured by the sensor for different directions of terrain-motion. Red arrows represent the vector of terrain motion while blue arrows represent the LOS motion measured by the radar system.

With this configuration it is possible to view the terrain in both ascending and descending geometry (though not simultaneously with the same sensor). It is therefore possible to combine the measured motion information to obtain an accurate estimate of the actual vertical motion and of the east-west component of the motion (Fig. 2.7).

Figure 2.7. Illustration of the method for obtaining actual motion by combing ascending and descending orbit information. a) In the case of vertical ground displacements the motion components on the ascending and descending directions are both negative (moving away from the sensor) b) while in the case of a horizontal (E-W) motion one vector is positive (moving toward the sensor) while the other is negative.

2.2.2 Satellite based SAR sensors

The first SAR sensor was the Seasat, managed by NASA's Jet Propulsion Laboratory. In 1989 some researchers developed the SAR data processing technique known as SAR Interferometry. The main source of useful radar imagery was provided by the European Space Agency (ESA), with the ERS-1, ERS-2 and Envisat-ASAR sensors (C-band: frequency = 5.3 GHz), which all together span nearly 20 years, from the end of 1991 to July 2011. The repeat cycle of both ERS and ASAR (Advanced Synthetic Aperture Radar) missions was 35 days. Envisat, the sensor used in this PhD project, was the largest Earth Observation spacecraft ever built, ensuring the continuity of data after ERS-2, despite a small (31 MHz) central frequency shift (frequency = 5.331 GHz and wavelength = 5,63 cm). It was launched in 2002, in a 98.5° sun-synchronous circular orbit at 800 km altitude, and carried ten sophisticated optical and radar instruments to provide continuous observation and monitoring of the Earth’s land, atmosphere, oceans and ice caps. The full scene Envisat images covers an area of 100 x 100 Km. Presently a C-band Canadian sensor (Radarsat-2), two German X-band instruments (TerraSAR- X and TanDEM-X), the Japanese L-band PALSAR-2 sensor on board ALOS-2, the ESA C-band Sentinel-1, and the four Cosmo-SkyMed X-band sensors managed by the Italian Space Agency (ASI) are operational. The list and frequency interval of the satellite bands are shown in Table 2.1.

Bands Frequency interval (GHz) Satellite

L 1.0 - 2.0 ALOS-2

S 2.0 – 4.0 -

ERS-1, ERS-2, Envisat, C 4.0 – 8.0 Radarsat-2, Sentinel-1 Cosmo-SkyMed, TerraSAR-X, X 8.0 – 12.0 TanDEM-X Table 2.1. Satellite frequency bands

Apart from building archives, the revisit time (temporal resolution) is relevant, because a short revisit time permits the monitoring of faster deformation. TerraSAR-X is operated in an 11-day repeat orbit, and since Cosmo-SkyMed uses multiple satellites even shorter intervals are possible. Nevertheless, good temporal coverage is only achieved for a very small fraction of the 24 total area, namely if acquisitions were programmed. As regards the data availability, in 2014 a new InSAR era has started, with the successful deployment of the European Sentinel-1A SAR satellite and a second Sentinel-1B mission (launched in 2015), able to provide SAR images with constant repeat pass (with 12 day intervals), for many years in the future. The improved SAR data availability, the larger coverage, the constant temporal sampling provided by the Sentinel-1 sensors will generate a strong increase in the observation of crustal deformation worldwide, providing new high quality data.

2.3 The DInSAR technique

SAR interferometry (InSAR) is a radar based technique that exploits the information contained in the phase of at least two SAR images acquired over the same area, at different times, in order to infer terrain deformation measurements (Hanssen, 2001). The first InSAR basic observable datum is called an interferogram, and represents the per-pixel phase difference between two SAR acquisitions. The two images (the first one is called master and the second one slave) of the same area can be acquired by the sensor in a revisit time. Figure 2.8 illustrates the repeat-pass acquisition geometry. The distance between the two (master and slave) satellite orbits is the interferometric baseline, and its projection perpendicular to the slant range (i.e. the line of sight distance between sensor and target) is the perpendicular baseline (B┴). This parameter plays an important role in the topography measurement. The temporal baseline is the time (in days) between two acquisitions. The differential phase delay is the so-called interferometric phase, a geometric quantity directly related to the elevation of the target and its movement (Rosen et al., 2000). The interferometric phase is also affected by the different phase delay propagation caused by the different atmospheric effects during both acquisitions, and by a term due to noise. In particular the interferometric phase can be decomposed in these components: (i) the phase caused by the deformation of the target ϕMov, (ii) the effect of the different positions of the sensor at each acquisition time, named topographic component ϕTopo, (iii) the effect of atmospheric propagation ϕAtm, and (vi) finally a noise term ϕNoise, that is the sum of the instrumental noise and the contribution of the physical changes of each measured object which can affect radar response. Hence, measuring deformations, i.e. to obtain ϕMov, requires solving the following equation:

ϕInt = ϕM − ϕS = ϕMov + ϕTopo + ϕAtm + ϕNoise + 2kπ (2)

where ϕInt is the interferometric phase, ϕM and ϕS are the phases of the master and slave images, and k is an integer number.

The goal of any DInSAR (Differential SAR Interferometry) technique is to derive ϕMovfrom the

ϕInt equation. This implies separating ϕMov from the other phase components of Eq. 2.

The topographic phase ϕTopo can be expressed as a function of the perpendicular baseline B┴:

4π B┴ ϕTopo = H (3) λ RM sin θ

where H is the altitude of the scatterer, RM is the range of the master, and θ is the off-nadir angle

(Fig. 2.8). Note that the sensitivity to topography increases with the perpendicular baseline B┴. This property can be exploited to generate a DEM (Digital Elevation Model), using high spatial baselines.

Figure 2.8. Geometry of acquisition of an interferometric SAR system (from Devanthéry, 2014), where M corresponds to the master satellite and S to the slave satellite. B is the interferometric baseline.

ϕTopo has to be removed from the interferometric phase to derive deformation measurements.

This is achieved by using an available DEM simulating ϕTopo, in order to obtain the so-called differential interferometric phase:

ϕDInt = ϕInt − ϕTopo_simulated = ϕMov + ϕRTE + ϕAtm + ϕNoise + 2kπ (4)

26 where ϕRTE is the residual topographic error (RTE) phase component, which is the difference between the true height of the point and the height given by the used DEM, ϕMov is the phase related to the difference of path length travel caused by surface deformations: 4 휋 ϕ = 푑 (5) Mov λ where d is the displacement in the LOS, λ is the wavelength, and 4π is related to the two-way path radar-target-radar.

2.3.1 Atmospheric effects and noise

The measured differences in travel times (or distances) can be influenced by un-modeled effects, e.g. an effect due to variable propagation velocity through the atmosphere, which is mainly linked to water vapor content in the troposphere (Zebker et al., 1997). The artefacts caused by variations in the atmosphere have a spatial dimension that spans from hundreds of meters to 100- 200 kilometers (Hanssen, 1998), depending on the phenomena originating the turbulence, the characteristics of a given area and the season. The vertical stratification effects of the atmosphere only affect areas with pronounced topography (Hanssen, 2001). Several sources as the differences in imaging geometry, the soil erosion, the vegetation growth, the agricultural practices or, more generally, the human activities may cause that the SAR phases at the two acquisitions are statistically decorrelated (Zebker and Villasenor, 1992), and thus unrelated to the changes in the distance from the sensor. The phase noise influences the quality of the interferometric phase. The main sources of noise are geometric, and temporal decorrelation (Hassen, 2001). Geometric decorrelation is caused by the different incidence angle of acquisition of the master and slave images, which originates a shift in the range spectra of the two SAR images. This spectral shift increases with the perpendicular baseline. Temporal decorrelation is caused by changes in the characteristics of the scatterer over time. For instance, vegetated areas show very fast decorrelation over time due to the high variability of the physical properties of the scatterers. Seasonal and anthropogenic changes also produce temporal decorrelation in interferograms with large temporal baselines. By contrast, arid areas and man- made structures usually display high correlation over time.

2.3.2 Phase unwrapping

Phase differences are only observed modulus 2π, and therefore the DInSAR phase ambiguity, given that the phase is modulus 2π (Fig. 2.9), has to be solved. The recovery of the integer multiples of 2π, and thus the determination of the phase gradient between any two interferogram pixels, represents a 2D phase unwrapping problem, that is one of the major problems of SAR interferometry. It consists in reconstructing the unwrapped phase by means of adding the appropriate number of cycles:

ϕInt_unwrapped = ϕInt_wrapped + 2kπ (6) where k is an integer number, called phase ambiguity. This means calculate the exact number k of multiples of 2π.

Figure 2.9. Interferometric wrapped phase (above, known between ± π) and unwrapped phase (below).

Some a priori assumptions need to be made in order to solve the phase unwrapping problem. The most common assumption in phase unwrapping algorithms is the phase continuity between neighbouring pixels:

|ΔϕInt_unwrapped(i,j) – ΔϕInt_unwrapped(k,l)| < π (7)

28 where (i,j) and (k,l) represent two neighbouring pixels. Note that, in C-band, π corresponds to a maximum relative displacement of about 1.4 cm. Higher relative displacement values lead to unwrapping errors, which are multiple of 2π and result in errors of the displacement measurements. These errors are particularly important in areas with low coherence, where neighboring pixels are sparse. Many phase unwrapping algorithms have been proposed in the literature (Chen and Zebker, 2000). The so-called minimum cost flow (Costantini, 1998 and 1999), which is formulated in terms of a global minimization problem with integer variables, is been used in this work. Other phase unwrapping methods consists in a spatial and a temporal phase unwrapping. The interferometric procedure proposed in Monserrat, (2012) and Devanthéry, (2014) includes a 2+1D phase unwrapping method.

2.4 The PSI technique

The DInSAR technique, that measures the change in distance between the ground and the sensor using pairs of SAR images, can be successfully used when the magnitude of the deformation is higher than the other contributions to the interferometric phase, which usually occurs in single deformation events reaching at least several centimeters of deformation. On the other hand, it does not properly work with slow deformation events or when the presence of atmospheric artefacts is significant. Moreover, significant limitations are due to changes in reflectivity and to atmospheric disturbances. In fact a draw-back of the classical DInSAR techniques, is the possible lack of coherence of the radar signal on vegetated areas and in different climatic conditions. An efficient solution to these problems has been provided by the multi- interferometric approaches grouped in the name of Persistent Scatterers Interferometry (PSI). These approaches were exploited from the early 2000s to go beyond the conventional InSAR methods by correcting for atmospheric, orbital and DEM errors in order to identify a sub-set of image-pixels where high-precision measurements can be carried out. At this specific points on the ground, the so-called persistent scatterers or PSs, it is possible to obtain very accurate displacement and velocity measurements with millimetric precision (Ferretti et al., 2001; Werner et al., 2003; Crosetto et al., 2008). A considerable amount of satellite SAR data can provide annual velocity values and time series of ground deformation on a dense grids of point-wise targets (Ferretti et al., 2001). In particular, the PSI technique allows the identification of radar targets which maintain the same electromagnetic signature in all images, and which preserve the

29 phase information in time (i.e. PSs, Fig. 2.10). Normally PSs correspond to buildings, infrastructure, antennas, conducts, and exposed rocks. They act as persistent scatterers because of their geometry, surface-roughness and electrical conductivity. Snow coverage, constructions works, street re-pavement, etc. can cause the complete or partial loss of PSs. The PS points can form a high spatial density network. Very high PS densities are concentrated in urban environments, while low density occurs in inhabited, agricultural landscape, vegetated area and in areas affected by foreshortening, layover and shadowing effects on the related SAR images. The exact location of PSs, therefore, cannot be predicted in advance of processing, but over urban areas their densities are usually measured in the hundreds per square kilometer. This technique strongly minimizes the loss of coherence between time delayed acquisitions, but unfortunately limits the use of the method to urbanized areas. After identification of the PSs in a data-set, the inherent phase data between all common scatterers across all scenes can be compared. If the phase data for a particular scatterer are the same across all scenes, then the scatterer is deemed not to have changed its relative location. Conversely, any difference in phase correspond to motion. This does not of course account for a number of variables, one of which is atmospheric refraction which, in effect, can change the signal path length and hence the phase data, giving false information. The statistical computation of an “atmospheric phase screen” to correct for this effect is consequently a vital part of the process. The PSI technique indeed takes conventional InSAR a step further by correcting for atmospheric, orbital and Digital Elevation Model (DEM) errors to derive relatively precise displacement and velocity measurements at specific points on the terrain. It is worth considering that PSI is a rather complex process whose results depend on multiple factors, i.e. the number of available SAR images, the density of PSs, the quality of PSs, the characteristics of the deformation at hand (spatial extent, type of temporal deformation, deformation rates, etc.), the distance from the reference point, the quality of the reference point, etc. The best PSI performances are obtained under ideal conditions, i.e. large data stacks, high PS density, zero to slow deformation and linear deformation. Different PSI techniques which exploit the redundancy offered by tens or hundreds of image pairs have been proposed in the last fifteen years. The first PSI approach, called Permanent Scatterer technique, was proposed by Ferretti et al. (1999 and 2001, patent of the PSInSAR™ algorithm). In addition to this, the Small Baseline Subset (SBAS) technique is one of the most important and well documented PSI approaches (Berardino et al., 2002; Lanari et al., 2004; Pepe et al., 2005, 2007, 2011). In the original SBAS approach (Berardino et al., 2002), interferometric pairs are chosen to minimize temporal and geometric decorrelation, allowing deformation time

30 series to be retrieved for distributed scatterers, i.e., neighboring radar resolution cells, which are not dominated by a single scatterer, and share the same backscattering properties. Another PSI method suitable for geophysical applications is described in Hooper et al. (2004). A simplified approach based on stepwise linear functions for deformation and least squares adjustment is proposed in Crosetto et al. (2005). Recently a new algorithm called SqueeSAR, which can jointly process PSs and distributed scatterers (DS), was proposed by Ferretti et al. (2011).

Figure 2.10. Schematic representation of the PSI theoretical basic principles. Modified form http://treuropa.com/.

2.4.1 Key issues and processing

PS velocity values are relative to a reference point, chosen within the PS dataset, which is considered stable. One of the key outputs of a PSI analysis is the average velocity of each persistent scatterer (PS) over the entire time period encompassed by the SAR data used, expressed as an average annual velocity (in mm/yr). Perhaps the most important distinction between conventional InSAR and PSI is the ability of the latter to create time series of motion for individual PSs temporally referred to the first acquisition date of the image stack, and spatially referred to a reference point or area within the radar coverage. They provide the

31 deformation history over the observed period, which is fundamental for many applications, and comprise a detailed graphical representation of the displacement, usually in mm, of the PS along the LOS as a function of the time between the first and last acquisition. Time series represent a powerful tool in the analysis of terrain-motions: they provide valuable information concerning individual PS behavior that cannot be discerned by simply observing the mean annual velocity. It is possible indeed to identify non-linear motions, seasonal trends, accelerations, and the times when changes in motion behavior took place. Another outcome of a PSI analysis is the so-called residual topographic error (RTE), which is the difference between the true height of the scattering phase center of a given PS and the height of the DEM in this point. The RTE is a key parameter in order to achieve an accurate PS geocoding. PSI requires a large number of SAR scenes acquired over the same area. Typically, a minimum of 15-20 images is needed to perform a C-band PSI analysis. Anyhow, the larger the number of available scenes the better the quality of the PSI deformation velocity and time series estimation. Besides, this number is critical for the estimability of those PSI models that comprise multiple unknown parameters, e.g. see Monserrat et al. (2011) and van Leijen (2014). PSI suffers severe limitations in the capability to measure fast deformation phenomena due to the ambiguous nature of the wrapped interferometric phases. When the differential deformation phase, between two subsequent acquisitions is bigger than , corresponding to a maximum differential deformation of /4 over the revisit interval, the actual deformation cannot be retrieved unambiguously. The maximum differential deformation rate measurable is 14.7 cm/yr for Envisat. The actual capability to estimate the deformation rates depends on the noise level of the data and the technique used to resolve phase ambiguities (van Leijen, 2014). Note that the above condition involves pairs of PSs and its effect on single PS depends on the spatial pattern of the specific deformation phenomenon. Many PSI approaches make use of a linear deformation model in their deformation estimation procedures. This assumption, required to unwrap the interferometric phases, can have a negative impact on the deformation estimates of all phenomena characterized by non-linear deformation behavior, for which the assumption is not valid. In particular, the PSI products may lack PSs, due to the model misfit, in areas where the deformation shows significantly non-linear motion.

The temporal coherence (γe) is a quality indicator of the goodness-of-fit of the model (height correction, topographic errors and deformation) to the observations (interferogram phase values; see the following paragraph (2) Deformation velocity and RTE estimation). The index is based

32 on the maximization of a function that takes into consideration the number of scenes used in the analysis and the coherence of the stable reflectors identified as possible PS. The measurement accuracies depend on various factors, such as the type of terrain and the properties of the SAR image data stack (Casu et al., 2006; Lanari et al., 2007; Hooper et al., 2012). The precision of the measurement can range from 0.1 to 1 mm/year depending on their dispersion of amplitude (Ferretti et al., 2007), and decreases with distance from the reference point. The dispersion of amplitude (DA; see the following paragraph (1) Candidate PS selection) index is a statistical parameter that evaluates the temporal variability of the scatterer response for each point and can be used to evaluate the phase noise (Ferretti et al., 2001). In this work a classical PS approach implemented at the Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) has been applied. The main steps of the procedure are shown in Fig. 2.11, which provides a detailed description of each step. However, for sake of completeness, here below the key aspects of the applied approach for the particular case study of this work are briefly discussed. The starting point is a set of N single look complex SAR images from which a set of M interferograms is calculated (M>>N).

Figure 2.11. Flow chart representing the general scheme of the PSI data analysis procedures used at CTTC.

(1) Candidate PS selection PSI techniques require a pre-selection of the points to be used to derive the deformation measurements. The selected points, named candidate PSs, are points with good interferometric phase quality, i.e. with a low level of ϕnoise. The applied selection methods involves the use of the Dispersion of Amplitude (DA). The DA criterion, proposed in Ferretti et al. (2001), is the most widely employed, e.g. see Werner et al. (2003), Kampes and Hanssen (2004), Kampes (2006), Hooper et al. (2004 and 2007), Crosetto et al. (2008) and van Leijen (2014). It consists in a measure of the scatterer stability by means of the analysis of the amplitude along the stack of N images. The DA is estimated as the standard deviation of the amplitude (σA) divided by the mean amplitude (mA):

σA DA = ⁄mA (8)

Bamler and Hartl (1998) show that there is a linear relation between the DA and the standard deviation of the phase noise for small values of DA: low DA values indicate low phase dispersion during the period of study. The DA is a pixel-wise selection method that works at full resolution, enabling the selection of isolated points. The method is successfully used when large stacks of images, at least 25 according to Ferretti et al. (1999), are available; the criterion can be biased if the number of images is smaller. Typically, a DA threshold of 0.2 is selected in urban areas, where the density of PSs is high. In non-urban areas, such as rural and vegetated areas, the DA threshold can be usually less strict to reach the required density of candidate PSs. Consequently, a higher noise component is accepted.

(2) Deformation velocity and RTE estimation All the PSI methods need to unwrap the interferometric phases to estimate the deformation time series over the selected PSs. Several methods have been proposed with different models, interferogram configurations and methodologies to filter the APS (APS stands for Atmospheric Phase Screen). The PSI method proposed in Ferretti et al. (2000 and 2001), and applied in this work, uses a linear deformation model. The phase is modelled as the sum of a RTE ( i.e. the height of each measured PS with respect the DTM used to calculate the network of differential interferograms) term and a deformation term:

4π 4π B﬩ ϕDint = vΔt + RTE (9) λ λ RMsinθ where v is the linear velocity and Δt is the temporal baseline of a given interferogram. This model avoid performing directly the phase unwrapping and it is the most widely used, e.g. see Adam et al. (2003) and Werner et al. (2003). All the available SLC images can be used if this model is implemented together with the DA, because the selected point scatterers are not affected by geometric and temporal decorrelation, allowing to use large baselines. The differential nature of the differential interferometric phase is exploited by considering pairs of candidate PSs (see Fig. 2.12). Given an edge “e” of a pair of candidate PSs, the differential k wrapped phase of an interferogram k, Δϕ obs_e, is computed, and therefore, for each edge, the differential deformation velocity and RTE are estimated. In detail, the differential interferometric k phase is modelled assuming a term related to the linear movement, Δϕ linear_mov_e, a term related

k k to the RTE, Δϕ RTE_e, and a residual error term, Δε e, related to the APS, noise and non-linear deformation effects:

4π 4π Bk Δϕk = Δϕk + Δϕk + Δεk = Δtk Δv + ┴ ΔRTE + Δεk (10) obs_e linear_mov_e RTE_e e λ e λ Rksinθk e e

where Δve is the differential linear velocity of the edge e, ΔRTEe is the differential residual k k topographic error, Δt and B┴ are the temporal and perpendicular baselines of the interferogram, Rk and θk are the slant range and incidence angle of the interferogram k, and λ is the radar wavelength. k The best fit between Δϕ obs_e and the model proposed in Equation 10 is occur when the residuals k Δε e, is minimized. An associated statistical parameter named temporal coherence (γe) evaluates the agreement between the observed phases (interferograms) and the used model (mostly the linear deformation model). The gamma function takes values between 0 and 1, reaching its k maximum when the residual errors (Δε e) are zero. The higher is the maximum γe, the better is the fit to the model of the wrapped phases. Note that the presence of non-linear motion during the period of study decreases the gamma value because a linear deformation is assumed.

Figure 2.12. Network of connected Persistent Scatterers (PSs). For each edge the differential interferometric phase is calculated and the differential deformation velocity and RTE are estimated. Figure adapted from Biescas et al. (2007).

(3) Removal of the deformation velocity and RTE phase components In this step, the contribution of the linear deformation and RTE of each selected point is removed from the original interferograms. This step helps to smooth the local phase gradients of the interferograms easing the phase unwrapping process.

(4) 2+1D phase unwrapping In this step the interferometric phase ambiguity by adding to each phase the integer multiples of 2 is solved. This procedure is described in detail in Monserrat (2012) and Devanthéry (2014). The phase unwrapping is done in two different steps. First the unwrapping is performed spatially, interferogram by interferogram, over the M interferograms. Errors can occur when the phase changes due to RTE are large or, similarly, where there are strong deformation gradients. Furthermore, a common source of unwrapping errors is given by phase variations due to the atmospheric component, particularly when the PSs density is low, e.g. in the case of isolated areas. Finally, the noise can also produce unwrapping inconsistencies over specific points. The phase unwrapping errors have to be detected in order to avoid errors in the estimated phases, and, consequently, in the deformation estimates. For this reason, for each unwrapped point a consistency check is performed, exploiting the temporal component of the interferograms network. Note that this step is performed only over the network of selected points. The result of this step is, for each selected point, the temporal evolution of the phase along the measured period.

(5) Atmospheric and orbital errors estimation and removal The atmospheric phase component removal is a fundamental step in any wide-area PSI processing. Its goal is the correct separation of the APS (Atmospheric Phase Screen) from the deformation signal. The N APS images are estimated and the APS component is removed from the M interferograms to achieve this goal. The atmospheric disturbance is characterized by a smooth spatial phase gradients and a random behavior in time. In this work, a low pass filter (LPF) is applied to remove the low frequencies associated with large spatial correlation phenomena. In this particular case a Butterworth filter is executed over the unwrapped phase images to estimate the low pass spatial component, that is the candidate APS images. Then for each point, a temporal filter is applied to separate the APS signal from the temporally correlated components, usually related to terrain deformation. The signal correlated in time is estimated and removed from the candidate APS with this second filter. Discriminating

37 between the part of the signal that is APS from that which is deformation is often not straightforward. In this work, in case of doubt in discerning between the two components, the candidate APS are considered APS. The PSI deformation velocity maps can sometimes be affected by tilts, which can be caused by uncompensated orbital errors or uncompensated low frequency atmospheric effects (e.g. see Bateson et al., 2010). The orbital errors are inaccuracies in the estimation of the satellite position, at the time of the acquisition. Note that these tilts have a corresponding effect in the deformation time series. A tilt in a given deformation velocity map could be due to uncompensated processing errors or to a real geophysical signal. One of the following two situations could happen: (i) a tilt in a given velocity map can be interpreted as geophysical signal, while in fact it is simply a result of residual processing errors, which can have a negative impact in the results of the geophysical analysis at hand; (ii) a tilt-free velocity map can be interpreted by a geophysicist as no signal, e.g. quiescence of a given phenomenon, while in fact the site may have undergone significant geophysical low-frequency deformations that have been removed (together with the other residual effects) during the PSI processing. Any PSI application focused on low spatial- frequency deformation signals should consider this limitation of the PSI technique. However, this limitation can be overcome by using multiple overlapping tracks or by fusing the PSI with external data from GNSS, levelling, gravimetry, etc., e.g. see Chen et al. (2010), Hung et al. (2011), Caro Cuenca et al. (2011), Hanssen et al. (2012) and Adam et al. (2013).

(6) Final estimation of the deformation velocity In order to derive the deformation maps and time series, the procedure starts with a set of M APS-free interferograms and a set of selected PSs. In this final step, a 2+1D phase unwrapping is performed over a dense set of PSs (a lower threshold of γe is exploitable in this second iteration) on the M APS and RTE-free interferograms. The RTE needs to be estimated and removed from the M APS-free interferograms to avoid errors in the 2D phase unwrapping. A spatial 2D phase unwrapping is performed separately on each interferogram using the Minimum Cost Flow method (Costantini, 1998 and 1999). The 2D phase unwrapping is an error-prone step, especially for isolated pixels, high phase gradient between pixels, and when the atmosphere or the noise components are high. A 1D phase unwrapping, whose goal is the detection and correction of the errors generated in the 2D phase unwrapping stage, is performed. For this purpose, it uses an iterative least squares procedure (Baarda, 1968; Förstner, 1986; Björck, 1996), which fully exploits the integer nature of the unwrapping errors. The 1D phase unwrapping involves a pixel wise procedure over the temporal dimension (Devanthéry et al., 2014). 38

(7) Output: velocity maps and time series The main outputs of the procedure are two: the deformation velocity map that provides the deformation rates in mm/yr for each measured PS, and the deformation time series describing the PS behavior along the measured period.

2.4.2 Comparison with existing survey techniques

The monitoring of ground deformation can be accomplished by several types of technique, including geodetic methods, such as leveling, GPS-based systems, theodolite and total station surveying, photogrammetric methods. These techniques have different characteristics in terms of precision, reliability, cost, automation and real-time capability. Even if some techniques offer highly automated solutions, e.g. the continuous GPS-based deformation monitoring of structures, they can usually only be used to measure a limited number of points or cover relatively small areas. PSI represents a particular type of monitoring tool, which can be competitive in different ground deformation applications. In fact, PSI offers the typical advantages of all remote sensing techniques, such as data acquisition over inaccessible areas, wide area coverage, periodic and low-cost data acquisition, and, at the same time, it can potentially monitor deformations at millimeter level of the whole Earth's surface. The potential reduction in the amount of ground- based observation by using PSI allows targeted and simplified logistics operations, reduces network maintenance costs, personnel time and cost and allows to retrieve information in dangerous and inaccessible zones. PSI measures terrain-motion along the LOS of the satellite and is more sensitive to vertical motions than to motions in the horizontal direction. The sensitivity is further reduced for horizontal motions in the north to south (azimuth) direction due to the satellite orbit geometry. By contrast, GPS has less sensitivity to vertical motions than horizontal, and can measure horizontal displacements which are hardly detectable by radar. Concerning the measurement density and the costs of the technique, PSI analysis shows benefits respect to the GPS one, because it allows surveying at a much higher density and is significantly cheaper. Drawbacks of PSI as compared with GPS measurements are the fact that the current SAR systems provide a limited sampling frequency (temporal resolution), related to orbital periodicity, while GPS networks have continuous observation capabilities. The integration of PSI and GPS measurements creates a dataset which overcomes the disadvantages of a sole GPS or PSI survey. The validation of PSI has received a lot of attention in the last fifteen years. Its results have provided valuable inputs for the research on PSI and 39 development activities, and have proved to be key for the acceptability of this new technique at scientific, technical and commercial level. Most of the PSI validation activities were based on the comparison of deformation velocities and time series with independent estimations of the same quantities acquired by levelling or GPS measurements. Validation of deformation velocity and time series, mainly based on levelling and GPS data, are described in Colesanti et al. (2003), Musson et al. (2004), Salvi et al. (2004), Hooper et al. (2004), Colesanti et al. (2005), Strozzi et al. (2005), Teatini et al. (2005), Casu et al. (2006), Zerbini et al. (2007), and Dehghani et al. (2013).

Chapter 3

Study of the superficial deformation of mud volcanoes of Azerbaijan through the DInSAR technique

3.1 Introduction

Mud volcanism is a process that leads to the extrusion to the topographic surface of material originated from deeply buried sediments, such as saline waters, gases (mostly methane), mud, and fragments or blocks of country rock. This mechanism is typically linked to in-depth hydrocarbon traps (Higgins and Saunders, 1974), and it builds up a variety of features, the most typical of them being the conical extrusive edifices that may vary in size from centimeter-scale to a few hundred meters in height and some kilometers across. Mud volcanoes usually occur in fold-and-thrust belts and submerged accretionary prisms, and develop at convergent plate margins where the active tectonic shortening affects the sediments with increasing stresses and temperatures leading to the maturation of the organic matter (Higgins and Saunders, 1974; Brown, 1990; Kopf, 2002). During the largest part of their life, the mud volcanoes show a background activity consisting in the quiet to vigorous expulsion of fluids and mud breccias. However, such a quiescent activity may be occasionally interrupted by paroxysmal events consisting of the violent release of gasses and of large mud flows. The Greater Caucasus in Azerbaijan hosts the greatest number of mud volcanoes on Earth (Jakubov et al., 1971; Guliyev and Feyzullayev , 1997). Some mud volcanoes may be tall up to 400 m, and long as much as 4-5 km, with dimension and morphological characteristics similar to those of magmatic volcanoes. These volcanoes 41 experienced several eruptions, for which there is a rather complete record since 1810 (Mellors et al., 2007). In particular, the second edition catalogue provides the chronology and a brief description of 387 eruptions, which took place between 1810 and 2007 at 93 mud volcanoes onshore and offshore Azerbaijan (Aliyev et al., 2009). Satellite based Synthetic Aperture Radar Interferometry (InSAR) have been commonly used to monitor and investigate the ground deformation connected to the eruptive phases of magmatic volcanoes. In particular, the satellite interferometry has been proved to be a powerful tool for monitoring ground deformations produced by different processes at active magmatic volcanoes, particularly: (i) uplift before eruption and co-eruptive subsidence (Amelung et al., 2000; Jay et al., 2014), (ii) pulses of inflation and deflation on volcanic systems (Biggs et al., 2009, 2011; Brunori et al., 2013; Chang et al., 2007; Pagli et al., 2006; Pritchard and Simons, 2004), (iii) subsidence associated with magma phase change (Caricchi et al., 2014), (iv) linked hydrothermal and magmatic systems (Wicks et al., 1998), (v) crustal rifting episodes (Pagli et al., 2012; Sigmundsson et al., 1997; Wright et al., 2012), (vi) downslope movements on the volcano flanks (Ebmeier et al., 2010), and (vi) lava thicknesses and extrusion rate (Ebmeier et al., 2012). InSAR techniques have also been employed to explore the ground deformation associated with the LUSI mud volcano in Indonesia, which is still erupting since 2006 (Abidin et al., 2009; Aoki and Sidiq, 2014; Fukushima et al., 2009; Rudolph et al., 2013). Using similar techniques, we have carried out research on ground deformation related to the activity of the mud volcanoes of Azerbaijan. For this purpose, the deformation of mud volcanic systems has been analyzed using the differential interferometry (DInSAR) technique applied to Envisat data.

3.2 Mud volcanism

Mud volcanism phenomenon is linked to many relevant aspects (see Mazzini et al., 2009), spanning from their role in predicting hydrocarbons reservoirs (e.g., Devlin et al., 1999), to the environmental implications related to their input of methane into the atmosphere (e.g., Kopf, 2002), and their hints into how fluids travel up through the shallow crust (Davies et al., 2008; Mazzini et al., 2009). ‘Mud volcano’ is a generic term that indicates the various morphologic features associated with the extrusion of subsurface material. As soon as it reaches the topographic surface, the extruded mud gives rise to scenic features, the conical edifices being the most typical (Fig. 3.1c and d).

Other manifestations are represented by mud pools (or ‘salses’) characterized by bubbling gas centers.

Figure 3.1. Mud volcanoes activity and morphological features. (a) Flaming eruption of the Lockbatan mud volcano (event of 25th October 2001). (b) Typical gas bubble topping a gryphon during the normal background activity; hammer (encircled) for scale. (c, d) Details of gryphon/vent alignments on the summit crater of a larger extrusive edifice (photos taken during the field survey in June 2013).

Conventional nomenclature subdivides the small sub-conical extrusive edifices in gryphons (≤3 m high; Fig. 3.1c, d) and mud cones (> 3 m high), while the term mud volcano should be used for the edifices reaching some tens of meters in height or few kilometers across. The largest mud volcanoes are the gigantic submarine serpentinite mud volcanoes (‘conical seamounts’) of the Marianas forearc, which may reach 25 km in diameter and exceed 2 km in height. Mud volcanoes are widespread worldwide, both on-shore and off-shore (Fig. 3.2). Mud volcanic provinces occurs in the Gulf of Mexico and Caribbean basin, in the Mediterranean Sea, Crimea and Caucasus, northern Iran, Pakistan, and Burma to Malaysia, Indonesia, and New Zealand (Kholodov, 2001). Smaller-scale manifestations of mud volcanism are known in California, Italy, Taiwan and Japan. Moreover mud volcanic provinces are localized in the Caspian Sea and in the Black Sea (Kholodov, 2001).

Various mechanisms have been proposed to control the development of mud volcanism. In particular, this process has often been considered to manifest subsurface intrusive processes such as mud or shale diapirism (Fig. 3.3; Brown, 1990; Morley and Guerin, 1996; Kopf, 2002). Mud diapirs are subsurface fluid-rich overpressured muddy masses that are driven upward in response to their buoyancy resulting from the bulk density contrast with respect to the denser surrounding overburden (Brown, 1990). The overpressure is produced by the organogenic activity and the subsequent gas production at depth (e.g., Higgins and Saunders, 1974). The rapid expansion and degassing (at ca. 1–2 km depth) of the methane dissolved in the mud further increases both the overpressure and buoyancy and the stress on the surrounding sediments causing their fluidification (Brown, 1990). Other models suggest that mud volcanism is sourced from the depletion of highly fluid-rich source layers at depth, rising up through intricate systems of conduits and pipes and networks of anastomosing fault-controlled planar pathways exploiting deeper fluid-rich source layers (Fowler et al., 2000; Cooper, 2001; Dimitrov, 2002; Morley, 2003; Planke et al., 2003; Davies et al., 2007; Mazzini et al., 2009; Roberts et al., 2010; Albarello et al., 2012). Fluid reservoirs may occur at various depths, and likely consist of zones with high density networks of mud-filled fractures.

Figure 3.2. Distribution of mud volcanoes on Earth (modified from Kholodov, 2002). Subaerial volcanoes: (I) large mud volcanic provinces—(1) Apsheron Peninsula, southwestern Gobustan, Nizhnyaya Kura depression (Azerbaijan), (2) Kerch and Taman peninsulas; (II) other mud volcanic provinces—(3) North Italy, (4) Sicily, (5) Albanian coast, (6) Romania, (7) West Turkmenian depression, (8) Gogran coast (Iran), (9) Makran coast of Iran and Pakistan, (10) Baluchistan and Punjab areas (Pakistan), (11) Assam and eastern Punjab areas (India), (12) Junggar region (China), (13) islands of western Burma, (14) middle course of the Irrawaddy River (Burma), (15) Andaman Islands, (16)

Kalimantan Island (Malaysia), (17) Kalimantan Island (Malaysia), (18) Timor Island (Indonesia), (19) New Guinea (Indonesia), (20) Sakhalin Island (Russia), (21) Hokkaido Island (Japan), (22) North Island (New Zealand), (23) Trinidad Island (Trinidad and Tobago), (24) Venezuela, (25) northern Columbia; (III) isolated mud volcanoes; subaqueous volcanoes: (IV) large mud volcanic provinces—(26) South Caspian Sea; (V) other mud volcanic provinces—(27) eastern Black Sea, (28) western Black Sea, (29, 30) Mediterranean Sea; (VI) isolated mud volcanoes and seeps.

Figure 3.3. Mud volcano schematic model. Elongation, elongated vent distributions, mud chamber elongation and summit caldera elongation patterns are showed. Mud dykes preferentially trend perpendicular to σ1 taking advantage of the crestal faulting along the anticline. Summit calderas and mud chambers also become elongate perpendicular to σ1. After Roberts et al. (2011b).

Mud volcanoes are thus closely associated with petroleum systems, and the development of overpressures in reservoir rocks is a necessary condition for triggering mud volcanism (Dimitrov, 2002). For this reason mud volcanoes often localize in correspondence of thrust- anticlines where sealing layers in the fold core may efficiently trap the rising hydrocarbon fluids and readily built-up overpressures (Yassir and Bell, 1996; Bonini, 2012). Mud volcanoes are frequently located also at rollover anticlines associated with listric normal faults (e.g., Graue, 2000).The development of overpressures in sediments is thus a necessary prelude to mud volcanism (Dimitrov, 2002). Tectonic stress provides an important source of overpressure (though difficult to quantify) as highlighted by the widespread mud volcano occurrence in many active compressional belts worldwide (e.g., Kopf, 2002). The state of stress is also inferred to control distribution, geometry, and shape of mud volcano features (Bonini, 2012). The activity of the mud volcanoes is often characterized by paroxysmal phases and occasionally by flaming eruptions (Fig. 3.1a) caused by the self-ignition of the methane, which may interrupt the quiescent phases. The background activity ranges from quiet to vigorous explosion and flow of mud and fluids, with the typical intermittent gas bubbles popping up through the saline muddy 46 water (Fig. 3.1b). Local and regional earthquakes with large enough magnitude have also been shown to increase eruption rate or initiate new eruptions (Mellors et al., 2007; Manga et al., 2009; Bonini, 2009; Manga and Bonini, 2012). However, violent eruptions can also occur in the absence of seismicity (Bonini, 2009). Mud volcanoes and magmatic volcanoes display very similar morphologic features, and for this reason many terms used for mud volcanism are often borrowed from the terminology of magmatic features. For instance ‘crater’ (Fig. 3.4a) is used to indicate the sub-circular collapsed areas topping the extrusive edifices, ‘caldera’ indicates the depressions formed as a consequence of the removal of subsurface material (Fig. 3.4b, c), and ‘vent’ refers to openings through which fluids are released.

Figure 3.4. Map view of Azerbaijan mud volcanoes (southern Greater Caucasus) from Google Earth imagery. (a) Large mud volcano edifices, with elongate elliptical shape: Touragai (N40°09′48″, E49°17′50″) and Kjanizadag (N40°08′26″, E49°22′50″). (b, c) sub-circular collapsed areas topping the extrusive edifices, and the ‘caldera’ depressions of Gotur (N40°8'55", E49°10'43.02") and Gyulbakt (N40°14'24", E49°30'26") mud volcanoes respectively.

Other similarities concern the internal structure and eruptive history. In particular, the relatively small magmatic ‘monogenic’ cones that form over a short time span may be analogous to the gryphons/cones that grow during a single period of activity, while the ‘polygenic’ volcanoes may be equivalent to the large mud volcanoes that result from discrete extrusive periods and that may remain active over a few million years (e.g., Yusifov and Rabinowitz, 2004; Evans et al., 2006). 3D seismic reflection data has allowed the imaging of the vertical architecture of an idealized mud volcano system composed of a cone-shaped extrusive edifice overlying a caldera depression connected by a feeder system to a deeper source layer (Davies and Stewart, 2005; Stewart and Davies, 2006).In addition, feeder complexes and the collapse of some Azerbaijan mud volcanoes may imitate the ‘sector collapse’ structures typical of igneous volcanoes (Roberts et al.,2010, 2011a), and the velocity of mud ascent (60–300 km yr−1) is comparable to that of silicate magmas (Kopf and Behrmann, 2000). Finally it has been hypothesized that both magmatic and mud eruptions share the same threshold seismic energy to be triggered off by distant earthquakes (Wang and Manga, 2010). Even though obvious differences do exist – mud volcanism being almost entirely driven by fluid pressure, and magmatic volcanism driven by pressure in addition to temperature, magma chemistry, volatiles, etc. – the processes governing both volcano systems may be viewed as dynamically similar. Both magmatic and mud volcanoes are in fact driven by gas escape (buoyancy supplied by exsolved gases, and the presence of an in-depth overpressured source; Manga et al., 2009; Wang and Manga, 2010). In this perspective, the stress indicator concepts formerly concerned for the igneous features could also be applied to the same features developed in mud volcanic settings. Alignments of active mud volcano vents have been indeed analyzed for estimating the ongoing stress-field in the shallow crust of Azerbaijan (Bonini and Mazzarini, 2010; Roberts et al., 2011b). Steep mud-filled intrusions have been referred to as hydraulic fractures (e.g., Morley et al., 1998; Morley, 2003) and interpreted as potential pathways for the transport of the mud–fluid mix to the surface (Davies et al., 2007). The crestal region of folds may also experience stretching above a neutral surface produced by ‘tangential– longitudinal strain folding’ (Ramsay, 1967). Normal faults may thus control the shape of mud calderas, which would elongate parallel to a local, shallow, fold axis-parallel SH (Bonini et al., 2012).

3.3 Geological setting of the Greater Caucasus

The Greater Caucasus thrust-and-fold belt lies northeast of the Black Sea and northwest of the South Caspian and accommodates most of the northeast-southwest convergence between Arabia and Eurasia (Jackson, 1992; Forte et al., 2014). The present structure of the Greater Caucasus is characterized by the outcropping of the Palaeozoic basement west of 44E, which is exposed in dominantly southward directed thrust slices (Allen et al., 2003; Fig. 3.5a). Mesozoic strata are present in tight or isoclinal folds across the range all around the exposed Palaeozoic core (Allen et al., 2003; Fig. 3.5a). These folds are commonly associated with thrusts, and the vergence of these structures is predominantly southward (Fig. 3.5b). However, there are north-directed thrusts on the northern margin, especially in the northeast (Dagestan). Pliocene-Quaternary strata are folded into elongated, linear, south-vergent anticlines on the southern side of the range, between the Kura and Rioni basins (Allen et al., 2004; Fig. 3.5b). Consequently, fault plane solutions show a predominance of thrusts dipping toward the range interior (Triep et al., 1995; Jackson et al., 2002). Late Cenozoic magmatism is Pliocene and Quaternary in age (Gazis et al., 1995). A number of researchers (e.g., Adamia et al., 1977; Gamkrelidze, 1986; Zonenshain and Le Pichon, 1986; Golonka, 2007) interpret the Greater Caucasus to have originated by closure of a Mesozoic to Cenozoic back-arc basin, which originally opened during north-dipping subduction of Neotethys along the southern margin of the Lesser Caucasus arc. The exact timing of initial stages of Cenozoic closure of the basin and the subsequent shortening is controversial (e.g., Adamia et al., 2011; Golonka, 2007). Evidence for Oligocene deformation in the western Greater Caucasus (e.g., Vincent et al., 2011) probably testifies the early subduction stages during the basin closure, with continent-continent collision occurring at ~5 Ma (Forte et al., 2014). This onset of collision is largely coincident with a major plate reorganization in the Arabia-Eurasia collision (e.g., Allen et al., 2004), which additionally marks the initiation of the North and East Anatolian Faults (e.g., Westaway, 1994), the rapid exhumation of the Alborz mountains (e.g., Axen et al., 2001)and the onset of the subduction of the South Caspian basin (e.g., Allen et al., 2002).The convergence rates between the Lesser and Greater Caucasus increase from ~2 mm/yr at 42°E to ~12 mm/yr at 51°E, as indicated by GPS data (Reilinger et al., 2006). Most of this shortening was inferred to occur along the Main Caucasus Thrust (Fig. 3.5a), a north-dipping thrust system near the southern edge of the Greater Caucasus (e.g., Allen et al., 2004; Reilinger et al., 2006).

Figure 3.5. (a) Synthesized geologic map of the Greater Caucasus and surrounding areas (modified from Forte et al., 2014). The thin white line indicates the gross position of the Main Caucasus Thrust (MCT). Red lines indicate locations of mapped faults. (b) Tectonic map of the Greater Caucasus (modified from Forte et al., 2014).

However, recent studies indicates that the deformation front at ~2–1.5 Ma propagated southward accompanied by the abandonment of the Main Caucasus Thrust (Forte et al., 2010, 2013). The propagation of thrusting toward the Kura basin is observed along the northern margin of the central and eastern Greater Caucasus (Sobornov, 1994; 1996). Similarly, in the southeastern flank of the Greater Caucasus (in Azerbaijan), the southernmost Kura fold-thrust belt accommodated most of this shortening east of 45°E (Forte et al., 2013; Fig. 3.5b) at an average rate of 8 mm/yr since ~1.5–2 Ma, (Forte et al., 2013). The Kura fold-thrust belt is an elongated range that is roughly parallel and represents the southeastern continuation of the Greater Caucasus chain front for ~400 km. This chain of ridges and valleys lies within Azerbaijan and eastern Georgia and separates the Alazani and Kura basins to the north and south, respectively

(Fig. 3.5b). The Kura Basin separates the Greater and Lesser Caucasus ranges to the north and south, respectively, and has a long axis that trends roughly parallel to the strike of the Greater Caucasus (Forte et al., 2010; Fig. 3.5b). The geologic history and role of the Kura Basin within the Arabia-Eurasia collision zone remain enigmatic, but the most the most likely scenario is that the Kura is a flexural-foreland basin loaded by the Greater Caucasus during Pliocene (Koçyiğit et al., 2001).Along its western margin (~45°E), NE-SW–trending seismic sections indicate a typical foreland geometry, in which Cenozoic siliciclastic sediments are ~5–8 km thick near the Greater Caucasus range front and then gradually thin southward toward the Lesser Caucasus (Ershov et al., 2003). The sedimentary fill of the Kura Basin has been described as molasse (Khain, 1966; Tevelev and Blyumkin, 1990).The overall Cenozoic sedimentary sequence within the Kura Basin is generally equivalent to that in the South Caspian basin, although the total thickness of the Cenozoic section varies significantly, from ~5–7 km in the Kura Basin (Agabekov et al., 1976; Mangino and Priestly, 1998; Forte et al., 2010) to >20 km in the South Caspian Basin (Berberian, 1983; Forte et al., 2010). In the Kura Basin, seismic sections and well log data both indicate that strata have a regional dip of 1°–2° N and that stratal thicknesses of Plio-Pleistocene sediments increase northward, toward the Greater Caucasus range front (Agabekov et al., 1976; Ershov et al., 2003; Forte et al., 2010).

3.4 Geological setting of the Azerbaijan mud volcanoes

The study area is centered on the southeastern part of the outermost sector of the Great Caucasus orogenic system, and in particular the Kura fold-and-thrust belt and the Kura Basin (Fig. 3.6a). The thrust-anticlines of the Kura fold-thrust belt are considered to be active (Forte et al., 2014), and host the mud volcanism phenomenon, that is linked to the hydrocarbon reservoirs. The Azerbaijani territory counts hundreds of mud volcanoes of various size and shape (Fig. 3.7). In Figure 3.8b, c, d some examples of mud volcano edifices of eastern Azerbaijan are shown, in particular gryphons, mud cones and mud volcano, respectively.

Figure 3.6. (a) Simplified structural sketch map of the Greater Caucasus - eastern Caspian Basin (modified after Jackson et al., 2002). KFTB: Kura fold-and-thrust belt. (b) Regional cross section through the Absheron Sill and the South Caspian Basin (vertical exaggeration 2x; simplified from Stewart and Davies, 2006).

The Azerbaijan mud volcanoes are typically associated with hydrocarbon traps in thrust anticlines (Jakubov et al., 1971; Guliyev and Feyzullayev, 1996; Reynolds et al., 1998; Fowler et al., 2000; Planke et al., 2003; Stewart and Davies, 2006; Bonini, 2012). In particular, mud volcanoes puncture the crest of the fold anticlines bounding the eastern Kura Basin (Baku and Apsheron Peninsula areas; Fig. 3.7; Jakubov et al., 1971; Guliyev and Feyzullayev, 1997), and those forming the submerged Apsheron Sill in the South Caspian Basin (cross section through the Absheron Sill and the South Caspian Basin in Fig. 3.6b; Fowler et al., 2000; Cooper, 2001; Stewart and Davies, 2006).

Figure 3.7. Main mud volcano fields around the Greater Caucasus front and Absheron area (adapted from Jakubov et al., 1971; Jackson et al., 2002; Bonini and Mazzarini, 2010; Bonini et al., 2013); these features are superposed onto a shaded digital terrain model (U.S. Geological Survey, available from: http://www.gisweb.ciat.cgiar.org/sig/90m_data_tropics.htm).

Most of these mud volcanoes are thought to have initiated their activity in the Pliocene around 3.5 Ma (Yusifov and Rabinowitz, 2004), thereby in close connection with the development of onshore and offshore folds that probably began to form during Early–Late Pliocene times (e.g., Devlin et al., 1999; Yusifov and Rabinowitz, 2004).These folds are generally considered to be active structures resulting from the folding of both the fluvial–deltaic sands of the latest Miocene-early Pliocene ‘Productive Series’ and the middle-late Miocene Diatom Suite (Fig. 3.8a). The folds are detached in the underlying Oligocene–Miocene overpressured shales of the ‘Maykop Series’ (Devlin et al., 1999; Jackson et al., 2002; Soto et al., 2011). The Maykop Series is a 200–1200 m-thick (up to 3000 m offshore) regionally continuous package of fine-grained clastic sediments with high content of organic material that provides the main regional source rock for hydrocarbons and for the mud volcano systems (Inan et al., 1997, Abrams and Narimanov, 1997; Hudson et al., 2008). The overlying marls and shales of the Diatom Suite (Fig. 3.8a), represent another important hydrocarbon source layer. The ‘Productive Series’ affords the majority of hydrocarbon reservoirs and consist of 5-7 km thick, fluvial-deltaic sand bodies, interbedded with mudstones deposited during short-lived lacustrine events (Allen et al., 2002;

Guliyev and Feyzullayev, 1996; Reynolds et al., 1998; Fowler et al., 2000; Stewart and Davies, 2006; Soto et al., 2011). Sedimentation of the Productive Series was followed by the deposition of the late Pliocene marine mudstones of the Akchagyl Series, and afterward by the Absheron Series and younger units (Fig. 3.8a), which were deposited in non-marine, commonly brackish conditions.

Figure 3.8. (a) Conceptual setting of Azerbaijan mud volcanism, which typically localizes at tight buckle folds detached on the overpressured Maykop shales (slightly modified from Bonini et al., 2013). Hydrocarbon gases, oil (mostly produced in the organic-rich Maykop Series), saline waters and mud are inferred to migrate into the fold core and collect in reservoirs within the Productive Series. This mud– fluid mix may eventually ascend toward the topographic surface along structurally-controlled pathways (i.e., outer-arc normal faults, faults and joints associated with the folding, etc.), giving rise to extrusive edifices constructed by mud breccias. Examples of onshore mud volcano edifices in eastern Azerbaijan: gryphons and mud cones at the (b) Kichik Maraza (Malomaraza; June 2013) and (c) Pirekyushkul mud volcano fields (June 2013). (d) Lateral view (looking west) of the ~400 m-tall Kjanizadag mud volcano (June 2013).

3.5 Previous studies on mud volcanoes through satellite interferometry

Very few studies have been carried out using the InSAR technique to analyze the mud volcanoes all around the world. Hommels et al. (2003) investigated the dynamics of the largest mud volcanoes (Touragai, Great and Lesser Kjanizadag) of Azerbaijan employing nine ERS-2 and ERS-1 scenes in combination with hyperspectral satellite ASTER imagery. They found preliminary indications of deformation in the analyzed dataset over a period of seven months. Around each volcano they found signals of uplift, which can be explained by increasing pressure at depth. Mellors et al. (2005) focused on the analysis of the Absheron Peninsula and the Lokbatan mud volcano. They employed eight ERS-1 and ERS-2 scenes from 1996 to 1999, and Radarsat images from 1999, but they did not observe any large-scale movement (>10 cm line-of- sight) during the analyzed period. InSAR techniques have also been successfully used to study the important and peculiar LUSI mud volcano in Indonesia. On May 29th 2006 the LUSI (Lumpur Sidoarjo) mud volcano started to erupt in the sub-district of Porong, East Java, Indonesia (Davies et al., 2007), in an zone adjacent to a hydrocarbon exploration well. An almost continuous eruption of a mud, water and gas mix has occurred since May 2006 that has caused significant damage to the livelihoods, the environment and the infrastructure. Subsidence occurs at steady rates of 0.1 and 4 cm/day in the central parts of the mud volcano (Abidin et al., 2009) and may take place by down-faulting of a caldera. Abidin et al. (2009) proposed that subsidence mainly occurs due to (1) the erupted mud loading, and (2) the collapse of the overburden due to the removal of mud from the subsurface. Moreover the authors identified the simultaneous presence of uplift during the eruption of the LUSI. The uplift deformation pattern aligns with the Watukosek fault system, which seems to be the cause of this localized vertical movement. Ground subsidence in the vicinity of the LUSI eruptive vent was well recorded by a Synthetic Aperture Radar (SAR) PALSAR onboard the Japanese ALOS satellite. Fukushima et al. (2009) obtained continuous maps of ground displacements around LUSI during the first year of the eruption, detecting a subsidence of the order of 70-80 cm of displacements away from the satellite (Fig. 3.9), and investigating about the possible causes of the subsidence.

Figure 3.9. Wrapped interferograms of the LUSI mud volcano (modified from Fukushima et al., 2009) dating (from left to right):19 May/4 October 2006; 4 October/19 November 2006; 19 November 2006/19 February 2007. The back-scatter signal on the mud-covered area is small and homogeneous, allowing to discriminate the extent of the flooded area.

Rudolph et al. (2013) used ALOS satellite imagery between October 2006 and April 2011 to attempt to estimate the longevity of the LUSI eruption. They found that the rate of ground deformation, and by implication, the pressure in the mud source region, has been decaying exponentially of 2.1 ± 0.4 mm/years. Aoki and Sidiq, (2014) derived the evolution of the displacement field associated with the LUSI eruption from SAR images taken from the ALOS satellite. They observed that the deformation associated with the mud eruption extends to the west as well as around the vent.

3.6 The used dataset and the DInSAR technique

The dynamic analysis of surface deformation at volcanoes is usually carried out by means of observations obtained during its seismic or eruptive activities. In the case of the mud volcanoes of Azerbaijan, detailed information exists only for specific cases. In order to widen this information, we have applied the satellite based DInSAR technique (see Chapter 2 for the detailed description).In particular, we performed a systematic search of all available satellite radar data in the online ESA catalogue and in particular in the Envisat mission (2003–2010). A dataset of 9 Envisat descending images (track 235-frame 2799) spanning from October 2003 to November 2005 have been processed in this study. The Envisat scene covers an area of 100 km2, in which the great majority of the Azerbaijan mud volcanoes are contained (Fig. 3.10). All the available interferometric pairs were processed at the Centre Tecnològic de Telecomunicacions de Catalunya (CTTC). The Shuttle Radar Topography Mission (SRTM) 56

Digital Terrain Model (DTM), with a spatial resolution of 90 m, was used to estimate the differential interferograms. Finally, a subset of 8 interferograms that showed a good phase unwrapping and covered the entire period of observations was selected for the analysis. Table 1 shows the main characteristics of the selected interferograms. It is worth noting that, with the exception of interferogram n°19, all the interferograms have temporal baselines shorter than 105 days, assuring working with interferograms with high coherence, which ease the phase unwrapping process. At the same time, the analyzed data set was heterogeneous enough to provide a robust discrimination of both atmosphere and topographic errors.

Temporal Date of the Perpendicular Interferogram n° Date of the slave baseline master baseline (m) (days) 11 02/10/2003 15/01/2004 -213 105 19 15/01/2004 25/11/2004 62 315 27 25/11/2004 30/12/2004 -212 35 1 30/12/2004 10/03/2005 -30 70 40 10/03/2005 14/04/2005 118 35 48 14/04/2005 28/07/2005 -160 105 51 28/07/2005 01/09/2005 -319 35 8 01/09/2005 10/11/2005 106 70 Table 1. Interferograms selected according to their coherence, spanning from October 2003 to November 2005.

Different approaches to solve quantitatively equation (2) (see Chapter 2) are described in the literature. However, these approaches were discarded for this work because all of them are based on the processing of large data stacks (a minimum of approximately 15 images is required). In this work, a simpler approach based on the simultaneous analysis of small sets of interferograms is employed. The first step of the applied procedure is to fully screen each interferogram in order to detect phase spatial variations. Once a phase spatial variation is detected, the significance of the contribution of each one of the non-deformation components is evaluated by using a pairwise logic criterion (Massonnet and Feigl, 1998) as briefly described here:

- φtopo: this component is linearly related to the perpendicular baseline of the interferogram. Hence, given a specific area, if the phase variation has opposite gradient with opposite

perpendicular baselines, then the observed pattern is mainly due to φtopo: otherwise, the φtopo

can be neglected. At this point, it is worth emphasizing that this analysis is adequate for two

reasons: first, the contribution of the φtopo is expected to be negligible since the studied areas are mostly flat, and secondly, the perpendicular baselines of the interferogram network are mostly small (< 216 m).

- φAtmo: Let’s assume that we observe a phase variation in an interferogramΔφ21= φ2 – φ1 and that the phase variation is mainly due to the atmospheric contribution of φ1. Then, the same

phase variation should appear in an interferogram Δφ1k= φ1 – φk but with opposite sign.

Therefore, analyzing different combinations of φ1and φ2makes possible the discrimination of

a significant φAtmo contribution.

- φNoise: this contribution is evaluated for each interferogram by means of the coherence. The coherence threshold has been chosen iteratively for each interferogram and each mud volcano separately. The final coherence threshold is based on a trade-off between the level of noise and the spatial point density, in order to obtain a reliable phase unwrapping. It is worth noting that the reliability of the unwrapped phase is qualitatively evaluated comparing the unwrapped interferogram with its corresponding wrapped version. In particular, for the Ayaz- Akhtarma mud volcano the selected coherence threshold is around 0.1 for all the used interferograms (see caption of Fig. 3.12). Regarding the other mud volcanoes, the selected coherence threshold is around 0.3 and 0.2 for the Khara-Zira and Akhtarma-Pashaly mud volcanoes, respectively, and 0.25 for both the Byandovan and the Akhtarma-Karadag mud volcanoes (see captions of Fig. 3.15, Fig. 3.17, Fig. 3.18 and Fig. 3.20). Phase unwrapping is applied to the interferograms that show negligible non deformation components. In this step, the absolute cycles of the phase are estimated, which leads to a correct interpretation of the measured deformation. The success of this step strongly depends on both the magnitude of the deformation and the density of coherent points. For this reason, only those interferograms with a trade-off between coherence and spatial density have been used. The analysis of interferograms led us to select a maximum temporal baseline of 105 days. Longer temporal baselines show poor correlation in most of the cases. The last step consisted of interpolating the unwrapped interferograms to obtain a continuous deformation field. The used interpolation method was a distance weighted bilinear interpolation. It is worth noting that the interpolated points are sparse and, in the worst case (interferogram 11), correspond to 15% of the total number of points. Hence, there are not expected interpolation artifacts.

To conclude, it is important to underline that the final deformation maps are represented in the Line-of-Sight (LOS) direction, i.e. the measured deformation at one point is the projection of the actual deformation along the satellite’s LOS.

3.7 Results

DInSAR observations have allowed us to detect significant deformation at four volcanic edifices in the Absheron Peninsula, in the Baku Archipelago and in the Gobustan region, namely the Ayaz– Akhtarma, Khara-Zira Island, Akhtarma–Pashaly, Byandovan (or Bandovan) and Akhtarma-Karadag mud volcanoes (Fig. 3.7 and Fig. 3.10a). After a general screening of the used Envisat frame (Fig. 3.10b), we focused our study on these five mud volcanoes because the displacement signals are the most evident and the deformation at these mud volcanoes clearly stands out. As no dedicated monitoring networks exist, the mud volcanic activity in the region remains poorly documented, and the associated hazard cannot be forecasted. A geological–structural field survey was carried out in selected areas of Azerbaijan during 2010 and during June 2013 (the latter field trip was carried out in the frame of this PhD project)to collect morphological and structural data on the volcanoes according to the preliminary results obtained from the DInSAR analysis. The results for each mud volcano are described below.

Figure 3.10.Location of the five studied mud volcanoes. (a) White boxes indicate the location of wrapped interferograms at Ayaz–Akhtarma (1 September 2005/10 November 2005), Khara-Zira Island (14 April 2005/28 July 2005), Akhtarma–Pashaly (25 November 2004/30 December 2004), Byandovan (25 November 2004/30 December 2004) and Akhtarma-Karadag (1 September 2005/10 November 2005). Each fringe represents 2.8 cm ofmotion in the satellite Line-of-Sight (LOS). (b) Red box: Envisat descending footprint (track 235-frame 2799).

3.7.1 Ayaz-Akhtarma mud volcano

The Ayaz-Akhtarma mud volcano is a large edifice with elliptical base (major axis about 2700 m) characterized by a wide flat top surface that might be a filled caldera (Fig. 3.11a, b). Current fluid expulsion occurs at several gryphons and cones clustering in a circular area (about 430 m in diameter) at the center of the crater (Fig. 3.11a, b). The major eruptions in the last 20 years

60 occurred in 2001, 2005, 2006 and 2007, even though the precise date of these eruptive events is unknown (Aliyev et al., 2009).

Figure3.11. Ayaz–Akhtarma mud volcano. (a) Panoramic view looking south. (b) Google Earth image (March 2004). (c) Fracture observed during the field survey carried out in June 2013. (d) Contour plot (Fisher distribution; equal area, lower hemisphere) of poles to fracture and fault segments. (e–f) Fault segments showing dominant vertical displacement (maximum vertical throw ~1 m). (g) En-echelon fractures pointing to some dextral strike–slip component of displacement. Images in (b) and (c) are extracted from Google Earth®; http://earth.google.it/download-earth.html.

Significant surface ground displacement within the mud volcano edifice is registered in the interferograms from October 2003 to November 2005 (Fig. 3.12), in connection with the strong eruptive activity of 2005 and 2006. Specifically, the wrapped interferograms (left column of Fig. 3.12) show a very complex pattern of concentric fringes in the eastern part of the mud volcano. The unwrapped interferograms (central column of Fig. 3.12) indicate that the deformation in the mud volcano shows a similar trend in all the analyzed interferograms. Although at different rates, the western part of mud volcano has moved away from the satellite and the eastern sector

61 towards the satellite. This implies that, assuming a purely vertical deformation, the eastern part of the volcanic edifices has lifted and the western sector has subsided. This deformation pattern is shown in Fig. 3.13, where the cumulative LOS displacements for the two orthogonal cross- sections (displayed at the bottom-right corner of Fig. 3.12) are represented. The figures of the right column of Fig. 3.12 represent the sum of the contributions of the interferograms. In particular, the measured ground uplift (LOS displacement) reached up to 20 cm in the period October 2003 - November 2005. In the hypothesis of a continuous deformation, the final cumulative ground uplift is probably underestimated because it has not been possible to analyze the interferograms with low coherence values, specifically those of January-November 2004 and April-July 2005 (white panels in Figure 3.12). The uplift has indeed increased in rate (up to 6 cm in 70 days) between July and November 2005 and it is located in a specific semicircular zone showing a larger diameter in comparison to the previous interferograms (Fig. 3.12). This observation suggests that the area of highest uplift probably corresponds to the center of inflation. The western area of the mud volcano is affected by a smaller subsidence which decreases in the last interferograms.

Fig. 3.12. Wrapped (panels on the left column), unwrapped (panels on the central column) and cumulative (figures on the right column) interferograms of the Ayaz–Akhtarma mud volcano. Unwrapped and relative cumulative interferograms with low coherence are not reported (white panels). The mean coherence threshold for all the used interferograms is 0.1± 0.05. The detected Line-of-Sight (LOS) ground displacement consists of a maximum motion toward the satellite (i.e. uplift, blue colors and negative sign in the chromatic scale, as the sensor-target distance decreases) in the eastern part of the mud volcano, and a maximum displacement away from the satellite (i.e. subsidence, red color and positive sign in the chromatic scales, as the sensor-target distance increases) in the western part of the volcano. The bottom right panel reports the traces of the cross-sections of Fig. 3.13.

Figure 3.13. Observed cumulative Line-of-Sight (LOS) displacements along cross-sections of the Ayaz– Akhtarma mud volcano. The traces A–A′ and B–B′ are indicated in Fig. 3.12 (bottom right panel).

Volumetric deformation rates of the order of those measured in this case might be associated with some surface brittle faulting/fracturing. A geological-structural field survey was carried out in June 2013 in order to investigate this possibility. The field survey allowed the identification of a main fault/fracture 600 m long and in a direction trending about N42°E (Fig. 3.11c, d). The north-eastern part of this brittle element is characterized by a normal component of vertical throw varying between 25 cm and 1 m (Fig. 3.11e, f). The remaining south-western tract consists of en-echelon fractures, which suggest a right strike-slip component of displacement (Fig. 3.11g). GPS reference points were collected and used to locate the fault/fracture on the geocoded radar digital data and on Google Earth images. This fault/fracture surveyed in 2013 can also be observed in Google Earth images dating back to 2004 (Fig. 3.11c), thereby suggesting that this brittle system has been active at least since 2004.

3.7.2 Khara-Zira Island

The Khara-Zira Island lies south-southwest of the Baku Archipelago and is one of the islands generated by mud volcanoes in the Caspian Sea (Mellors et al., 2007; Aliyev et al., 2009; Fig. 3.7 and Fig. 3.10). The Khara-Zira Island is elliptical in shape (with a maximum length of ~2.4 64 km) and is topped by a rather flat surface (Fig. 3.14a). According to the catalogue of mud volcano eruptions, a major paroxysmal phase occurred on 20th November 2006 (Aliyev et al., 2009). The catalogue describes this event as an explosion with inflammation of gas with a 50 to 200 m high fire column. A large amount of mud breccias was outburst and the altitude of the island increased significantly. The eruption zone can be easily identified on satellite images slightly after the eruption date (Fig. 3.14a). Unfortunately, the time span covered by the interferograms showing good coherence (October 2003 to November 2005) can only be used to observe the initial stages of pre-eruptive deformation. The wrapped interferograms show concentric, elliptical fringes located on the most active part of the mud volcano. All the interferograms show a relative ground uplift affecting most the mud volcano surface (Fig. 3.15, panels on the left and central columns). The cumulative ground uplift exceeds 10 cm in the two years analyzed (Fig. 3.16). At the north-western part of the island, the displacement displays opposite sign in two interferogram pairs, namely n. 19-27, and n. 51-08 (Fig. 3.15). The two interferogram pairs share a common master or slave date, which suggests that the change in the sign of the displacement might be the result of peculiar atmospheric conditions on two dates (25 November 2004 and 01 September 2005 for the first and second pair, respectively). Such rapid signal changes are likely caused by local atmospheric disturbances, and thus, may not represent actual ground displacement.

Figure 3.14. Morphological characteristics of the analyzed mud volcanoes. (a) Optical satellite image of the Khara-Zira Island (March 2007). (b) Optical satellite image of the Akhtarma–Pashaly mud volcano (March 2004). (c) Lateral view of the top surface of the Akhtarma–Pashaly mud volcano with gryphons and mud cones (May 2010). (d) Plan- and (e) oblique view of the Byandovan mud volcano (September 2009). Images are extracted from Google Earth®; http://earth.google.it/download-earth.html.

Figure 3.15. Wrapped (panels on the left column), unwrapped (panels on the central column) and cumulative (panels on the right column) interferograms of the Khara-Ziramud volcano. Themean coherence threshold for all the used interferograms is 0.3 ± 0.05. The Line-of-Sight (LOS) ground displacement infers a maximum motion toward the satellite (i.e. uplift) in thesouth-eastern part of the mud volcano (for explanation see Fig. 3.12 caption). The bottom right panel reports the traces of the cross- sections of Fig. 3.16.

Figure 3.16. Observed cumulative Line-of-Sight (LOS) displacements along cross-sections of the Khara- Zira mud volcano. The traces A–A′ and B–B′ are indicated in Fig. 8 (bottom right panel).

3.7.3 Akhtarma-Pashaly mud volcano

The Akhtarma-Pashaly is one of the largest mud volcanoes in Azerbaijan and is located over the crest of one of the thrust anticlines bordering the active south-eastern margin of the Great Caucasus. It has a roughly elliptical shape and is topped by a wide flat area with a major axis about 2.2 km long (Fig. 3.14b). The Akhtarma-Pashaly mud volcano had an intense period of activity in the eighties, with two eruptions in 1982 and one in 1986. Since then, we have no information of any relevant volcanic activity until 01 April 2013, when the volcano expelled a massive amount of mud and breccias that covered the entire perimeter of the edifice. During the field survey carried out in June 2013, a few months after the eruption, the surface of the volcano appeared as a large expanse of mud breccias, exceeding one meter in thickness in some areas. The mud breccias covered part of the pre-existing morphological features, including the mud cones and gryphons that erected from the top of the mud volcano (Fig. 3.14c). The Akhtarma-Pashaly mud volcano appears to undergo a period of quiescence in the time span covered by the analyzed interferograms that was interrupted by an isolated intense deformation event between November and December 2004. This deformation event was anticipated by a

68 phase of slight uplift during the previous 10 months. Unfortunately, the interferograms covering the period before the deformation event show low coherence. The strong deformation event (interferogram n. 27, Fig. 3.17) is revealed by fringes with complex patterns. A wide area in the central part of the volcano was affected by a strong subsidence (6 cm of LOS displacement in about one month), while a portion at its southern border experienced uplift (4.5 cm of LOS displacement in about one month). The deformation reduces drastically during the next 70 days until it vanishes (interferogram n. 01; Fig. 3.17). On the basis of the available eruption catalogue (Aliyev et al., 2009), the observed deformation would only represent a transitory pulse during the quiescence phase that lasted about 27 years, and it does not coincide with an eruption (which is indeed not found on the catalog list).

Figure 3.17. Wrapped (panels on the left column) and unwrapped (panels on the central column) interferograms of the Akhtarma–Pashalymud volcano. Themean coherence threshold for all the used interferograms is 0.2± 0.05. Displacement away from the satellite (i.e. subsidence) is detected in the central–western part of themud volcano. An area that moves toward the satellite (i.e. uplift) is perceptible in the southern part of themud volcano (see Fig. 3.12 caption for explanation). Observed displacements along cross-sections (panels on the right column). The traces are indicated on the unwrapped interferograms.

3.7.4 Byandovan mud volcano

The Byandovan mud volcano is an extrusive edifice with a diameter of approximately 1.5 km and 55 m high relative to the adjacent Caspian Sea level (altitude of -29 m). The so-called “Byandovan mountain” (Fig. 3.14c, d) is the most famous of the three mud volcanoes present in the Shirvan National Park, a natural reserve established in July 2003. This mud volcano lies on the NW-SE-trending Kelameddin-Byandovan anticline (Zeinalov, 2000). The catalogue of recorded mud volcano eruptions (Aliyev et al., 2009) reports two important eruptions that occurred in 1932 and 1989. Subsequently, the volcano went through a phase of quiescence. Satellite radar imagery spanning from October 2003 to November 2005 show that the volcano edifice has deformed, although no significant eruptive events are reported in the catalogue of mud volcano eruptions. The analyzed interferograms show a complex sequence of feeble episodes of ground displacement. In particular, the interferograms reveal a series of short-term pulses that alternate ground deformations with opposite sign (i.e., uplift and subsidence), localized at the main extrusive edifice. It is not possible to attribute the change of sign of the deformation to atmospheric conditions on a particular date, as the anomaly is found in many interferograms that do not share the master or slave images. In particular, a main ground displacement event was identified in an interferogram between November and December 2004 (interferogram n. 27, Fig. 3.18). The deformation consisted of relative uplift of almost 3 cm in 35 days at the mud volcano edifice, and subsidence of almost 2 cm in 35 days along a localized narrow, linear area located approximately along the coastline (Fig. 3.18).

Figure 3.18. Wrapped (panels on the left column) and unwrapped (panels on the central column) interferograms of the Byandovan mud volcano (see Fig. 3.12caption for explanation). The mean coherence threshold for all the used interferograms is 0.25± 0.05. The mud volcano edifice has experienced a complex sequence of uplift and feeble subsidence. Observed displacements along cross- sections (panels on the right column). The traces of the cross-sections are indicated on the unwrapped interferograms. 71

3.7.5 Akhtarma-Karadag mud volcano field

The Akhtarma–Karadag mud volcano field is located on an east–west trending anticlinal structure 87 km SW of Baku, near the Caspian coast. It consists of an elongate mud volcano field and is composed of morphological features of various size. A quite large mud volcano edifice, called Pilpilya, lies westward on the same anticline (Fig. 3.19a, b). This fold measures about 2.15 km in length, 0.8 km in width, and 96 m in elevation. The volcanic edifices affect the crest of an anticline that locally strikes about ENE and folds Early Pliocene (Productive Series) to Early Pleistocene (Absheron) sediments (Alizadeh, 2008). The summit area of the anticline is characterized by kilometre-scale normal faults that deform the fold crest, and have been previously referred to as an ‘incipient sector collapse’ structure (Roberts et al., 2011a, b).The faults are distributed into two oppositely-dipping fault sets defining a graben that closely follows the trend of the anticline axis (Fig. 3.19a, b). This graben is probably the product of outer-arc extension and collapse of the fold crest. The normal faults are characterized by relatively fresh and steep scarps that suggest recent formation (Fig. 3.19d, e)and show an apparent vertical separation of the ground almost ~1.5-2 m. Such throw decreases eastward to a centimeter-scale separation in the fracture zone (Fig. 3.19). The Akhtarma-Karadag mud volcano activity is mainly related to gryphons, mud cones (about 10 m high) and decameter scale vent populations that align with the fold-parallel faults (Bonini et al., 2012, Fig. 3.19b).The cones and vents are generally located at the normal faults, defining decameter-scale alignments trending sub-parallel to the anticlinal axis. Satellite radar imagery reports some evidence of deformation of the volcano edifice, although no significant eruptive events are reported in the catalogue of mud volcano eruptions for the analyzed time span. The interferograms from 30/12/2004 to 10/03/2005 and from 01/09/2005 to 10/11/2005 show two short-term pulses of ground displacement in that time interval. The deformation affects mainly the anticlinal crest and coincides with the down-faulted area of the graben. In particular the interferograms reveal subsidence (of the order of 1.8 cm in 70 days) in the south-western area of the central graben, while a portion at its east-northern side experienced uplift. Intermingle, the recent activity and fluids expulsion seems to focus on this part of the mud volcano field. The shape of the deformed area is consistent with the geometry of the observed structures and help us to establish the rate of deflection/inflection for the investigated time span. To summarize, the DInSAR and structural data suggest that the normal faults accommodate the amplification of the anticline and that they control the rise to the surface of the fluids accumulated in the fold core.

Figure 3.19. Akhtarma-Karadag mud volcano field. (a) Panoramic 3D view looking north. (b) Map view. (c) Vent alignment on the crest of the Akhtarma-Karadag anticline (photo viewpoint in b). (d-e) Normal faults observed during the field survey carried out in June 2010 (photo viewpoint in b). Images in (a) and (b) are extracted from Google Earth®; http://earth.google.it/download-earth.html.

Figure 3.20. Wrapped (panels on the left column) and unwrapped (panels on the central column) interferograms of the Akhtarma-Karadag mud volcano (see Fig. 3.12 caption for explanation). Unwrapped interferograms with low coherence are not reported (white panels). The mean coherence threshold for all the used interferograms is 0.25± 0.05.

3.8 Discussion

The interferograms allow the detection of ground displacement related to mud volcano activity for two distinct cases, particularly (1) during the time span encompassing pre-eruptive phases (Ayaz-Akhtarma and Khara-Zira Island mud volcanoes), and (2) short-lived pulses interrupting a period of quiescence (Akhtarma-Pashaly, Byandovan and Akhtarma-Karadag mud volcanoes).

Regarding the pre-eruptive activity, the analysis of the interferograms has allowed observing the deformation phases connected to the 2005 eruptive event of the Ayaz-Akhtarma mud volcano, as well as the activity up to one year before the eruption of the Khara-Zira Island on 26 November 2006. In both cases, the deformation consists mostly of a relative uplift of the main active zone of the mud volcano. It has long been recognized that mud volcano activity is driven by internal fluid pressure changes, (e.g., Brown, 1990; Dimitrov, 2002; Davies et al., 2007; Mellors et al., 2007) and thus, the deformation of both the Ayaz-Akhtarma and the Khara-Zira Island mud volcanoes is likely linked to an increase of internal pressure, which may also induce the circulation of pressurized fluids into the system. The deformation at the Ayaz-Akhtarma mud volcano is characterized by two adjacent zones of local subsidence and uplift. The uplift phenomenon predominates and is located to the east of the area in which diffuse seepage and discharge of fluids and solid material occur. Uplift continuously grows in intensity, especially in the last two interferograms (Fig. 3.12). Unfortunately, the exact date of the eruption is unknown, and thus, we cannot directly relate our results to the eruptive event, which has been dated approximately 2005 by Aliyev et al (2009). In other cases, the eruption is generally accompanied by a clear signal of subsidence due to the discharge of material and release of gas pressure, as occurs at the LUSI mud volcano (e.g., Abidin et al 2009; Fukushima et al., 2009; Rudolph et al. 2013; Aoki and Sidiq 2014) and at magmatic volcanoes (e.g., Amelung et al., 2000; Sigmundsson et al., 2010; Hreinsdóttir et al., 2014; Jay et al., 2014). By contrast, a nearly continuous uplift of the main active zone is observed at the Ayaz-Akhtarma mud volcano. Two scenarios may explain this behavior: (1) the eruption was a minor event, or (2) the eruption was associated with subsidence but it is unrecorded because it occurred after the date of the last interferogram (10 November 2005). As the eruption catalogue reports that the eruptive event was characterized by the outburst of a huge amount of mud breccias (Aliyev et al., 2009), the second hypothesis is what we believe to be true. Abidin et al. (2009) detected the simultaneous presence of subsidence and uplift during the eruption of the LUSI mud volcano, as explained in section 3.4 of this Chapter, and the uplift aligns with the Watukosek fault system, which caused localized vertical movement. At the Ayaz- Akhtarma, faults also play a fairly important role in the deformation pattern. Interestingly, the observed fault/fracture system broadly occurs within the area which separates sectors with different rates of subsidence. Particularly, this fault/fracture system down-faults the sector affected by the highest subsidence (Fig. 3.21). The fault/fracture may represent either (1) a shallow structure originated to accommodate the ground displacement of the area with the

75 maximum subsidence rate, or (2) the surface expression of a deeper fault intercepting the subsurface fluid reservoir. In the first case, the local subsidence observed at Ayaz-Akhtarma could correspond to a deflation zone, and may be the response to a redistribution of fluids. In the second case, subsidence would be directly related to the movement of a pre-existing normal fault that has possibly been reactivated by the increased fluid pressure regime connected with the pre- eruptive phases. In the latter hypothesis, the surface fractures/faults would represent secondary fault splays connected to the deeper main fault.

Figure 3.21. The location of the fault/fracture detected during the field survey at the Ayaz–Akhtarma mud volcano is shown in an interferogram (superimposed on the Google Earth image) and in a graph displaying the final cumulative Line-of-Sight (LOS) displacements of a cross-section. The trace of the cross-section A–A′ is indicated in Fig. 3.14 (bottom right panel). Structural field analysis of outcrop-scale faults allowed us to infer the fault dip and to identify the down-faulted block.

Similar uplift-subsidence patterns have also been observed in magmatic volcanoes, where uplift is typically associated with magma intrusion in the shallow crust or with hydrothermal fluid injection and circulation. In several large calderas and magmatic volcanoes, such as Yellowstone and Campi Flegrei, volcanic uplift can also generate complex, temporally and spatially, varying patterns of ground deformation, and may be concurrent with the presence of areas of subsidence

(e.g., Wicks et al., 1998; Chang et al., 2007; Hutnak et al., 2009). Apart from the obvious differences related to the presence of magma and the larger dimensions of the magmatic systems, the uplift-subsidence patterns identified at the Ayaz-Akhtarma show some similarities to those observed in the Yellowstone caldera, where the area affected by subsidence could correspond to a deflation zone and the brittle fracturing/faulting accommodates the differential volumetric variations (Chang et al., 2007). Similarly, Brunori et al. (2013) identified deformation patterns of the Cerro Blanco/Robledo Caldera during a resting phase period. In particular, the caldera subsides with decreasing velocity while a positive velocity field is detected in the northwestern part of the system outside the caldera. Regarding the short-lived pulses at the Akhtarma-Pashaly, Byandovan and Akhtarma-Karadag mud volcanoes, these are characterized by ground deformation of considerable magnitude although they do not culminate in an eruption (we have no reports of eruptions around the period analyzed with the interferograms). In particular, the Akhtarma-Pashaly mud volcano underwent an isolated intense deformation event that lasted 3 months approximately, while the Byandovan mud volcano experienced a series of short-term deformation pulses that came to an end within approximately one month. As regards the Akhtarma-Karadag mud volcano field, two events of deformation (each lasting maximum 2 months) localized in the graben affecting the anticline crest. The detected subsidence and uplift patterns are probably controlled respectively by the activity of the crestal normal faults, and the inflation of the rising mud-fluid mixture. Important similarities have been found between these above mentioned cases and a study of the Murono mud volcano, in Japan, carried out with high-precision leveling (Kusumoto et al., 2014). The Murono mud volcano was in a state of quiescence, but the analysis revealed the occurrence of uplift and subsidence events that reached approximately 26 mm and 14 mm, respectively. Most of the benchmarks experienced recurrent uplift and subsidence in a relatively short period of time (6 months) highlighting the occurrence of episodic changes in internal overpressure that may transmit the effects up to the topographic surface. Pulsed fluids moving through long-lived mud volcano systems have been indeed found to be important in the eruptive history as a mechanism of fluid-mud redistribution and expulsion (Evans et al., 2006). In general, it is likely that the quiescence periods of mud volcanoes are cluttered with short-duration phases of deformation, as those mentioned here. Similar types of surface deformation events as discrete, short-duration pulses (shorter than one year) are also experienced by magmatic volcanoes, such as some of the East African Rift where the pulses are repeated over time as multiple inflation and deflation events (Biggs et al., 2009, 2011). Multiple pulses of uplift, several of which did not end in eruptions, occurred, for example,

77 at Sierra Negra in the Galapagos Islands (Amelung et al., 2000), Eyjafjallajökull in Iceland (Pedersen and Sigmundsson, 2006; Sigmundsson et al., 2010) and Cordón Caulle in Chile (Jay et al., 2014). Volcanic deformation preceding eruptions can indeed occur episodically at magmatic volcanoes and a peak in deformation rate does not necessarily imply that an eruption is about to occur within a relatively short time period (a few years or months). In particular, Pritchard and Simons (2004) evidenced that important deformation may occur even during phases of quiescence, and Hutnak et al. (2009) used numerical simulations and inferred that the circulation of hydrothermal fluids may explain the observation that the deformation of some calderas did not culminate in magmatic eruptions. Apart from evident differences (i.e., the role of temperature, crystallization and volatiles), magmatic and mud volcanoes share some processes and properties, including (i) comparable morphologic and internal structure (Stewart and Davies, 2006), (ii) a similar eruptive history (Evans et al., 2006), and (iii) a similar seismic energy density to be triggered off by earthquakes (Wang and Manga, 2010). The similarities in the time-space evolution of ground deformation evidenced for the studied mud volcanoes of Azerbaijan thus strengthen the notion that similar processes govern both igneous and mud volcano systems. The present study indicates that satellite radar interferometry represents a suitable tool for studying mud volcano activity, and the results contribute to a wider understanding of the processes driving ground deformation at mud volcanoes. In the absence of monitoring systems and detailed historical information about mud volcanic activity, satellite-based observation of the superficial deformation could play a relevant role in assessing the hazards related to mud volcanoes. Sentinel-1 radar data, will lead to new challenges in the analysis of ground deformation and will enhance the potential of radar satellite- based technique thanks to improved acquisition parameters and increased satellite repeat times (~12 days).

3.9 Concluding remarks

Since the nineties the Differential Satellite Interferometry (DInSAR) has been extensively used for providing detailed information about the nature of the deformation connected to the eruptive phases of magmatic volcanoes. This work demonstrates that the same methodology can also be used to study the activity phases of the mud volcanoes with excellent results. Moreover, the results of this study highlights some important similarities in the deformation patterns of the mud volcanoes and the magmatic ones. To summarize, the analyzed interferometric data provide new

78 evidence on the deformation patterns of five mud volcanoes in Azerbaijan. The interferograms were selected according to their coherence and span the period from October 2003 to November 2005. The main outcomes of this study are the following: 1. The observed ground deformation patterns of mud volcanic edifices are often characterized by the simultaneous presence of uplift and subsidence areas, which might correspond to inflation and deflation zones, respectively. 2. Important deformation events, driven by fluid pressure and volume variations, can happen in connection with main eruptions. Pre-eruptive deformation consists of marked uplift and occasional minor subsidence that is probably related to subsurface redistribution of pressurized fluids. Ground uplift has been detected to manifest up to one year before the eruption. 3. Ground deformation may also occur as short-lived pulses interrupting a period of quiescence. Discrete pulses of surface deformation could be repeated through the long eruptive history of mud volcanoes as a mechanism of subsurface fluid-mud inflation and redistribution, and may not be directly related to main eruptive events. 4. The results of this interferometric study show that mud and magmatic volcanoes display some similarities in the behavior of ground deformation during quiescence and pre-eruptive stages.

Chapter 4

Study of the ground deformation of the Apennines foothills and of the Emilia thrust-folds using the PSI technique

4.1 Introduction

This chapter reports the study of the recent tectonic activity of the outermost structures of the Northern Apennines fold-and-thrust belt, namely the Pede-Apennine thrust front, which coincides with the morphological boundary between the Northern Apennines and the flat Po Plain, and those in the Po Plain. In particular, lines of thrust folds are buried beneath the thick sedimentary cover of the Po Plain, which represents the shortened foreland of both the Northern Apennines and the Southern Alps. Since the fifties, the buried compressional structures have been extensively explored through commercial seismic reflection lines and well logs for the petroleum industry. This sector is characterized by historical and recent seismicity, with earthquake magnitude up to 6 (Rovida et al., 2011), such as for the 2012 Emilia seismic sequence that nucleated at the toe of the Apennine wedge (e.g. Pondrelli et al., 2012). Compressional strains are prevalent in the Po Plain, as shown by the fault plane solutions of instrumental earthquakes (e.g., Pondrelli et al., 2006; Scognamiglio et al., 2012), in agreement with geological and GPS evidence (e.g. Devoti et al., 2011; Cenni et al., 2013). The strain rates are however rather low (Valensise and Pantosti 2001; Burrato et al., 2003; Basili et al., 2008; Maesano et al., 2015). 80

Assessing the seismic hazard of the external Northern Apennines is crucial due to the concentration of cities, infrastructures, and industrial activities that occur in this sector of the Emilia Romagna Region. Probably due to the scattered nature of local seismicity, much of the Po Plain has entered the Italian seismic code only in 2004, and the new provisions have become mandatory recently, in 2010. In addition the strong impact of human activity on this area (mainly exploitation of water resources) has locally caused significant vertical ground motions (e.g. Stramondo et al., 2007; Bitelli et al., 2014), which complicates the understanding of the seismotectonic setting of the studied region. The Differential Synthetic Aperture Radar Interferometry (DInSAR) has been extensively used for active tectonic deformation assessment (Massonnet et al., 1993; Massonnet, and Feigl, 1998; Price and Sandwell, 1998; Wright et al., 2004; Lyons and Sandwell, 2003). As explained in Chapter 2, DInSAR techniques measures the ground displacement in the Line Of Sight (LOS), and the multi-interferometric approaches grouped in the name of Persistent Scatterers Interferometry (PSI) techniques allows to obtain very accurate displacement measurements at specific points on the ground with millimetric precision (Ferretti et al., 2001, 2007; Werner et al., 2003; Crosetto et al., 2008). PSI technique strongly minimizes the loss of coherence between time delayed acquisitions, but unfortunately limits the use of the technique to mostly urbanized areas (for details see Chapter 2). PSI has been proved to be effective in detecting surface deformation of wide regions involved in low tectonic movements (where it is possible to find a considerable amount of PS) in various geodynamic environments, including orogenic fronts (Bürgmann et al., 2006; Massironi et al., 2009; Vilardo et al., 2009; Bell et al., 2011; Grandin et al., 2012; Peyret et al., 2013; Béjar-Pizarro et al., 2013). As showed in Chapter 2, besides the original Permanent Scatterer technique (PSInSARTM) patented by Ferretti et al. (2001), many different techniques have been developed to process multi-temporal SAR data stacks, such as the Small Baseline Subset (SBAS) described in Berardino et al. (2002) and Lanari et al. (2004), the SqueeSAR method proposed in Ferretti et al. (2011), and other methods described in Kampes (2006), Hooper et al. (2004), Costantini et al. (2000), and Mora et al. (2003). In this work we applied the PS approach (described in Chapter 2) implemented at the Geomatics Division of the Centre Tecnològic de Telecomunicacions de Catalunya (CTTC; Crosetto et al., 2011; Monserrat, 2012; Devanthéry, 2014), to obtain new insights into the present-day deformation pattern and anthropogenic subsidence of the sector between Piacenza and Reggio Emilia. Finally, in order to validate the PSI measurements we have compared our results with GPS data.

4.2 Geological setting of the Northern Apennines

The Northern Apennines chain is a complex fold-and-thrust belt that developed since the Late Cretaceous in a convergence setting due to the closure of the Ligurian-Piemontese (or Western Tethys) Ocean. Since the Middle-Late Eocene the system evolved in a continental collisional setting (Boccaletti et al., 1971; Vai, 2001, and references therein). Thus, the area interposed between the European and the African plates experienced a complex history due to the changes of tectonic stresses related to the plates movements. The continental collision gave rise to a complex eastward vergent thrust system composed of a nappe pile forming the Northern Apennines thrust-fold belt. The uppermost tectonic units of this stack are the ocean-derived Ligurian Units (Jurassic-Eocene; Fig. 4.1) which were deposited onto oceanic crust (i.e., the Ligurian–Piedmontese ocean, Principi and Treves, 1984), together with their satellite basins (the Oligocene-Early Miocene Epiligurian Sequence; Fig. 4.1; Ricci Lucchi, 1987; Boccaletti et al., 1990). The Ligurian Units tectonically overlie the Sub-Ligurian Units (Paleocene-Eocene), which were deposited in the transitional zone from the ocean to the continental passive margin of the Adria Plate. These units tectonically rest in turn over the Tuscan and Romagna Units (Triassic-Miocene) deposited on the continental passive margin of the Adria Plate and consisting of a succession of Oligocene–Miocene siliciclastic foredeep sediments (with thickness from 2500 to 4000 m), and a lower succession of Mesozoic–Cenozoic carbonate rocks (Abbate and Bruni, 1987; Sani et al., 2009). The migration of the nappe pile towards the foreland (i.e. to the east and north-east), was accompanied by the gradual eastward migration of the foredeep, whose sediments were progressively included into the nappe stack (Ricci Lucchi, 1986; Boccaletti et al., 1990). The siliciclastic sedimentation of the foredeep basins consisted of turbidite sediments supplied from the Alps, which are represented by the Macigno Formation and the Cervarola-Falterona Unit (Oligo-Miocene), in the Tuscan Domain, and by the Marnoso Arenacea Formation (Burdigalian-Tortonian) in the Umbria-Marche Domain (e.g., Boccaletti et al., 1990). During the Pliocene, the foredeep basin progressively migrated towards the foreland, i.e. towards the Adriatic sector, following the eastward propagation of deformation (e.g., Boccaletti et al., 1990). At the end of the Early Pleistocene, the geography of the Northern Apennines chain was similar to the present. Since the Middle Pleistocene, the sedimentation in the external sector was mainly related to the Po and Apennine rivers. At present, in northern Italy, the convergence between Africa and Eurasia plates is still active (e.g. Serpelloni et al., 2005; Devoti et al., 2011; Vannoli et al., 2014)

Figure 4.1. Tectonic scheme of the Northern Apennines (after Pieri and Groppi1981; Cerrina Feroni et al., 2002; Boccaletti et al., 2011).

4.2.1 The Emilia-Romagna Apennines and buried thrust folds

The study area extends for about 100 km (Fig. 4.2) in a northwest-southeast direction, between Piacenza and Reggio Emilia, and comprises the Pede-Apennine margin and a wide part of the Po Plain, which is a syn-tectonic sedimentary wedge filling the Pliocene–Pleistocene Apenninic foredeep (Pieri and Groppi, 1981; Cremonini and Ricci Lucchi, 1982). The current geological setting of this area results from the activity of the outer thrusts in the foreland basin (Pieri and Groppi, 1981; Boccaletti et al., 1985, 2004, Fantoni and Franciosi, 2010). The latter shows a monoclinal setting and corresponds to a large and mostly undeformed structural element (Mariotti and Doglioni, 2000). Along the Pede-Apennine foothills, the outcropping rocks belong mainly to the Messinian evaporites and the post-evaporite succession (Formazione di Tetto, Formazione a Colombacci, Argille Azzurre), sealed by the Pleistocene continental deposits of the Po Plain. The sedimentary sequence of the Po Plain consists of syn-tectonic deposits, with a general regressive trend that marks the transition from marine to continental environments (Ricci Lucchi et al., 1982). The continental sequence starts with the deposition of the Lower Emilia- Romagna Synthem (Sintema Emiliano Romagnolo Inferiore; SERI) (650-450 kyr), and ends with the deposition of the Upper Emilia-Romagna Synthem (Sintema Emiliano Romagnolo Superiore; SERS), which mainly consists of fluvial sediments (450 kry to present) (Boccaletti et al., 2004).

Figure 4.2. Structures and subsurface geology of the study area (white box) from the Structural Model of Italy (Bigi et al., 1983). PTF: Pede-Apennine Thrust Front. 84

In this section, a brief description of the geological setting of the external margin of the Apennine chain is given. The outermost part of the Northern Apennines may be subdivided into three sectors corresponding to (1) the Apennine axial zone, (2) the Pede-Apennine margin (Pede-Apenninic thrust front, PTF), and (3) the Emilia and Ferrara fold systems buried below the Po Plain (Fig. 4.2).

1. The axial zone of the chain is the morphologically highest area of the Apennines and essentially consists of the tectonic pile of Sub-Ligurian, Ligurian and Epiligurian Units, above the turbidite successions of the Tuscan, and Umbria-Marche units. This zone is characterized by a major system of thrust faults, developing in correspondence to the main Apennine divide. The basement and the carbonate succession although not outcropping, are both folded and doubled (Boccaletti et al., 2011). The presence of ‘‘tectonic windows’’ and out-of-sequence structures, which may locally reverse the original superposition relations of the Ligurian units onto the Tuscan units, are determined by recent (Pliocene-Pleistocene) compressive tectonics. Furthermore, recent activity of compressive structures is evidenced by deformation or differential uplift of Quaternary deposits or river terraces (Marabini et al., 1986; Bonini 2007). The Northern Apennines axial belt is mostly characterized by moderate rates of uplift (or stability), which are estimated in less than 1–1.5 mm/yr, whereas slightly higher rates have been recognized in the Piacenza and Reggio Emilia regions (Cenni et al., 2013).

2. The Pede-Apennine Thrust Front (PTF) actually separates the emerged portions of the Apennines from the portions buried beneath the sediments of the Po Plain (Boccaletti et al., 1985; Toscani et al., 2009) and consists of a rather continuous NE verging thrust system of stacked thrust sheets (Pieri and Groppi, 1981; Rossi et al., 2002). However, the geomorphic surface expression of this system can be identified only in some zones such as near Bologna and southwest of Reggio Emilia and Modena (Benedetti et al., 2003; Boccaletti et al., 2004; Bonini, 2007). The PTF have been reactivated during Middle-upper Pleistocene and in a more internal position with respect to some more external buried thrusts, and therefore the PTF is probably an out-of-sequence thrust (Vai, 2001). The analysis of the seismic profiles suggests that external thrusts (PTF, Ferrara arc) affect at depth the carbonate sequence and possibly also the basement (Boccaletti et al., 2011; Fig. 4.3).

Along the Pede-Apennine foothills, the outcropping rocks mainly belong to the post- evaporite succession (Formazione di Tetto, Formazione a Colombacci, Argille Azzurre), consisting of prevailing mudstones deposited on the inner foreland margin. This sequence is included between two unconformities of regional importance: the base rests on the Messinian evaporites (Formazione Gessoso-Solfifera), whereas the roof is sealed by continental deposits of the Po Plain. This sequence consists of syn-tectonic deposits, containing numerous stratigraphic discontinuities and transgression/regression cycles, with a general trend that marks the transition from a marine (deep and shallow sea level, cycles) to a continental environment (Ricci Lucchi et al., 1982). The large-scale cycles show in fact a regressive trend and end with the recent alluvial deposits of the Upper Emilia-Romagna Synthem (SERS; Fig. 4.3). From a morphostructural point of view, the Pede-Apennine margin is characterized by two sectors with distinct patterns, extending (i) north-west and (ii) southeast of Bologna. (i) The north-western sector (Emilia Apennines) is marked by the presence of the major Pede-Apennine thrust front (PTF; Boccaletti et al. 1985; Benedetti et al., 2003; Boccaletti et al., 2010; Bonini, 2013), characterized by a prominent morpho-tectonic signature (section A- A’ of Fig. 4.2). At Stradella, 50 km south of Milan, the PTF system outcrop as a reverse and active fault which is traceable for about 35 km (Benedetti et al., 2003). Various authors (eg. Castellarin et al., 1985; DISS Working Group, 2010) report steep inclinations for the PTF (generally >40°, sometimes even 55°) in this sector of the margin. (ii) The southeastern sector of the PTF system (Romagna Apennines) is instead characterized by a monoclinal setting with north-eastern dip, forming tilted surfaces that involve Quaternary deposits (Boccaletti et al., 2011). Other authors (Picotti and Pazzaglia, 2008; Bruno et al. 2011) argue instead, that the PTF is anything but superficial or even outcropping in some locations of the Apennine foothill (e.g. in Bologna).

Figure 4.3. Recent and active structures of the Emilia-Romagna. Subsurface geology in the Po Plain is illustrated as isobaths of the base of the Upper Emilia-Romagna Synthem (450,000 year; modified from Boccaletti et al., 2011).

Along the Emilia Apennines foothills, there are geological indications of recent tectonic activity. The calculation of geomorphic indices of landforms (Gualtierotti, 1998) and the analysis of the evolution of the erosional-depositional system of the chain-Po Plain (Amorosi et al., 1996) were suitable tools to document the activity of the PTF and of the associated active anticlines. Moreover additional insight of current tectonic deformation is shown by fault scarps locally affecting Holocene alluvial fans and by the uplift of different orders of fluvial terraces (Boccaletti et al., 2004; Gunderson et al., 2014). In particular the uplift of the tectonic terraces suggest that the mean incision rates is ~6 mm/year for the 18,000–13,000 year interval, and ~3.5 mm/year for the last 13,000 year (Gualtierotti, 1998). These values give important indications about the activity of the Pede-Apennine thrust system, although they may be also influenced by the fluvial dynamics related to climatic changes. Late Middle Pleistocene to Holocene uplift rates of the mountain front in the northern Apennines have been estimated relying primarily on fluvial incision through deformed geomorphic markers, in particular at the Reno Valley (Picotti and Pazzaglia, 2008), and at Salsomaggiore anticline (Wilson et al., 2009).The value of the uplift ranges from 0.3 to 2.2 mm/year (Ponza et al., 2010). Additionally, the active tectonics of growing anticlines associated with the PTF has been documented using a quantitative analysis of the 87 topography. The analyzed structures appear to be growing at rates of 0.1-0.3 mm/year (Ponza et al., 2010). The principal geological and geomorphological characteristics of the PTF are listed from the western part, moving towards the easternmost part of the margin. Near Piacenza and northwest of Parma, the active structures coincide with the SE continuation of the Broni-Stradella thrust: in particular just south of Piacenza the buried front of the morphological high named Chero-Carpaneto (Fig. 4.3) described by Benedetti et al. (2003), shows late Quaternary activity. South-west of Parma, the PTF and the related anticline coincide with the Apennine morphological margin and with the tectonic window of Salsomaggiore (Fig. 4.3). Between Parma and Bologna, the PTF can be traced as a continuous active front, and geological evidence of active structures are present in the hills just south of the margin: the Middle Pleistocene deposits are folded and faulted (as shown by seismic profiles) and the Late Pleistocene deposits of the Po Plain terraces are tilted (Boccaletti et al., 2011). Near Quattro Castella (Fig. 4.3), prominent faceted spurs (with a mean height of about 50 m, Boccaletti et al., 2011) and a laterally continuous basal scarp occur in correspondence with the PTF. Moreover the activity of this thrust-fault is highlighted by the strong fluvial erosion in the hanging-wall, by the geomorphic indices of landforms (Boccaletti et al., 2004), and by the analysis of seismic profiles. West of Quattro Castella, the PTF bifurcates giving rise to blind thrusts whose recent activity is evidenced by analysis of seismic lines (Barbacini et al., 2002). The greatest depth of the base of Holocene sediments occurs in front of the Apennine margin, between Reggio Emilia and Bologna, suggesting an Holocene activity of the PTF. Other indicators of PTF activity occur at some thrust-related anticlines buried below recent alluvial deposits, such as in correspondence of the Castelvetro di Modena anticline (Fig. 4.3). Here, the argille Azzurre Formation (Pliocene) and the Upper Emilia-Romagna Synthem (SERS, Middle Pleistocene–Holocene) are uplifted and deformed, and the growing anticline produced the deflection of minor streams and inversion of the local drainage direction (Boccaletti et al., 2004). South-east of Bologna, the Imola Sands and the alluvial sediments of the Lower Emilia-Romagna Synthem, (Middle Pleistocene) are clearly deformed and are characterized by vertical or overturned attitude, and are cut by reverse faults (Ghiselli and Martelli 1997; Boccaletti et al. 2004). Active structures affecting Middle-Late Pleistocene deposits are present in eastern Romagna (Ghiselli and Martelli, 1997). In the south-eastern sector of the Romagna Apennines–Po Plain margin, between Bologna and Forlì, the evidence of activity consists mainly in the

blind back thrust localized at the base of the Pliocene sequence and in the Marnoso- Arenacea south of Faenza and Forlì (Fig. 4.3), and the Cesena ‘high’ (Boccaletti et al., 2011).

3. The last sector of the outermost part of the Northern Apennines corresponds to the portion of the Northern Apennine buried below the foredeep deposits. These buried structures consist of a series of thrusts and related folds which have a typical arcuate shape in map view (Fig. 4.2 and Fig. 4.3). Proceeding from west to east, three groups of thrust and folds (Pieri and Groppi, 1981) are identified (mainly through the seismic profiles analysis): the Monferrato folds, the Emilia Folds, and the Ferrara Folds. The latter are generally divided into three different areas namely the Ferrara Arc, Romagna Arc and Adriatic Folds (Pieri and Groppi, 1981). These structures are characterized by anticlines and synclines associations that deform the Plio- Quaternary formations of the Po Plain sedimentary infill (Fig. 4.4). The large sedimentation rates of the foredeep implies a large accommodation space (i.e. a considerable subsidence across the whole Po Plain), which was possible as a consequence of the tectonic load of the Northern Apennines stack (e.g. Bartolini et al., 1996; Ghielmi et al., 2013). Tectonic subsidence rate is currently up to 2–3 mm/yr (Arca and Beretta 1985; Carminati and Martinelli 2002; Stramondo et al., 2007; Cuffaro et al., 2010) and has always been larger than the tectonic uplift rates along the active thrust-related folds. The synclines host the sedimentation of the previously mentioned syn-tectonic deposits with the so called “growth strata” sedimentary geometry. The analysis of these deposits through the interpretation of seismic sections (for example: sections A-A’ and B-B’ of Fig. 4.4), allows the reconstruction of the deformation events that led to the creation of the folds arches. In particular it is possible to reconstruct the deformation history of the buried Apennines from Messinian until present (Argnani et al., 2003).

Figure 4.4. Transverse geological cross sections of the external Northern Apennines. The traces of cross sections are indicated in Fig. 4.2 (modified from Pieri and Groppi, 1981). Q: Quaternary; Plms: Upper Middle Pliocene; Pli: Lower Pliocene; Ms: Upper Miocene; Mm: Middle Miocene; Mi: Lower Miocene; PG: Paleogene; Mz: Mesozoic; L: Ligurian Units.

Starting from a mostly linear compressional front in the lower Messinian, the blind thrusts begin to propagate according to a piggyback sequence (Castellarin et al., 1985), evolving towards the foreland direction (Fig. 4.5). The formation and the migration of the thrusts cause the inclusion of the foredeep sediments into the overthrust system. The current setting, turns out to be almost completed before the end of the Pliocene (Fig. 4.5; Argnani et al., 2003). The subsurface data collected through boreholes and seismic profiles carried out for hydrocarbon and water research, allowed to reconstruct the main Quaternary unconformities. The base of the Upper Emilia-Romagna Synthem (SERS; approximately 450.000 years old), is generally folded and recognizable at regional scale (RER and ENI-Agip 1998). At the top of the Emilia and Ferrara Folds, this surface is very shallow, while it becomes deeper in the footwall of the buried thrust fronts of the PTF-Romagna Folds system and of the Emilia and Ferrara Folds.

Figure 4.5. Evolution of buried thrust fronts. (a) Incipient PTF front with approximately linear trend in the Messinian; (b) advancing front in Pliocene; (c) deformations similar to those currently observable around the Late Pliocene. The dashed line represents a thrust front presumably of Pleistocene age (modified from Argnani et al., 2003).

The Holocene base has been recognized throughout a wide sector of the Po Plain trough C14 analysis on samples collected during numerous drillings (available in the accompanying notes of the geological maps of the Po plain, http://www.regione.emilia-romagna. it/wcm/geologia/canali/cartografia/sito_cartografia/sito_cartografia.htm). West of Reggio Emilia and in western Romagna, the Holocene base is located at a depth less than 9 m, whereas it is deeper (>20 m) along the Adriatic coast, between Ravenna and the Po River delta, as well as in some sectors of the Modena and Bologna plain. Between Reggio Emilia and Bologna and in Romagna, it plunges rapidly immediately north of the Apennines–Po Plain margin (Boccaletti et al., 2011). The good correspondence between the distribution of the anticlines and the geometry of the Holocene base suggest a tectonic control on the geometry of Holocene deposit – i.e., this deposits are thinner over the crest of anticlines (Boccaletti et al., 2004).

Some peculiarities of the hydrographic pattern indicate the ongoing activity of the Ferrara Folds. Some examples are: the direction changes of the Po River near the Bagnolo- Novellara high, and of the Secchia and Panaro rivers near the Mirandola high (see Burrato et al., 2003; Boccaletti et al., 2004; Boccaletti et al., 2011).

4.2.2 Seismicity and main active structures of the outermost Northern Apennines

Several earthquakes have hit the Northern Apennine front and the Po Plain through time (ISIDe Working Group 2010; Rovida et al., 2011, Vannoli et al., 2014), thus demonstrating that the geodynamic processes which led to the subduction of Adria plates below European plate is still active (e.g., Carminati and Doglioni, 2012). More specifically, in the study area, the historical and instrumental catalogues (Fig. 4.6; Rovida et al., 2011; ISIDe Working Group 2010) show that the seismicity concentrates along the buried thrust fronts (Emilia Folds and the Ferrara Folds). The earthquakes registered in the area are characterized by moderate to strong magnitude and compressive/transpressive focal mechanism solutions (e.g. Pondrelli et al., 2006; Fig.4.6). The main recent seismic events are the earthquakes of 15 October 1996 (Mw 5.4) (Ciaccio and

Chiarabba, 2002) and 20 and 29 May 2012 (Mwmax 6.1) (Fig. 4.6; Pondrelli et al., 2012). Seismicity is not evenly distributed, but increases from west to east, following the similar increase in the convergence rates across the whole Po Plain suggested by geodetic, seismological and tectonic evidence (D’Agostino et al., 2008; ISIDe Working Group 2010; Rovida et al., 2011). In particular, the cumulative slip rate (calculated over the past 1.81 Ma) are ~0.70 mm/yr and 0.95 mm/yr for the Emilia and the Ferrara thrust front, respectively. Such an eastward increase in the strain rate (see also Boccaletti et al., 2011) matches the GPS observations, which document an increase of shortening from about 0.5 mm/year across the western Emilia thrust front to about 2.4 mm/year across the Ferrara thrust front (Michetti et al., 2012, Maesano et al., 2015) as well as the number of historical earthquakes and the released seismic moment (Burrato et al., 2004; ISIDe Working Group, 2010; Rovida et al., 2011; Vannoli et al., 2014).

Figure 4.6. Seismotectonic map of the study area. The main structural elements are modified from the Structural Model of Italy (Bigi et al., 1983). The instrumental earthquakes between 1985 and 2015 (yellow circles) recorded by the Istituto Nazionale di Geofisica e Vulcanologia (INGV) are reported (ISIDe catalogue; http://iside.rm.ingv.it). Historical events are indicated by blue squares (CPTI11 catalogue, http://emidius.mi.ingv.it/CPTI11/; Rovida et al., 2011). Black crosses represent the orientation of the minimum horizontal stress (Sh min) deduced from breakout and focal mechanism analysis (Montone et al., 2012). Focal mechanisms are from Regional Centroid Moment Tensor (RCMT, Pondrelli et al., 2006), and quick RCMT on-line (available at: http://autorcmt.bo.ingv.it/quicks.html) project of INGV.

Shallow seismicity (<10 km) is widespread, both in the chain and in the outer sectors. Between 15 and 35 km depth, the distribution of the epicenters is characterized by a lower density with respect to the shallow one (Boccaletti et al., 2004), with compressive and strike-slip focal mechanisms and with compression directions mainly NE-SW. Deeper seismicity (depths greater than 35 km) is widespread in the axial sector and in the Apennine foothills (Boccaletti et al., 2004). The seismicity of the study area is briefly analyzed for the main sectors of the external Northern Apennines:  In the axial zone of the chain, dominant extensional focal mechanisms (Boccaletti et al., 2011 and references therein) suggest the presence of an extensional stress field that obviously postdates the development of the previous compressive structures. Other authors (Picotti and Pazzaglia 2008; Brown et al., 2009) suggest the presence of high- angle normal faults in the Northern Apennines foothill, also through the seismic sections analysis. They interpret these structures as potential seismogenic sources related to an

extensional, shallow stress regime coexisting/accommodating deeper compressive crustal structures.  The Pede-Apennine Thrust Front has a seismogenic potential that is suggested by historical seismicity of varying intensity and with macroseismic epicenters located over the foothills. Important events have occurred on 5 June 1501 (Mw≈6.0, Appennino Modenese), 1 March 1505 (Mw≈5.6, Bolognese), 9 December 1818 (Mw≈5.2, Langhirano), and 4 March 1898 (Mw≈5.4, Valle del Parma), 20 April 1929 (Mw≈5.3, Bolognese), 14 September 2003 (Mw≈5.3, Monghidoro), 23 December 2008 (ML 5.1, Langhirano) (DISS Working Group, 2010; Vannoli et al., 2014).  Regarding the structures buried beneath the Po Plain, the cumulative slip rates increase eastward (Maesano et al., 2015). A similar eastward trend is also shown by the amount of historical earthquakes and by the released seismic moment (Burrato et al., 2004; ISIDe Working Group, 2010; Rovida et al., 2011; Vannoli et al., 2014). The buried thrust system of the Emilia Folds caused a number of damaging earthquakes over historical times, the most important of which occurred on 11 June 1438 (Mw 5.6, Parmense), 11 September 1831 (Mw 5.5, Reggiano), 13 March 1832 (Mw 5.5, Reggiano), 15 July 1971 (Mw 5.7, Parmense), and 9 November 1983 (ML 4.8, Parmense) (Vannoli et al., 2014). The westernmost thrust-faults that belong to the Ferrara-Romagna Arc extend to the NE of Reggio nell’Emilia, forming the lateral ramp of the arc. A number of earthquakes of Mw about 5 was presumably generated by this structure: 10 February 1547 (Mw 5.1, Reggio Emilia), 20 June 1671 (Mw 5.2, Rubiera), 12 February 1806 (Mw 5.2, Novellara), 25 Dec 1810 (Mw 5.3, Novellara), 15 October 1996 (Mw 5.4, Rio Saliceto), and 18 June 2000 (Mw 4.5, Parmense; Vannoli et al., 2014). The 2012 Emilia earthquakes of 20 May (Mw 6.1) and 29 May (Mw 5.9) (Pondrelli et al., 2012) was the largest earthquakes ever recorded instrumentally in the Northern Apennine

thrust front, and were characterized by purely reverse slip, with SH-max axes perpendicular to the western Ferrara-Romagna Arc thrust faults (e.g. Scognamiglio et al., 2012, among others). The aftershock sequence (about 2.500 events) consists of seven M>5 earthquakes, ranging mostly within 12 km of depth (Vannoli et al., 2014). The sequence activated 2 main thrust fault segments for a cumulative length of ~50 km. These seismogenic sources form the shallow, outer thrust front of the NE-verging Northern Apennines.

4.2.3 The Pede-Apennine mud volcanoes

In the outcropping external Northern Apennines, mud volcanism occurs in the form of surface seeps distributed along a belt sub-parallel and adjacent to the potentially seismogenic Pede- Apennine thrust front (Fig. 4.7; Boccaletti et al., 1985, 2004; Selvaggi et al., 2001; Calderoni et al., 2009). In particular. The mud volcanoes, basically localize over the hangingwall of the PTF, most often above the crest of thrust-related anticlines. (Bonini et al., 2007, 2012). The typical conical extrusive edifices, locally called “salse”, have relatively small dimensions: their height rarely exceeding 3 m (Fig. 4.8b, c). The mud volcano emissions normally cluster in a subsided elliptical areas, called “caldera”, of up to 0.6 – 0.7 km of diameter (Bonini, 2012; Manga and Bonini, 2012). The mud volcanoes form over the Ligurian Units (LU in Fig. 4.8a), which act as impermeable barrier generating an important source of overpressures (most of the extruded rock fragments belongs to Ligurian Units, e.g., Kopf, 2002). In the Romagna Apennines, where these rock units are absent, only gaseous (methane) emissions occur (e.g., Bonini, 2007). Geochemical analyses have shown that the greater fraction of the expelled fluids consists of formation water and methane generated in the Marnoso Arenacea sequence (Minissale et al., 2000; Pieri, 2001; Capozzi and Picotti, 2002; Martinelli and Judd, 2004; Capozzi and Picotti, 2010). Mud volcanism along the Pede-Apennine margin is regarded as essentially controlled by tectonic structures (Martinelli et al., 1995; Capozzi and Picotti, 2002; Martinelli and Judd, 2004; Bonini, 2012). Open macroscopic and mesoscopic size fractures or faults, are indeed necessary to allow the vertical mobilization of the mud and gases. Mud volcano features have also been used as stress indicators, and the obtained SH trajectories trend sub-parallel to the World Stress Map (WSM) database, as well as the P-axes of earthquakes distributed around the Pede-Apennine margin (Bonini, 2012).

Figure. 4.7. Location and geological setting of the mud volcanoes of the northern Apennines (slightly modified from Manga and Bonini, 2012). 1, Rivalta; 2, Torre; 3, San Polo d’Enza; 4, Casola Querzola; 5, Regnano; 6, Montegibbio; 7, Nirano; 8, Montebaranzone; 9, Centora; 10, Puianello; 11, Canalina; 12, Ospitaletto; 13, Dragone di Sassuno; 14, S. Martino in Pedriolo; 15, Sellustra valley; 16, Casalﬁumanese; 17, Bergullo; 18, Rubano; 19, Pedriaga; 20, Macognano.

The Pede-Apennine mud volcanoes currently show a quiescent activity. There are, however, historical chronicles that reported of violent eruptions and outbursts (Manga and Bonini, 2012). Evidence for large past eruptions is also recorded by large decimeter rock clasts preserved in erupted mud. In order to transport these large clasts to the surface, the inferred ascent speed of mud and hence eruption rate was at least three orders of magnitude greater than at present (Manga and Bonini, 2012). This phenomenon could represents a local natural hazard, as the eruptions may have the ability to damage infrastructures that characterize such a densely populated areas (Manga and Bonini, 2012). However, the mud volcanoes of Emilia Romagna region are also a touristic attractions, drawing tens of thousands of visitors each year (Castaldini et al., 2005).

Figure 4.8. The Pede-Apennine mud volcanoes. (a) Ideal vertical section through a fold and conceptual relationships between brittle structures, fluid pressure and mud volcanism at the core of a thrust anticline, inspired to the Northern Apennines setting (modified from Bonini, 2012). The encircled numbers indicate (1) bottom of the main fluid conduit or diatreme, and pressurized (2) deep and (3) shallow fluid reservoirs. MA, Marnoso Arenacea sandstones; LU, Ligurian Units; EPL, Epi-Ligurian sequence; and PPT, Pliocene–Pleistocene claystones. (b) Nirano and (c) Regnano mud volcanoes.

4.3 Previous works using satellite interferometry in the study area

The surface deformation in the Po Plain had been already studied by means of the major satellite based interferometric techniques. Strozzi et al. (2001) studied the subsidence phenomena in Bologna city and in the surrounding area by means of DInSAR. The phenomenon of the subsidence has also been monitored through the SqueeSAR™ technique, by Bitelli et al. (2014). They used Radarsat radar images over the whole Emilia-Romagna plain territory in order to obtain a subsidence map at regional scale (Fig. 4.9). This map has been very useful for a comparison with our results (see section 4.5.2). Teatini et al. (2005) studied the subsidence of the Venice coastland by integrating different monitoring systems as leveling, GPS and InSAR (using ERS1 and ERS2 images) techniques.

Figure 4.9. Soil vertical movement velocity map (mm/year) for the period 2006-2011 from Bitelli et al. (2014).

In order to study the temporal evolution of a deformation phenomenon in the area near Bologna, Stramondo et al. (2007) used the SBAS (Small BAseline Subset) technique employing ERS1 and ERS2 images from 1992 to 2000. They highlighted that the subsidence in the Po Plain is due to both natural-tectonic and human-induced causes. Tectonic subsidence is the result of sediment compression and loss of interstitial fluids produced by loading, natural compaction and tectonics. The detected amount of such displacement lies in the range of 2–5mm/year in the area close to the piedmont Sabbiuno anticline, which coincides with an active thrust. In particular, Stramondo et al., (2007) argued that the main cause for the movements observed in this area is the tectonic (interseismic) activity. Concerning the human-induced causes, these can be identified in the hydrocarbon and groundwater extraction, producing sediment compaction with at least one order of magnitude higher with respect to the long-term natural processes. Near Bologna, indeed, the industrial and agricultural activities lead to a larger exploitation of water resources, and therefore the surface subsidence rates reaches a maximum of 59 mm/year in the NE industrial and agricultural areas mm/yr (Stramondo et al., 2007). The Po Plain has been investigated in zones nearby our area of study by means of InSAR monitoring systems in order to detect evidence for ongoing ground uplift indicative of tectonic

98 activity. In particular, Perrone et al., (2013) studied the boundary between the Western Po Plain and the Cottian Alps through the PS-InSAR (Permanent Scatterers Interferometric Synthetic Aperture Radar) technique. They detected uplift in the hanging-wall of active thrust faults, and compared the results with geological, seismological and GPS data in order to analyze the current tectonics and the crustal mobility. Pezzo et al. (2014) studied the earthquake cycle in a portion of the Po Plain affected by the seismic sequence of 2012 by means the satellite interfometry (in particular by the SBAS processing). This zone overlaps in part with the study area of this Ph.D. thesis (area between Parma and Reggio-Emilia), and therefore, this work has been helpful to validate our results. The mosaicked Envisat and ERS mean velocity maps relative to the 1992-2010 time span show the strong subsidence patterns due to the water pumping in the Po Plain, and a relative uplift pattern of the Apennines front, with respect to the Po Plain. Moreover, they observe a rough correspondence between the growth anticlines and the uplift pattern. Finally, satellite interferometry has been used in several studies to analyze the coseismic deformation produced by the main earthquakes of the 2012 Emilia sequence (Bignami et al., 2012; Pezzo et al., 2013; Tizzani et al., 2013; Fig. 4.10), by exploiting the high sensitivity to deformation of the X-band and the short temporal baselines of the COSMO-SkyMed imagery, and by exploiting the widest spatial coverage of Radarsat imagery. The coseismic surface displacement along the LOS (Fig. 4.10) corresponded to a maximum uplift of about 20 cm (Bignami et al., 2012), and the location of the coseismic deformation essentially corresponds to the Mirandola and Ferrara folds, located under the Po alluvial plain. The activity of these structures supports the long-term geomorphic analyses that attribute to the growth of the same folds the wide northward bend of the Po River and the deviation of the Secchia and Panaro rivers (Burrato et al., 2003). Pezzo et al. (2013), hypothesized that the evolution of the hydrographic network have been progressively controlled by repeated coseismic uplift events, rather than by aseismic creep, because no evident topographical bulge corresponds to the buried anticlinal crests. This implies that the net fold growth is lower than the sedimentation rate. In fact, for the Mirandola anticline, Scrocca et al. (2007) evaluate a relative tectonic uplift of 0.16 mm/yr in the last 125 Ka, much lower than the estimated sedimentation rate (0.89 mm/yr).

Figure 4.10. Radarsat wrapped differential interferogram from Bignami et al., (2012). Red stars, position of May 20 and 29 mainshocks; red lines, position of the main thrust fronts; black rectangles, surface projection of modeled faults. Inset: The N-S simplified geological section runs across the epicentral area of the May 29 mainshock, showing the geometry of the buried outer thrust fronts of the northern Apennines. PTF: Pedeapenninic Thrust Front; MTF: Mirandola Thrust Front.

4.4 The used dataset and PSI technique

As already stated in Chapter 2, the Persistent Scatterers technique takes advantage of long temporal series of SAR data, acquired over the area of interest, to filter out atmospheric artefacts and to identify a sub-set of image-pixels where high-precision measurements can be carried out (the persistent scatterers, PSs). The method allows the generation of a mean deformation velocity map with millimeter precision, and a deformation time series for each PS. For our aim we selected 34 Envisat images on descending acquisition geometry (track 437, frame 2709; Fig. 4.11) spanning from September 2004 to September 2010. We computed all the possible interferometric pairs (about 300 interferograms) from the available dataset. The topographic phase has been removed by using the 90-m pixel size SRTM DEM.

100

Figure 4.11. Footprint of the Envisat descending images (red rectangle; track 437, frame 2709), used in this work, and the study area (yellow rectangle).

In the order to obtain a more complete dataset and to improve the robustness of the results, 21 Envisat ascending (track 215, frame 891), covering a time period ranging from July 2004 to March 2010, has been processed. The results of the ascending dataset will not be included in the following sections of the data analysis, because they revealed some particular inconsistencies with the results of the descending dataset and with the other studies. The nature of the problems of the ascending dataset will be described in the Appendix at the end of this chapter (section 4.7). The used PS approach (implemented at the CTTC) has been described in Chapter 2. The main steps of the procedure are shown in figure 2.10. As stated in the paragraph 2.4.1, the point selection criterion was based in the dispersion of amplitude (DA) parameter. In this work, the PS analysis was applied to points with DA lower than 0.3. After the atmospheric phase component (APS) estimation applied to a sequence of spatial and temporal filters, and after removing it from the original interferograms, the deformation velocity and residual topographic error (RTE) have been estimated. In this work, only the points with temporal coherence (γ) higher than 0.7 have been used. Although this rather high threshold can drive to a critical loss of measurements in areas affected by strong non-linear deformations, we consider that this is not a critical issue for this study, because we are interested mostly on linear deformation pattern.

101

The main outputs of the procedure are two: (1) the deformation velocity map that provides the deformation rates in mm/yr for each measured PS, and (2) the deformation time series describing the PS behavior during the measured period. Finally, we geocoded the data, representing them by averaging pixels of 100 x 100 m.

4.5 Data Analysis

Fig. 4.12 shows the PSI linear deformation velocity map of ground movements of the PS projected onto the satellite LOS. The reference point has been selected in a stable area at Collecchio, where a GPS station (COLL) measured about zero deformation from 2007 to 2014. The linear deformation velocity map is characterized by a high density of PS in the cities and in the urban settlements, while PS density gradually degrades outside. This can be explained by the frequent surface changes due to agricultural activities in the Po Plain, which affect the temporal correlation of the SAR signal. The velocity map (Fig. 4.12) shows some clear patterns ascribable to different phenomena. The red zones (negative values) have moved away from the satellite and the cyan parts of the map (positive values) indicate zone that moved toward the satellite. We consider only vertical movements, given that the LOS is inclined at an angle close to the vertical and the horizontal displacements are significantly underestimated. This implies that, assuming a purely vertical deformation, the red zones are indicative of subsidence, while the cyan zones indicate a relative ground uplift. Given the sensitivity of the sensor and the standard deviation analysis of the velocity values for the stable areas, the values of mean velocity between -1 and 1 mm/yr have been substantially considered as stable (green color). The time series for each PS have been computed to estimate the temporal evolution of the deformation. In particular, the time series show the relative LOS displacement of each PS relative to the first image. The analysis of the displacement time series have also been useful for the comparison with GPS data (for a description of the time series results, see the following

sections 4.5.1 and 4.5.2).

102

Figure 4.12. (a) PS mean velocity map in the satellite LOS, from Envisat archives in descending orbit. The reference point of the PSI linear deformation velocity is also shown (black star). The data have been plotted onto a shaded relief from a 10 m pixel Digital Terrain Model provided by the Regione Emilia- Romagna. (b) Same map of panel (a) superimposed onto the Structural Model of Italy (Bigi et al., 1983).

103

4.5.1 PSI and GPS data comparison

Owing to the capability to detect millimetric displacements over long periods and large areas, advanced DInSAR techniques can be considered complementary to conventional techniques (leveling, GPS) for monitoring ground movements. In particular, advanced DInSAR analyses make it possible to obtain very precise vertical displacement measurement on a dense network of point of measurements, without installing any ground equipment. The integration of different methodologies have been widely employed to map land deformation in different contexts, such as tectonic assessment (e.g., Burgmann et al., 2006; Wright et al., 2012; Peyret et al., 2013), volcanic activity (e.g., Pagli et al., 2006; Lagios et al., 2013) and subsidence phenomena (e.g., Zerbini et al., 2006; Bock et al., 2012). In order to improve the reliability of the outcomes shown in the previous paragraph, we compared the PSI results with those derived from the analysis of the GPS permanent stations that are present in the study area. The GPS benchmarks occur at Parma (PARM), Reggio Emilia (REGG), Collecchio (COLL), and Gualtieri (RE01) towns (Fig. 4.13, Baldi et al. 2009; 2011; Cenni et al., 2013). The daily position time series of these four sites have been estimated (together with several permanent stations located in the Italian peninsula and surrounding regions) with the method described in Cenni et al. (2013). The GPS time series have been performed using the series of geographical components (North, East and Vertical) of the site position, without outliers and steps due to instrument changing or/and earthquakes (see Cenni et al., 2012; 2013 for the processing details).

Figure 4.13. Location of the GPS benchmarks that occur in the study area. PARM, south of Parma; REGG in the northern part of Reggio Emilia city; COLL at Collecchio; RE01 north-west of Novellara. 104

In order to compare the PSI time series with the GPS ones, the North, East and Vertical GPS velocity components have been projected along the Envisat LOS in descending geometry, as illustrated in figure 4.14.

Figure 4.14. Projection along the LOS of the North, East and Vertical GPS components.

As stated earlier, owing to its roughly stable pattern, the Collecchio benchmark (Cenni et al. 2012; 2013; COLL in Fig. 4.13) has been chosen as reference point (supposed motionless) for the PSI data processing. In order to refer SAR and GPS at the same reference point, we subtracted the North, East and Vertical components of COLL site to each of the remaining GPS stations included in the study area. Successively we compared the LOS GPS time series of these stations obtained through this procedure with those estimated from the PSI data (Fig. 4.15). The PSI time series were performed among the average of the PSs nearest to the GPS stations, possibly within a square of 400 x 400 m around the estimated position of the GPS receiver, as it may be difficult to identify a radar target in the same location of a GPS station. It is worth noting that the GPS time series also contain a periodicity, which is not present in the filtered PS time series. This periodicity may be due to the characteristics of the substrate or to the characteristics of the GPS station. The pattern of the GPS and PSI time series are well comparable. Specifically, the PARM and the REGG time series show a very good agreement (Fig. 4.15). A poor correlation exists instead for the RE01 time series (Fig. 4.15). The PSI time series is rather noisy

105 and may be influenced by some issues affecting the quality of the data: the RE01 station is indeed located on the edge of the frame, where the spatial filter used to remove the atmospheric error may not have correctly operated. Also, in this area, the number of PSs is rather low.

Figure 4.15. Comparison between the PSI and the GPS time series (from September 2004 to September 2010) projected along the satellite Line of Sight, taking COLL benchmark as reference.

4.5.2 PSI results overview and interpretation

In order to better observe the areas affected by ground deformation and to interpret the results, we have interpolated the velocity values to obtain an iso-kinematic map (Fig. 4.16a). The latter was derived in GIS environment by using an inverse distance interpolation weighted method (IDW). The IDW method is commonly used to interpolate points that do not necessarily have relationship or influence over neighboring data values (e.g. Vilardo et al., 2009). Therefore IDW ensures the preservation of local variations for spatially non-homogeneous distributions of points (Franke,1982; Mueller et al., 2004). We used a quadratic weighting power within a 600 m radius neighborhood that has been chosen iteratively. The velocity values were subdivided into 4

106 classes to highlight the ground deformation patterns (see histogram in Fig. 4.16b). In order to interpret the possible causes of the observed ground deformation, the PSI results were compared with the geological, tectonic and seismological data available in literature. We observe negative values of velocity that correspond to strong subsidence patterns in a wide area of the Po Plain, located in the eastern part of the study area, particularly around the cities of Reggio Emilia, Correggio and Carpi. The detected maximum subsidence rates are respectively - 12 mm/yr, -14 mm/yr and -13 mm/yr. The subsidence detected in the eastern part of the study area is a well-known phenomenon, which had already been highlighted by other studies (Strozzi et al., 2001; Carminati and Martinelli, 2002; Stramondo et al., 2007; Baldi et al. 2009, 2011; Bock et al., 2012; Bitelli et al., 2014). The subsidence in the Po Plain is due to both natural and human-induced causes. Natural subsidence is the result of sediment compaction and loss of interstitial fluids produced by loading and tectonics. The detected amount of such displacement is of the order of 2–5 mm/yr in the area close to the piedmont Sabbiuno anticline, nearby Bologna (Stramondo et al., 2007). The human-induced causes can be identified in the hydrocarbon and groundwater extraction, producing sediment compaction with at least one order of magnitude higher than that due solely to long-term natural processes (Carminati and Martinelli, 2002). Near Bologna, industrial and agricultural activities lead to a larger exploitation of water resources, and therefore the surface subsidence rates reaches 40–50 mm/yr (Stramondo et al., 2007). It is possible that a tectonic subsidence signal may also exist in the study area, but the remarkable negative velocity values (-13/-14 mm/yr of subsidence) are dominantly attributable to the exploitation of water resource. More specifically, this pattern is mainly concentrated where the thickness of soft sediment is greater (i.e., at the towns Correggio and Carpi) and the depth of the base of Pliocene increases (Fig. 4.12b and Fig. 4.16a), particularly in correspondence of the large syncline south of the Ferrara arc (Fig. 4.2). Our results were also compared with the soil vertical movement velocity map of the Po Plain sector of the Emilia Romagna Region for the period 2002-2006, created through the use of Envisat and Radarsat data (Bissoli et al., 2010; http://www.arpa.emr.it/), and with the updated regional subsidence map for the period between 2006 and 2011, obtained with the integrated use of continuous GPS stations and SqueeSAR™ analysis of Radarsat images (Bitelli et al., 2014). Both the shape of the areas subjected to subsidence, and the detected mean velocity values (e.g., from -12.5 to -17 mm/yr of subsidence in the towns of Correggio and Carpi) are in good agreement with our results. A relative uplift pattern has been detected in several zones of the study area, where the deformation velocities are moderate and range from 1 to ~2.8 mm/yr (Fig. 4.16a). Interestingly,

107 most of the uplift areas occur in correspondence of structures belonging to the main thrusts systems described above, particularly the Pede-Apennine Thrust Front (PTF), the Emilia Folds and the Ferrara Folds. Given this spatial coincidence, we infer a correlation between the activity of some thrust-related anticlines, buried below the Po Plain deposits, and the ground deformation pattern immediately above these structures (Fig. 4.16a). As already mentioned in the paragraph 4.3, Pezzo et al., (2014) studied a wide area in the Po Plain through the use of ERS, Envisat and COSMO-SkyMed images processed with the SBAS technique. Importantly they observed a rough correspondence between the growth anticlines and the positive velocity values. Our study area overlaps with the velocity map of Pezzo et al., (2014) between Parma and Reggio Emilia, and we find a good agreement for most of the observed ground deformation. In addition, most of the uplifting areas over the Emilia folds detected in this study have been also reported as local zones of tectonic uplift in the geomorphological map of Castiglioni et al. (1997).

Figure 4.16. (a) Interpolated LOS linear deformation velocity map (IDW method). The GPS benchmarks RE01, COLL, REGG, PARM are indicated in blue. (b) Frequency histogram showing the statistical distribution of velocity values and the classification criterion of the PS velocity values.

Figure 4.17 shows the time series of the uplift zones, representing the LOS motion component of PS as a function of time. The time series were calculated as an average of the four PS centered in each zone of uplift. Most of the time series have an approximately linear trend with constant angle of slope (that represents the mean velocity value), indicating a roughly constant movement towards the satellite during the six years of measurements. Such patterns may represent a 108 relatively slow tectonic movement, although other movements involving the superficial layers of the subsoil may combine with the tectonic signal. Specifically, the cessation or a significant reduction of water pumping (e.g., Gambolati and Teatini, 2015) could be invoked to explain the origin of the ground uplift movement in the surveyed areas. The uplift induced by the groundwater recovery after long-term intensive pumping originated from relaxation of elastically compressed aquifer materials (elastic rebound; Allen and Mayuga, 1969; Waltham, 2002) has been monitored through InSAR technique (Schmidt and Burgmann, 2003; Bell et al., 2008). This phenomenon is characterized by a transition from subsidence to uplift (Chen et al., 2007), as reported in Venice, Italy (e.g., Gatto and Carbognin, 1981). This specific trend is not present in the time series (Fig. 4.17), nor it is reported in literature for the study area, and thus we have no evidence to support this hypothesis. Another anthropogenic factor that may cause ground uplift may be related to underground gas storage. The storage of natural gas can be realized in depleted gas fields, which are present in the area of study. However, the concession areas of storage shown in the map of the mineral rights in force of the Emilia Romagna (downloaded at http://unmig.sviluppoeconomico.gov.it/unmig/cartografia/cartografia.asp; updated to August 31, 2013) do not correspond to those affected by uplift, and therefore this possibility can be ruled out.

Figure 4.17. PSI filtered time series of the uplift zones in the analyzed time span (from September 2004 to September 2010).

The zones affected by uplift are described below by reference to the main thrusts systems: 109

(i) Pede-Apennine Thrust Front. The surface deformation pattern of the morphological boundary between the Apennine and the Po-Plain is characterized by a substantial stable setting during the investigated time span (2004-2010). There are however some important exceptions, particularly the town of S. Giorgio Piacentino, in the westernmost part of the study area, is characterized by an uplift pattern covering an area of ~30 km2, reaching a maximum velocity of about 2.8 mm/yr (Fig. 4.12 and Fig. 4.16a). These uplift values are in agreement with the soil vertical movement velocity map of Bitelli et al. (2014), which indicate for this area values between 2.5 and 5 mm/yr. As shown in Figure 4.12 and 4.16a, the ‘S. Giorgio Piacentino uplift zone’ is located over the hanging-wall of a thrust fault. Moderate positive velocity values (up to 1.3 mm/yr) are also recorded around Sassuolo, which is situated on the morphological boundary between the Po Plain and the Apennines (Fig. 4.12 and 4.16a). Moderate negative velocity values (mean of about -2 mm/yr) are instead detected in correspondence of the Taro Valley and Salsomaggiore. This pattern may be due to a deformation (pumping-induced subsidence due to the compaction of the river sediments) or rather to a residual atmospheric disturbance due to the particular location of this zone, which is between the hilly region and the plain, at the border of the study area. Owing to these conditions, the spatial filter used to remove the atmosphere component might have not worked correctly, therefore we have not enough information to discard the presence of an atmosphere disturbance. (ii) Emilia Folds. A relatively large area (approximately 45 km2) of the Po Plain between Parma and Fidenza shows an uplift pattern that ranges from 1.3 to 2 mm/yr (Fig. 4.12 and 4.16a). This area also includes the small village of Fontanellato and is therefore referred herein to as ‘Fontanellato uplift zone’. Here, several thrust faults form a morphological high (Bigi et al., 1983), and some faults are considered to be active (Boccaletti et al., 2004, 2011). This sector is however essentially devoid of instrumental earthquakes, while it was hit by the historical earthquake of 1438 (Mwm ≈ 5.6, Guidoboni et al., 2007) (Fig. 4.6). The thrust faults affect the sedimentary succession up to the most recent levels, as illustrated in the interpreted seismic profile 1-1’ (Fig. 4.18a), in which the SERS reflectors are clearly deformed. Cross section A-A’ in Fig. 4.18a represents a LOS velocity profile traced parallel to the seismic section 1-1’. We used an interpolated map of the velocity values (IDW method) with a 1200 m radius (i.e., twice the radius used for the map in Fig. 4.16a) to obtain a continuous profile of the ground deformation. The LOS profile has been projected onto the seismic line (about 3 km to the west) along structural strike (i.e., the mean trend of the thrust-faults). The comparison between the LOS and the seismic profiles shows that the Fontanellato uplift zone basically coincides with the crest of the buried thrust anticline (Fig. 4.18).

110

Figure 4.18. Interpreted seismic profiles located near the deformation zones identified through the PS data. (a) Section 1-1’, 3 km west of ‘Fontanellato uplift zone’; SERS: Upper Emilia-Romagna Synthem, SERI: Lower Emilia-Romagna Synthem, Q1, Q2 and Q3 = Quaternary Marine Sequence (after Regione Emilia-Romagna, and ENI AGIP, 1998). (b) Section 2-2’, 5 km east of the ‘Monticelli uplift zone’ (interpreted seismic line PR-314-80v, Progetto ViDEPI, http://unmig.sviluppoeconomico.gov.it/videpi/cessati/sismica_titoli.asp). The graphs above the seismic lines illustrate the cross sections of the LOS velocity patterns (see Fig. 4.16a for the location). The LOS profiles have been projected onto the seismic sections along the mean trend of the thrust faults.

111

The Monticelli village and surroundings areas (Fig. 4.12) are affected by uplift from 1.6 to 2 mm/yr. This zone is clearly located at the top of a thrust-anticline that represents a morphological high of the substrate, such that the Quaternary sediments are here only a few tens of meters thick (Petrucci et al., 1975; Calda et al., 2007). The Monticelli anticline is connected with an active thrust fault (Boccaletti et al., 2004), and is well imaged along a seismic reflection profile 2-2’ (Fig. 4.18b). In particular, the Monticelli uplift zone occurs over the thrust imaged in the seismic line 2-2’, which is situated almost 5 km to the east. Cross section B-B’ in Fig. 4.18b (obtained using the same procedure as for section A-A’) illustrates a LOS velocity profile along a transect crossing the Monticelli anticline, and parallel to the seismic section 2-2’. The comparison between the LOS profile and the seismic section shows that the Monticelli uplift zone corresponds to the crest of the thrust-related anticline (Fig. 4.18b). An uplift zone of ~30 km2 occurs east-southeast, approximately along the same structure, near Cavriago and in the southwestern part of Reggio Emilia city (Fig. 4.16a). In this sector, ground uplift ranges from 1 to 1.8 mm/yr, and occurs over an area close to a NW-SE trending thrust system. A number of historical earthquakes struck the Reggio Emilia city, the largest of which was the Mwm=5.14, of 1547 (Fig. 4.6; Guidoboni et al., 2007). It is possible that the structures producing the current uplift are the same that originated the historical earthquakes. Finally, a moderate uplift signal affects the northwestern part of Parma. (iii) Ferrara Folds. Although the area of study comprises only the western part of the Ferrara arc, it is worth noting that some signals of uplift (up to 1-1.5 mm/yr of velocity) occur in correspondence to the arcuate-shape main thrust front, at the town of Novellara. Several instrumental and historical earthquakes are reported in this zone (Fig. 4.6), and together with geomorphic data mentioned in the previous paragraphs (Castiglione et al., 1997; Burrato et al., 2003; Scrocca et al., 2007; Boccaletti et al., 2011), indicate that this structure is potentially seismogenic. To summarize, the PSI results reveal that some buried thrust anticlines along the Pede-Apennine margin and Po Plain are characterized by surface uplift. In these structural settings, ground uplift is often taken as evidence for ongoing tectonic activity (e.g. Massironi et al., 2009; Champenois et al., 2012; Grandin et al., 2012; Perrone et al., 2013; Peyret et al., 2013), an interpretation that is in good agreement with the various evidence (geological, seismological, morphological) supporting the activity of these structures (e.g., Benedetti et al., 2003; Burrato et al., 2003; Boccaletti et al., 2004, 2011; Basili et al., 2008). The areas of surface uplift are also characterized by historical and current seismic activity (in Fig. 4.19 the comparison of LOS deformation velocity map with instrumental and historical seismicity is shown).

112

Figure 4.19. Comparison of LOS deformation velocity map with instrumental (white circles) and historical (blue squares) earthquakes (for legend and references see legend and caption of Fig. 4.6).

In particular, the macroseismic epicentre of several historical earthquakes are localized along the Emilia Folds, and partly along the Ferrara Folds. The instrumental earthquakes are instead rather infrequent along the Emilia Folds, whereas in the Ferrara Folds show large clusters that followed the main seismic events of October 1996 and May 2012. On the basis of these observations, a working hypothesis may involve that the active Emilia and Ferrara thrust folds would be characterized by inter-seismic periods basically dominated by aseismic creep. The analysis of the 2012 seismic sequence apparently corroborates this possibility, in that the area above the thrust fault rupture was uplifting before the earthquake but it was essentially devoid of previous instrumental events. In particular the analysis of the deformation maps of Pezzo et al. (2014) and of Bitelli et al. (2014) reveal that the areas above the thrust faults that ruptured on 20 and 29 May 2012 (just outside our area of study) were slightly uplifting in the period 1992 - 2010, thereby prior to the earthquakes. Cenni et al., 2013 and 2014 identified minor uplift of the order of ~0.5 mm/yr in the zone of Mirandola through GPS data, for the period 01 January 2001 - 30 April 2012. Also the GPS vertical velocities map of Devoti et al. (2011) indicate minor positive values for this area. On this basis, a preliminary interpretation of the 2012 earthquake sequence would be that fault rupturing was preceded by aseismic uplift. The

113 lessons from the latter case may suggest that the nowadays uplifting thrust folds discussed in this study may exhibit a similar behaviour, even though we cannot establish whether these will accommodate deformation through aseismic creep or through episodic earthquakes. In this regard, Sentinel-1 satellites will provide new high quality data (i.e. improved SAR data availability, larger coverage, shorter temporal sampling) for developing and testing models of earthquake preparation processes, which will generate a strong increase in the observation of crustal deformation worldwide (e.g., Tolomei et al., 2015).

4.6 Concluding remarks

The Pede-Apennine margin of the Northern Apennines and the southern part of the Po Plain are tectonically active areas. This work has attempted to correlate the superficial deformation signals measured by radar satellite-based sensor with the known geological features. On the basis of the analysis of both the linear deformation velocity maps and the PSI time series, the ground deformation has been analyzed over a wide sector of the Po Plain, specifically between Piacenza and Reggio Emilia. The PSI technique revealed some areas of relative ground uplift with velocities ranging between 1 and 2.8 mm/yr. The uplift deformation zones are mostly located in correspondence of the belts of subsurface active thrust folds, and thus we hypothesize a correlation between the observed uplift deformation pattern and the growth of the thrust-related anticlines. As the uplift pattern corresponds to known geological features, it can be used to constrain the seismo-tectonic setting. The ongoing uplift identified for some thrust folds does not imply that this will necessarily lead to an earthquake, yet the results of the current analysis should be taken into proper account when evaluating the seismic hazard of the study region. The PS analysis also reveals an important subsidence area, the origin of which is primarily linked to the exploitation of water resources. The areas subjected to the maximum velocities of subsidence approximately coincides with a large synform south of the Ferrara arc, where the sedimentary package is thicker.

114

4.7 Appendix: Envisat ascending dataset

The Envisat ascending dataset has been also considered for this study, and the same technique adopted for the descending dataset has been applied. 21 Envisat ascending images (track 215, frame 891; Fig. 4.20), covering a time period ranging from July 2004 to March 2010, have been processed.

Figure 4.20. Footprints of the Envisat ascending (track 215, frame 891) and descending images (track 437, frame 2709) used in this work (red boxes), and the study area (yellow box).

Fig. 4.21 shows the PSI mean deformation velocity map of ground movements of the PS projected onto the satellite LOS. Similarly to the descending dataset, the reference point has been selected in a stable area near Collecchio. The velocity map (Fig. 4.21) can be read as the descending one; assuming a purely vertical deformation, the red zones (negative values) are indicative of subsidence, while the cyan zones (positive values) indicate a relative ground uplift. The results of the ascending dataset were not considered reliable, because they revealed some important inconsistencies with the studies from the literature and with our results of the descending dataset. Therefore the ascending data have not been used for the interpretation. In particular, the comparison between the velocity map resulting from the ascending data processing, and the previously illustrated descending velocity map (paragraph 4.5; Fig. 4.12) 115 showed some differences. Several areas showing clear ground motion in the ascending velocity map do not fully coincide with the zones affected by ground deformation in the descending map, nor with the soil displacement maps available in the literature (Bitelli et al., 2014; Pezzo et al., 2014). Despite the comparison with the descending dataset (actually used in this thesis) is not entirely inconsistent, the identified discrepancies led us to investigate further about the reliability of the ascending results. The processing chain of PSs has been indeed controlled and re- performed in order to discover potential errors.

Figure 4.21. (a) PS linear velocity map in the satellite LOS, from Envisat archives in ascending orbit. The reference point of the PSI linear deformation velocity is also shown (white star). The underlying image is from Google Earth. The red lines show the main tectonic structures from the Structural Model of Italy (Bigi et al., 1983).

In order to estimate the evolution of the deformation (Fig. 4.21) during the analyzed time span, the time series for each PS have been computed. In most cases, these time series are not in agreement with the corresponding one of the descending dataset, as we can see in the Fig. 4.22.

116

Figure 4.22. Comparison between ascending and descending PSI non-filtered time series. Some time series show remarkable inconsistencies between the two acquisition geometries.

The final assumption is that the results of the ascending datasets would be biased by some issues that are probably linked to some processing step, or to a tilt in the ascending deformation velocity map, which can be caused by uncompensated orbital errors (as mentioned in Chapter 2, paragraph 2.4.1, point 5). These issues affect the reliability of the deformation measurements. To summarize, as we can not avoid and correct the orbital errors which probably affect the Envisat ascending images, we decided to not use the results of ascending dataset for the interpretation steps and to draw conclusions.

117

Chapter 5

General conclusion

This Ph.D. thesis employs the Satellite Radar Interferometry (InSAR) techniques for studying the ongoing activity of both mud volcanoes and the outermost structures of fold-and-thrust belts. Mud volcanoes indeed usually develop at convergent plate margins, and are linked to thrust- related anticlines. The two study areas are orogenic fronts where the mud volcanism occurs, namely the Northern Apennines margin (Emilia-Romagna region; Northern Italy), and the southern Great Caucasus margin (Azerbaijan). Mud volcanism is a process typically linked to hydrocarbon traps and leads to the extrusion of subsurface mud, fragments of country rocks, saline waters and gases, building up a variety of conical edifices. The Pede-Apennine mud volcanoes are represented by mud pools, gryphons and mud cones, and can be up to 3-4 m tall, with few tens of meters of diameter. The Azerbaijan mud volcanoes are imposing edifices, up to 400 m high, with diameter up to 5 km.

5.1 DInSAR analysis applied to mud volcanoes of Azerbaijan

The Differential Synthetic Aperture Radar Interferometry (DInSAR) technique has been proved to be effective for detecting significant superficial deformation related to the activity of the large mud volcanoes of Azerbaijan, while it has not been suitable to study the small Pede-Apennine mud volcanoes. The first part of this thesis concerned the analysis of the ground deformation of

118 the Eastern Great Caucasus mud volcanoes using Envisat images that cover the period from October 2003 to November 2005. The results allowed us to investigate the activity of five mud volcanoes, and we obtained very interesting outcomes that may contribute to a deeper understanding of the processes driving ground deformation at mud volcanoes. Important deformation events, caused by fluid pressure and volume variations, have been observed both (1) in connection with main eruptive events in the form of pre-eruptive uplift (which reach a cumulative value of 20 and 10 cm at the Ayaz-Akhtarma and Khara-Zira mud volcanoes, respectively), and (2) in the form of short-lived deformation pulses that interrupt a period of quiescence. Both deformation patterns show important similarities with those identified in some magmatic systems: the pre-eruptive uplift has been observed in many magmatic volcanoes as a consequence of magma intrusion or hydrothermal fluid injection, and discrete short-duration pulses of deformation are also experienced by magmatic volcanoes and are repeated over time as multiple inflation and deflation events. The similarities in the time-space evolution of ground deformation, evidenced for the studied mud volcanoes of Azerbaijan, thus support that similar processes, driven by fluid pressure and volume variations, may govern both igneous and mud volcano systems.

5.2 PSI analysis of active fold-and-thrust belts: the Po Plain case study

The Persistent Scatterers Interferometry (PSI) technique has been applied to obtain some insights into the ongoing tectonic activity of compressive structures controlling mud volcanism. PSI is usually used for detecting surface deformation of wide regions involved in low tectonic movements in various geodynamic environments, including orogenic fronts. The second part of this thesis thus focused on the study of the activity of the outermost structures of the Northern Apennines, where mud volcanism may also occur. In the densely urbanized Po Plain area it was possible to find a considerable amount of PSs, leading to very interesting results. The analysis of the PSI linear deformation velocity map and of the PSI time series have allowed to observe ground deformation over a vast sector of the Po Plain (between Piacenza and Reggio Emilia) in the time span from 2004 to 2010. The time series of the GPS stations that occur in the study area have been compared with our results, showing a good correlation with the PS time series. The PSI analysis reveals the occurrence of a well-known subsidence area on the rear of the Ferrara arc, which is mostly connected to the exploitation of water resources. In some instances, the PS velocity pattern shows some ground uplift above active thrust-related anticlines (with mean

119 velocities ranging from 1 to 2.8 mm/yr) of the Emilia and Ferrara folds, and part of the Pede- Apennine margin. We hypothesize a correlation between the observed uplift deformation pattern and the growth of the thrust-related anticlines. As the uplift pattern corresponds to known active tectonic features, it can be used to constrain the seismo-tectonic setting. The lessons from the May 2012 seimic sequence indicate that some ground uplift preceeded the thrust-fault rupture (see section 4.6). A working hypothesis may involve that the active Emilia and Ferrara thrust folds would be characterized by inter-seismic periods basically dominated by aseismic creep.

5.3 General remarks

The satellite radar interferometry techniques have been proved to be effective for studying active geological processes and provide some unique capabilities for assessing geological rates of deformation and therefore for evaluation of geological hazards. It is worth noting that this research project have been successfully carried out using Envisat images free of charge, provided by the European Space Agency. The potential of these data has been fully exploited for studying two types of geological phenomena, i.e. the mud volcanism, and the activity of tectonic structures, using DInSAR and PSI techniques, according to the specific boundary conditions (i.e. the characteristics of the study areas, and the amount of available images). Unfortunately, it wasn’t possible to study the ongoing activity of both mud volcanoes and tectonic structures of thrust anticlines within a same study area, and in the same geographic context. This would have allowed us to investigate the possible correlation between active compressive structures and mud volcanism. This comparison might be the goal of future studies by means of interferometry techniques, which could be carried out exploiting other sensors with improved acquisition parameters (e.g. COSMO-SkyMed and TerraSAR-X satellites). Moreover, new technologies may allow to overcome the limitations of the employed methodologies and data, such as the low spatial resolution of Envisat, and the small amount of the available images for the Azerbaijan study area. In this regard, Sentinel-1 satellites, launched on 3th April 2014, will enhance the potential of radar satellite-based techniques, and it is expected to improve the SAR data availability, the spatial coverage, and the temporal sampling, thereby providing new high quality data. Several future developments could be realized hereinafter following the path traced by this study, applying satellite radar interferometry, both for the estimation of the active crustal deformation and of the earthquake preparation processes, and for the understanding of not fully known processes, such as mud volcanism.

120

121

Bibliography

Abbate, E., Bruni, P., 1987. Modino-Cervarola o Modino e Cervarola? Torbiditi oligo-mioceniche ed evoluzione del margine nord-appenninico. Mem. Soc. Geol. Ital. 39, 19–34.

Abidin, H.Z., Davies, R.J., Kusuma, M.A., Andreas, H., Deguchi, T., 2009. Subsidence and uplift of Sidoarjo (East Java) due to the eruption of the Lusi mud volcano (2006–present). Environ Geol, 57:833– 844, doi 10.1007/s00254-008-1363-4.

Abrams, M.A., Narimanov, A.A., 1997. Geochemical evaluation of hydrocarbons and their potential sources in the western South Caspian depression, Republic of Azerbaijan. Marine and Petroleum Geology 14, 451–468.

Acoglu, A.M., Cakir, Z., Meghraoui, M., Belabbes, S., El Alami, S.O., Ergintav, S., Akyüz, H.S., 2006. The 1994–2004 Al Hoceima (Morocco) earthquake sequence: conjugate fault ruptures deduced from InSAR. Earth and Planetary Science Letters 252, 467–480.

Adam, N., Kampes, B., Eineder, M., Worawattanamateekul, J., Kircher, M., 2003. The development of a scientific permanent scatterer system. Proc. of ISPRS Workshop High Resolution Mapping from Space, Hannover (Germany).

Adam, N., Rodriguez Gonzalez, F., Parizzi, A., Brcic, R., 2013. Wide area persistent scatterer interferometry: current developments, algorithms and examples. Proceedings of IGARSS 2013, Melbourne, 1857-1860.

Adamia, S., Lordkipanidze, M. B., Zakariadze, G. S., 1977. Evolution of an active continental margin as exemplified by the alpine history of the Caucasus. Tectonophysics, 40, 183–199.

Adamia, S., Alania, V., Chabukiani, A., Kutelia, Z., Sadradze, N., 2011. Greater Caucasus (Cavcasioni): A long-lived North-Tethyan back-arc basin,Turk. J. Earth Sci., 20, 611–628.

Agabekov, M.G., Kerimov, K.M., Moshashvili, A.B., Khain, V.Y., 1976. New data on the structure of the central part of the Kura trough: Geotectonics, v. 10, p. 350–354.

Albarello, D., Palo, M., Martinelli, G., 2012. Monitoring methane emission of mud volcanoes by seismic tremor measurements: a pilot study. Nat. Hazards Earth Syst. Sci., 12, 3617–3629.

Alberti, A., Alessandro, V., Pieruccini, U., Pranzini, E., 1993. Landsat TM data processing for lithological discrimination in the Caraculo area (Namibe Province, SW Angola). Journal of African Earth Sciences, 17, 3, 261-274.

Aliyev, A.A., 2008. Geological Map of Azerbaijan Republic, Scale 1:500,000, with Explanatory Notes. Baki Kartoqrafiya Fabriki. National Academy of Sciences of Azerbaijan Republic. Baku.

Aliyev, A.A., Guliyev, I.S., Rahmanov, R.R., 2009. Catalogue of recorded of mud volcano eruptions of Azerbaijan (1810–2007). Second Edition. ‘Nafta-Press’ Publishing-House, Baku, pp. 109.

122

Allen, D.R., Mayuga, M.N., 1969. The mechanics of compaction and rebound, Wilmington Oil Field, Long Beach, California, U.S.A. Proceedings of the Tokyo Symposium on Land Subsidence 2, 410–423, International Association of Scientific Hydrology and UNESCO, September 1969, Tokyo.

Allen, M.B., Jones, S., Ismail-Zadeh, A., Simmons, M.D., Anderson, L., 2002. Onset of subduction as the cause of rapid Pliocene – Quaternary subsidence in the South Caspian Basin. Geology 30, 775–778.

Allen, M.B., Vincent, S.J., Alsop, G.I., Ismail-Zadeh, A, Flecker R., 2003. Late Cenozoic deformation in the South Caspian region: effects of a rigid basement block within a collision zone. Tectonophysics, n. 366; 223–239. doi:10.1016/S0040-1951(03)00098-2.

Allen, M.B., Jackson, J., Walker, R., 2004. Late Cenozoic reorganization of the Arabia-Eurasia collision and the comparison of short-term and long-term deformation rates: Tectonics, v. 23, TC2008, doi: 10.1029/2003TC001530.

Amelung, F., Jónsson, S., Zebker, H., Segall, P., 2000. Widespread uplift and ‘trapdoor’ faulting on Galápagos volcanoes observed with radar interferometry. Nature, v. 407, p. 993–996, doi: 10.1038/35039604.

Amorosi, A., Farina, M., Severi, P., Preti, D., Caporale, L., Di Dio, G., 1996. Genetically related alluvial deposits across active fault zones: an example of alluvial fan-terrace correlation from the upper Quaternary of the southern Po Basin, Italy. Sediment. Geol. 102, 275–295, doi:10.1016/0037- 0738(95).00074-7.

Aoki, Y., Sidiq, T.P., 2014. Ground deformation associated with the eruption of Lumpur Sidoarjo mud volcano, east Java, Indonesia. Journal of Volcanology and Geothermal Research 278–279, 96–102.

Arca, S., Beretta, G.P., 1985. Prima sintesi geodetico-geologica sui movimenti verticali del suolo nell’Italia Settentrionale (1897–1957). Boll. Geod. Sc. Aff. 44, 125–156.

Argnani A., Barbacini G., Bernini M., Camurri F., Ghielmi M., Papani G., Rizzini F., Rogledi S., Torelli L., 2003. Gravity tectonics driven by Quaternary uplift in the Northern Apennines: insights from the La Spezia-Reggio Emilia geo-transect. Quaternary International, 101– 102, 13–26.

Atzori, S., Manunta, M., Fornaro, G., Ganas, A., Salvi, S., 2008. Postseismic displacement of the 1999 Athens earthquake retrieved by the differential interferometry by synthetic aperture radar time series. J. Geophys. Res., 113, B09309, doi:10.1029/2007JB005504.

Avdeev, B., Niemi, N. A., 2011. Rapid Pliocene exhumation of the central Greater Caucasus constrained by low-temperature thermochronometry, Tectonics, 30, TC2009, doi:10.1029/2010TC002808.

Axen, G.J., Lam, P.S., Grove, M., Stockli, D.F., Hassanzadeh, J., 2001. Exhumation of the west-central Alborz Mountains, Iran, Caspian subsidence, and collision-related tectonics, Geology, 29(6), 559–562.

Baarda, W., 1968. A testing procedure for use in geodetic networks. Rijkscommissie voor Geodesie, Delft, The Netherlands.

123

Baldi, P., Casula, G., Cenni, N., Loddo, F., Pesci, A., 2009. GPS-based monitoring of land subsidence in the Po Plain (Northern Italy). Earth Plan. Sci. Lett., 288, 204-212, doi: 10.1016/j.epsl.2009.09.023.

Baldi., P., Casula, G., Cenni, N., Loddo, F., Pesci, A., Bacchetti, M., 2011. Vertical and horizontal crustal movements in central and northern Italy. Bollettino di Geofisica Teorica ed Applicata, Vol. 52, n. 4, pp. 667-685; doi: 10.4430/bgta0021.

Bamler, R., Hartl, P., 1998. Synthetic aperture radar interferometry. Inverse problems, 14 (4), R1-54.

Barbacini, G., Bernini, M., Papani, G., Rogledi, S., 2002. Le strutture embricate del margine appenninico emiliano tra il T. Enza ed il F. Secchia—Prov. di Reggio Emilia (con carta strutturale alla scala 1:50.000). In: Atti del III seminario sulla cartografia geologica. Bologna 26–27 febbraio 2002. Serv. Geologico, Sismico e dei Suoli, Reg. Emilia-Romagna—CNR, Istituto di Geoscienze e Georisorse, pp 64–69.

Bartolini, C., Caputo, R., Pieri, M., 1996. Pliocene–Quaternary sedimentation in the Northern Apennine Foredeep and related denudation. Geological Magazine, 133(3), 255−273.

Basili, R., Valensise, G., Vannoli, P., Burrato, P., Fracassi, U., Mariano, S., Tiberti, M.M., Boschi, E., 2008. The Database of Individual Seismogenic Sources (DISS), version 3: summarizing 20 years of research on Italy’s earthquake geology, Tectonophysics 453, 20–43, doi:10.1016/j.tecto.2007.04.014.

Bateson, L., Novali, F., Cooksley, G., 2010. Terrafirma user guide: a guide to the use and understanding of Persistent Scatterer Interferometry in the detection and monitoring of terrain-motion. Terrafirma project, ESRIN/Contract no. 19366/05/I-E, available on-line at www.terrafirma.eu.com/users.htm.

Béjar-Pizarro, M., Socquet, A., Armijo, R., Carrizo, D., Genrich, J., Simons, M., 2013. Andean structural control on interseismic coupling in the North Chile subduction zone. Nature Geoscience Letters, Vol 6 , 462-467, doi: 10.1038/NGEO1802.

Bell, J.W., Amelung, F., Ferretti, A., Bianchi, M., Novali F., 2008. Permanent scatterer InSAR reveals seasonal and long-term aquifer system response to groundwater pumping and artificial recharge. Water Resour. Res., 44, W02407, doi:10.1029/2007WR006152.

Bell, A.F., Naylor, M., Heap, M.J., Main, I.G., 2011. Forecasting volcanic eruptions and other material failure phenomena: An evaluation of the failure forecast method. Geophysical Research Letters, Vol. 38, L15304, Doi:10.1029/2011gl048155.

Benedetti, L., Tapponnier, P., Gaudemer, Y., Manighetti, I., Van der Woerd, J., 2003. Geomorphic evidence for an emergent active thrust along the edge of the Po Plain: the Broni-Stradella fault. J Geophys Res 108, n. 2238, doi:10.1029/2001JB001546.

Berardino, P., Fornaro, G., Lanari, R., Sansosti, E., 2002. A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms. IEEE Transactions on Geoscience and Remote Sensing, vol. 40, no. 11, pp. 2375–2383.

Berardino, P., Costantini, M., Franceschetti, G., Iodice, A., Pietranera, L., Rizzo, V., 2003. Use of differential SAR interferometry in monitoring and modelling large slope instability at Maratea (Basilicata, Italy). Engineering Geology, 68 (1-2), 31 - 51.

124

Berberian, M., 1983. The southern Caspian: A compressional depression floored by a trapped, modified oceanic crust. Canadian Journal of Earth Sciences, v. 20, p. 163–183.

Biescas, E., Crosetto, M., Agudo, M., Monserrat, O., Crippa, B., 2007. Two radar interferometric approaches to monitor slow and fast land deformations. Journal of Surveying Engineering, 133, 66-71.

Biggs, J., Wright, T., Lu Z., Parsons, B., 2007. Multi-interferogram method for measuring interseismic deformation: Denali Fault, Alaska. Geophys. J. Int. 170, 1165–1179, doi: 10.1111/j.1365- 246X.2007.03415.x

Biggs, J., Anthony E., Ebinger C., 2009. Multiple inflation and deflation events at Kenyan volcanoes, East African Rift, Geology, 37, 979–982, doi:10.1130/G30133A.1.

Biggs, J., Bastow, I. D., Keir, D., Lewi, E., 2011. Pulses of deformation reveal frequently recurring shallow magmatic activity beneath the Main Ethiopian Rift. Geochemistry Geophysics Geosystems, v.12, n. 9, Q0AB10, doi:10.1029/2011GC003662, ISSN: 1525‐2027.

Bigi, G., Bonardi, G., Catalano, R., Cosentino, D., Lentini, F., Parotto, M., Sartori, R., Scandone, P., Turco, E., 1983. Structural Model of Italy 1:500,000, CNR Progetto Finalizzato Geodinamica.

Bignami, C., Burrato, P., Cannelli, V., Chini, M., Falcucci, E., Ferretti, A., Gori, S., Kyriakopoulos, C., Melini, D., Moro, M., Novali, F., Saroli, M., Stramondo S., Valensise, G., Vannoli, P., 2012. Coseismic deformation pattern of the Emilia 2012 seismic sequence imaged by Radarsat-1 interferometry. Annals of Geophysics, 55, 4, 2012; doi: 10.4401/ag-6157.

Bissoli, R., Bitelli, G., Bonsignore, F., Rapino, A., Vittuari, L., 2010. Land subsidence in Emilia- Romagna Region, northern Italy: recent results”, Land Subsidence, Associated Hazards and the Role of Natural Resources Development, IAHS Publ. 339, 307-311.

Bitelli, G., Bonsignore, F., Del Conte, S., Novali, F., Pellegrino, I., Vittuari, L., 2014. Integrated use of Advanced InSAR and GPS data for subsidence monitoring. Proceedings of IAEG XII Congress.

Björck, Å., 1996. Numerical methods for least square problems. Siam: Philadelphia, PA, USA.

Boccaletti, M., Elter P., Guazzone G., 1971. Plate tectonic models for the development of the Western Alps and Northern Apennines. Nature, Phys. Sc., 234, 108-111.

Boccaletti, M., Coli, M., Eva, C., Ferrari, G., Giglia, G., Lazzarotto, A., Merlanti, F., Nicolich, R., Papani, G., Postpischl, D., 1985. Considerations on the seismotectonics of the Northern Apennines, Tectonophysics 117, 7–38.

Boccaletti M., Calamita F., Deiana G., Gelati R., Massari F., Moratti G., Ricci Lucchi F., 1990. Migrating foredeep-thrust belt system in the Northern Apennines and Southern Alps. Palaeo., Palaeo., Palaeo., 77, 3-14.

125

Boccaletti, M., Bonini, M., Corti, G., Gasperini, G., Martelli, L., Piccardi, L., Tanini, C., Vannucci, G., Molinari, F., Moratti, G., Rosselli, S., Severi, P., 2004. Carta sismotettonica della Regione Emilia- Romagna, scala 1:250.000. S.El.Ca., Firenze, 2004.

Boccaletti, M., Corti, G., Martelli, L., 2011. Recent and active tectonics of the external zone of the Northern Apennines (Italy). Int. J. Earth Sci. 100, 1331–1348, doi:10.1007/s00531-010-0545-y.

Bock, Y., Wdowinski, S., Ferretti, A., Novali, F., Fumagalli, A., 2012. Recent subsidence of the Venice Lagoon from continuous GPS and interferometric synthetic aperture radar. Geochemistry Geophysics Geosystems Research Letter, Vol. 13, n. 3, doi:10.1029/2011GC003976.

Bonini, M., 2007. Interrelations of mud volcanism, fluid venting, and thrust-anticline folding: examples from the external northern Apennines (Emilia-Romagna, Italy). J Geophys Res 112:B08413. doi:10.1029/2006JB004859.

Bonini, M., 2009. Mud volcano eruptions and earthquakes in the Northern Apennines and Sicily, Italy, Tectonophysics, 474, 723–735.

Bonini, M., 2012. Mud volcanoes: indicators of stress orientation and tectonic controls. Earth Sci. Rev. 115, 121–152.

Bonini, M., Tassi, F., Feyzullayev, A.A., Aliyev, C.S., Capecchiacci, F., Minissale, A., 2013. Deep gases discharged from mud volcanoes of Azerbaijan: New geochemical evidence. Marine and Petroleum Geology 43, 450-463.

Bonini, M., 2013. Fluid seepage variability across the external Northern Apennines (Italy): Structural controls with seismotectonic and geodynamic implications. Tectonophysics 590, 151–174.

Bovenga, F., Nutricato, R., Refice, A., Wasowski, J., 2006. Application of multi-temporal differential interferometry to slope instability detection in urban/peri-urban areas. Engineering Geology 88, 218-239.

Brenner, A., Roessing, L., 2008. Radar imaging of urban areas by means of very high resolution SAR and interferometric SAR. IEEE Transactions on Geoscience and Remote Sensing, 46, pp. 2971–2982.

Brown, K.M., 1990. The nature and hydrological significance of mud diapirism and diatremes for accretionary systems. J. Geophys. Res. 95, 8969-8982.

Bruno, P.P.G., Pazzaglia, F.J., Picotti, V., 2011. Evidence for active folding and faulting at the northern Apennines mountain front near Bologna, Italy from high resolution seismic reflection profiling. Geophys. Res. Lett., 38, L16302, doi:10.1029/2011GL047828.

Brunori, C.A., Bignami, C., Stramondo, S., Bustos, E., 2013. 20 years of active deformation on volcano caldera: Joint analysis of InSAR and AInSAR techniques. International Journal of Applied Earth Observation and Geoinformation 23, 279–287.

Bürgmann R., Rosen P.A., Fielding E.J., 2000. Synthetic aperture radar interferometry to measure Earth’s surface topography and its deformation. Ann. Rev. Earth Planet. Sci., 28, 169-209.

126

Bürgmann, R., Hilley, G., Ferretti, A., Novali, F., 2006. Resolving vertical tectonics in the San Francisco Bay Area from permanent scatterer InSAR and GPS analysis. Geology 34, 221–224.

Burrato, P., Ciucci, F., Valensise, G., 2003. An inventory of river anomalies in the Po Plain, Northern Italy: evidence for active blind thrust faulting, Ann. Geophys. 5, 865–882, doi:10.4401/ag-3459.

Burrato, P., Vannoli, P., Fracassi, U., Basili, R., Valensise, G., 2012. Is blind faulting truly invisible? Tectonic-controlled drainage evolution in the epicentral area of the May 2012, Emilia-Romagna earthquake sequence (Northern Italy). Ann. Geophys. 55, 4, 525–531, doi:10.4401/Ag6182.

Calda, N., Valloni, R., Bedulli, F., 2007. Three-dimensional representation of permeability barriers and aquifers recharge in the Pleistocene deposits of the Parma alluvial plain. Mem. Descr. Carta Geol. d’It. LXXVI (2007), pp. 97-108.

Calderoni, G., Di Giovambattista, R., Burrato, P., Ventura, G., 2009. A seismic sequence from Northern Apennines (Italy) provides new insight on the role of fluids in the active tectonics of accretionary wedges. Earth and Planetary Science Letters 281, 99–109.

Capozzi, R., Picotti, V., 2002. Fluid migration and origin of a mud volcano in the northern Apennines (Italy): The role of deeply rooted normal faults. Terra Nova, 14, 363–370.

Capozzi R., Picotti, V., 2010. Spontaneous fluid emissions in the Northern Apennines: geochemistry, structures and implications for the petroleum system; Geological Society, London, Special Publications, v. 348; p. 115-135. doi:10.1144/SP348.7.

Caricchi, L., Biggs, J., Annen, C., Ebmeier, S., 2014. The influence of cooling, crystallization and re- melting on the interpretation of geodetic signals in volcanic systems. Earth and Planetary Science Letters 388, 166–174.

Carminati, E., Martinelli, G., 2002. Subsidence rates in the Po Plain, northern Italy: the relative impact of natural and anthropogenic causation. Engineering Geology, 66, 241−255.

Carminati, E., Doglioni, C., 2012. Alps vs. Apennines: The paradigm of a tectonically asymmetric Earth. Earth-Science Reviews 112, 67–96.

Carnec, C., Massonnet, D., King, C., 1996. Two examples of the application of SAR interferometry to sites of small extent. Geophysical Research Letters, 23, 3579 - 3582.

Caro Cuenca, M., Hanssen, R., Hooper, A., Arıkan, M., 2011. Surface deformation of the whole Netherlands after PSI analysis. Proceedings of FRINGE 2011, Frascati (Italy).

Cassano, E., Anelli, L., Fichera, R., Cappelli, V., 1986. Pianura Padana: Interpretazione integrata di dati geofisici e geologici. AGIP. 73° Congresso Società Geologica Italiana, Roma, 27 pp.

Castaldini, D., Valdati, J., Ilies, D.C., Chiriac, C., Bertogna, I., 2005. Geo-Tourist map of the natural Reserve of Salse di Nirano (Modena Apennines, Northern Italy). Il Quaternario 18, 245–255.

127

Castellarin, A., Eva, C., Giglia, G., Vai, G.B., 1985. Analisi strutturale del fronte appeninico-padano. Giorn Geol Ser 3 47 (1–2):47–76.

Castello, B., Selvaggi, G., Chiarabba, C., Amato, A., 2006. CSI Catalogo della sismicità italiana 1981– 2002, versione 1.1. INGV-CNT, Roma, available at: http://legacy.ingv.it/CSI/.

Castiglioni, G.B., Bondesan, A., Bondesan, M., Cavallin., A., Gasperi, G., Persisco, A., 1997. Geomorphological map of the Po Plain, Map of relief and vertical movements of Po Plain (1 : 250.000 scale). S.EL.CA., Firenze.

Casu, F., Manzo, M., Lanari, R., 2006. A quantitative assessment of the SBAS algorithm performance for surface deformation retrieval from DInSAR data. Remote Sensing of Environment, 102(3), 195-210.

Catani, F., Farina, P., Moretti, S., Nico, G., Strozzi, T., 2005. On the application of SAR interferometry to geomorphological studies: Estimation of landform attributes and mass movements. Geomorphology, 66, 119-131.

Cenni, N., Mantovani, E., Baldi, P., Viti, M., 2012. Present kinematics of Centraland Northern Italy from continuous GPS measurements. J. Geodyn. 58, 62–72, http://dx.doi.org/10.1016/j.jog.2012.02.004.

Cenni, N., Viti, M., Baldi, P., Mantovani, E., Bacchetti, M., Vannucchi, A., 2013. Present vertical movements in Central and Northern Italy from GPS data: Possible role of natural and anthropogenic causes. Journal of Geodynamics 7, 74–85. http://dx.doi.org/10.1016/j.jog.2013.07.004.

Cenni, N., Loddo, F., Zucca, F., Meisina, C., Baldi, P., Belardinelli, M.E., Bacchetti, M., Mantovani, E., Viti, M., Casula, G., Stramondo, S., Severi, P., Pierini, L., Bonsignore, F., 2014. The spatio-temporal pattern of subsidence in the Po basin monitored by different techniques. Atti del 33° Convegno Nazionale GNGTS. 001-260 volume 1, pp.134-139.

Cerrina Feroni, A., Martelli, L., Martinelli, P., Ottria, G., 2002. Carta geologico-strutturale dell’Appennino emiliano-romagnolo in scala 1:250.000. Regione Emilia-Romagna—C.N.R., Pisa. S.EL.CA, Firenze.

Champenois, J., Fruneau, B., Pathier, E., Deffontaines, B., Lin, K.C., Hu, J.C., 2012. Monitoring of active tectonic deformations in the Longitudinal Valley (eastern Taiwan) using Persistent Scatterer InSAR method with ALOS PALSAR data. Earth Planet. Sci. Lett., 337, 144-155, http://dx.doi.org/10.1016/j.epsl.2012.05.025.

Chang, W., Smith, R.B., Wicks, C., Farrell, J.M., Puskas, C.M., 2007. Accelerated uplift and magmatic intrusion of the Yellowstone caldera, 2004 to 2006. Science, v. 318, p. 952–956, doi: 10.1126/science 1146842.

Chen, C.T.,. Hu, J.T,. Lu, C.Y,. Lee, J.C,. Chan, Y.C., 2007. Thirty-year land elevation change from subsidence to uplift following the termination of groundwater pumping and its geological implications in the Metropolitan Taipei Basin, Northern Taiwan, Eng. Geol., 95, 30–47.

Chen, C.W., Zebker H.A., 2000. Network approaches to two-dimensional phase unwrapping: intractability and two new algorithms. March Journal of the Optical Society of America, A 17, 401–414.

128

Chen, Q., Liu, G., Ding, X., Hu, J. C., Yuan, L., Zhong, P., Omura, M., 2010. Tight integration of GPS observations and persistent scatterer InSAR for detecting vertical ground motion in Hong Kong. International J. of Applied Earth Observation and Geoinformation, 12(6), 477-486.

Chini, M., Pierdicca, N., Emery, W.J., 2009. Exploiting SAR and VHR optical images to quantify damage caused by the 2003 Bam Earthquake. IEEE Transactions on Geoscience and Remote Sensing, 47, pp. 145–152.

Ciaccio, M.G., Chiarabba, C., 2002. Tomographic models and seismotectonics of the Reggio Emilia region, Italy. Tectonophysics 344, 261– 276.

Colesanti, C., Ferretti, A., Prati, C., Rocca, F., 2003. Monitoring landslides and tectonic motions with the Permanent Scatterers Technique. Engineering Geology, 68, 3-14.

Colesanti, C., Mouelic, S.L., Bennani, M., Raucoules, D., Carnec, C., Ferretti, A., 2005. Detection of mining related ground instabilities using the Permanent Scatterers technique—a case study in the east of France. International J. of Remote Sensing, 26 (1), 201-207.

Colesanti, C. and Wasowski, J., 2006. Investigating landslides with satellite Synthetic Aperture Radar (SAR) interferometry. Engineering Geology, 88 (3 - 4), 173 - 199.

Cooper, C., 2001. Mud volcanoes of Azerbaijan visualized using 3D seismic depth cubes: the importance of overpressured fluid and gas instead of non-existent diapirs (abs.). Proceedings of the European Association of Geoscientists and Engineers, Conference on ‘Subsurface Sediment Mobilisation’, Gent, Belgium, September 2001, p. 71.

Costantini, M. 1998. A novel phase unwrapping method based on network programming. IEEE Transactions on Geoscience and Remote Sensing, 36 (3), 813–821.

Costantini, M., Farina, A., Zirilli, F., 1999. A fast phase unwrapping algorithm for SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 37(1), 452– 460.

Costantini, M., Iodice, A., Magnapane, L., Pietranera, L., 2000. Monitoring terrain movements by means of sparse SAR differential interferometric measurements. In: Proceedings of IGARSS 2000, 20th IEEE international Geoscience and Remote sensing symposium, Honolulu, Hawaii, USA, 24–28 July 2000, pp 3225–3227

Cremonini, G., Ricci Lucchi, F., 1982. Guida alla geologia del margine appenninico-padano, Guide Geologiche Regionali, Società Geologica Italiana.

Crosetto, M., Crippa, B., Biescas, E., 2005. Early detection and in-depth analysis of deformation phenomena by radar interferometry. Engineering Geology, 79 (1), 81–91.

Crosetto, M., Biescas, E., Duro, J., Closa, J., Arnaud, A., 2008. Generation of advanced ERS and Envisat interferometric SAR products using the stable point network technique. Photogrammetric Engineering and Remote Sensing, 74 (4), 443–450.

129

Crosetto, M., Monserrat, O., Cuevas, M., Crippa, B., 2011. Spaceborne differential SAR interferometry: Data analysis tools for deformation measurement. Remote Sensing, 3 (2), 305-318.

Cuffaro, M., Riguzzi, F., Scrocca, D., Antonioli, F., Carminati, E., Divani, M., Doglioni, C., 2010. On the geodynamics of the Northern Adriatic Plate. Rend. Fis. Acc. Lincei 21, 1, S253– S279, Doi:10.1007/S12210-010-0098-9.

D’Agostino, N., Avallone, A., Cheloni, D., D’anastasio, E., Mantenuto, S., Selvaggi, G., 2008. Active tectonics of the Adriatic region from GPS and earthquake slip vectors. J. Geophys. Res. 113, B12413, doi:10.1029/2008JB005860.

Davies, R.J., Swarbrick, R.E., Evans, R.J., Huuse, M., 2007. Birth of a mud volcano: East Java, 29 May 2006. Geological Society of America Today 17, 4–9.

Dehghani, M., Zoej, M. J.V., Hooper, A., Hanssen, R.F., Entezam, I., Saatchi, S., 2013. Hybrid conventional and Persistent Scatterer SAR interferometry for land subsidence monitoring in the Tehran Basin, Iran. ISPRS J., 79, 157-170.

Del Ventisette, C., Righini, G., Moretti, S., Casagli, N., 2014. Multitemporal landslides inventory map updating using spaceborne SAR analysis. International Journal of Applied Earth Observation and Geoinformation, 30 (1), pp. 238-246.

Devanthéry, N., 2014. High-resolution deformation measurement using Persistent Scatterer Interferometry. Ph.D. thesis.

Devanthéry, N., Crosetto, M., Monserrat, O., Cuevas-González, M., Crippa, B., 2014. An approach to Persistent Scatterer Interferometry. Remote Sensing, 6 (7), 6662-6679.

Devlin, W., Cogswell, J., Gaskins, G., Isaksen, G., Pitcher, D., Puls, D., Stanley, K., Wall, G., 1999. South Caspian Basin: young, cool, and full of promise. Geological Society of America Today 9 (7), 1–9.

Devoti, R., Esposito, A., Pietrantonio, G., Pisani, A.R., Riguzzi, F., 2011. Evidence of large scale deformation patterns from GPS data in the Italian subduction boundary. Earth Planet. Sc. Lett. 311, 3-4, 230-241, doi:10.1016/j.epsl.2011.09.034.

Dimitrov, L.I., 2002. Mud volcanoes - the most important pathway for degassing deeply buried sediments. Earth-Science Reviews 59, 49-76.

DISS Working Group, 2010. Database of individual seismogenic sources (DISS), version 3.1.1: a compilation of potential sources for earthquakes larger than M 5.5 in Italy and surrounding areas. http://diss.rm.ingv.it/diss/, INGV 2010 - Istituto Nazionale di Geofisica e Vulcanologia - doi:10.6092/INGV.IT-DISS3.1.1.

Dong, Y., Forster, B., Ticehurst, C., 1997. Radar backscatter analysis for urban environments. International Journal of Remote Sensing, 18, pp. 1351-1364.

Drury, S.A., 1987. Image Interpretation in Geology. Chapman and Hall, London, p. 283.

130

Ebmeier, S.K., Biggs, J., Mather, T.A, Wadge, G., Amelung, F., 2010. Steady downslope movement on the western flank of Arenal volcano, Costa Rica. Geochemistry Geophysics Geosystems, Volume 11, Number 12, Q12004, doi: 10.1029/2010GC003263 ISSN: 1525‐2027.

Ebmeier, S. K., Biggs, J., Mather, T. A., Elliott, J. R., Wadge, G., Amelung, F., 2012. Measuring large topographic change with InSAR: Lava thicknesses, extrusion rate and subsidence rate at Santiaguito volcano, Guatemala. Earth and Planetary Science Letters 335–336, 216–225.

Ershov, A.V., Brunet, M.F., Nikishin, A.M., Bolotov, S.N., Nazarevich, B.P., Korotaev, M.V., 2003. Northern Caucasus basin: Thermal history and synthesis of subsidence models. Sediment. Geol., 156, 95– 118.

Evans, R.J., Davies, R.J., Stewart, S. A., 2006. Internal structure and eruptive history of a kilometer-scale mud volcano system, South Caspian Sea. Basin Research, 19, 153–163.

Fantoni, R, Franciosi, R., 2010. Tectono-sedimentary setting of the Po Plain and Adriatic foreland. Rend. Lincei 21 (1), S197–S209. http://dx.doi.org/10.1007/s12210-010-0102-4.

Feigl, K., Sergent, A., Jacq, D., 1995. Estimation of an earthquake focal mechanism from a satellite radar interferogram: application to the December 4, 1992 Landers aftershock. Geophysical Research Letters 22, 1037–1040.

Ferretti, A., Prati, C., Rocca, F., 1999. Permanent scatterers in SAR interferometry. Proc. of Remote Sensing, International Society for Optics and Photonics, Florence (Italy).

Ferretti, A., Prati, C., Rocca, F., 2000. Nonlinear subsidence rate estimation using permanent scatterers in differential SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 38 (5), 2202- 2212.

Ferretti, A., Prati, C., Rocca, F., 2001. Permanent scatterers in SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 39(1), 8-20.

Ferretti, A., Savio, G., Barzaghi, R., Borghi, A., Musazzi, S., Novali, F., Prati, C., Rocca F., 2007. Submillimeter Accuracy of InSAR Time Series: Experimental Validation. IEEE Transactions on Geoscience and Remote Sensing, Volume 45, Issue 5, 1142-1153.

Ferretti, A., Fumagalli, A., Novali, F., Prati, C., Rocca, F., Rucci, A., 2011. A new algorithm for processing interferometric data-stacks: Squee SAR. IEEE Transactions on Geoscience and Remote Sensing, 49(9), 3460-3470.

Filikov, S. V., Gomez, F., Al-Ghazzi, R., and Karam, G., 2006. GPS constraints on continental deformation in the Africa-Arabia-Eurasia continental collision zone and implications for the dynamics of plate interactions. Journal of Geophysical Research, v. 111, B05411, doi: 10.1029/2005JB004051.

Förstner, W., 1986. Reliability, gross error detection and self-calibration. ISPRS Commission III Tutorial on Statistical Concepts for Quality Control, International Archives of Photogrammetry and Remote Sensing, 26(3/4), 1-34.

131

Forte, A.M., Cowgill, E., Bernardin, T., Kreylos, O., Hamann B., 2010. Late Cenozoic deformation of the Kura fold-thrust belt, southern Greater Caucasus,Geol. Soc. Am. Bull., 122(3/4), 465–486.

Forte, A.M., Cowgill, E., Murtuzayev, I., Kangarli, T., Stoica, M., 2013. Structural geometries and magnitude of shortening in the eastern Kura fold-thrust belt, Azerbaijan: Implications for the development of the Greater Caucasus Mountains,Tectonics, 32, 688–717,doi:10.1002/tect.20032.

Forte, A.M., Cowgill. E., Whipple, K.X., 2014. Transition from a singly vergent to doubly vergent wedge in a young orogen: The Greater Caucasus. Tectonics, 33, 2077–2101, doi:10.1002/2014TC003651.

Fowler, S.R., Mildenhall, J., Zalova, S., Riley, G., Elsley, G., Desplanques, A., Guliyev, F., 2000. Mud volcanoes and structural development on Shah Deniz. Journal of Petroleum, Science and Engineering 28, 189–206.

Franke, R., 1982. Scattered data interpolation: test of some methods. Mathematics of Computations, 33, 181−200.

Fryer, P., Wheat, C.G., Mottl, M.J., 1999. Mariana Blueschist mud volcanism: implications for conditions within the subduction zone. Geology 27, 103–106.

Fukushima, Y., Mori, J., Hashimoto, M., Kano, Y., 2009. Subsidence associated with the LUSI mud eruption, East Java, investigated by SAR interferometry. Marine and Petroleum Geology26, 1740–1750.

Funning, G.J., Parsons, B., Wright, T.J., Jackson, J.A., Fielding E.J., 2005. Surface displacements and source parameters of the 2003 Bam (Iran) earthquake from Envisat advanced synthetic aperture radar imagery. J. Geophys. Res., 110, B09406, doi:10.1029/2004JB003338.

Gabriel, A.K., Goldstein, R.M., Zebker, H.A., 1989. Mapping small elevation changes over large areas: differential radar interferometry. Journal of Geophysical Research, 94(B7), 9183-9191.

Gambolati, G., Teatini, P., 2015. Geomechanics of subsurface water withdrawal and injection, Water Resour. Res., 51, 3922–3955, doi:10.1002/2014WR016841.

Gamkrelidze, I.P., 1986. Geodynamic evolution of the Caucasus and adjacent areas in alpine time. Tectonophysics, 127, 261–277.

Gatto, P., Carbognin, L., 1981. The Lagoon of Venice: natural environmental trend and man-induced modification. Hydrological Sciences-Bulletin 26 (4), 379–391.

Gazis, C.A., Lanphere, M., Taylor, H.P., Gurbanov A., 1995. Ar 40/Ar39 and O18/O16 studies of the Chegem ash-flow caldera and the Eldjurta granite: Cooling of two late Pliocene igneous bodies in the Greater Caucasus mountains, Russia, Earth Planet. Sci. Lett., 134, 377 – 391.

Ghielmi, M., Minervini, M., Nini, C., Rogledi, S., Rossi, M., 2013. Late Miocene–Middle Pleistocene sequences in the Po Plain—Northern Adriatic Sea (Italy): the stratigraphic record of modification phases affecting a complex foreland basin, Mar. Petr. Geol. 42, 50–81, doi:10.1016/j.marpetgeo.2012.11.007.

132

Ghiselli, F., Martelli, L., 1997. Evoluzione dei campi di stress lungo il margine appenninico-padano dal Pleistocene all’Attuale: analisi strutturale dei depositi pleistocenici affioranti tra Castell’Arquato (Piacenza) e Rimini. Il Quaternario, 10 (2), 437-444.

Goldstein, R.M., Zebker, H.A., Werner, C., 1988. Satellite radar interferometry: Two‐dimensional phase unwrapping. Radio Science, 23(4), 713-720.

Golonka, J., 2007. Geodynamic evolution of the South Caspian Basin. In Oil and Gas of the Greater Caspian Area. Edited by P. O. Yilmaz and G. H. Isaken, pp. 17–41, Am. Assoc. of Pet. Geol., Tulsa, Okla.

Grandin, R., Doin, M.P., Bollinger, L., Pinel-Puysségur, B., Ducret, G., Jolivet, R., Sapkota, S.N., 2012. Long-term growth of the Himalaya inferred from interseismic InSAR measurement. Geology, v. 40; no. 12; p. 1059–1062; doi:10.1130/G33154.1.

Graue, K., 2000. Mud volcanoes in deepwater, Nigeria. Marine and Petroleum Geology 17, 959–974.

Gualtierotti, A., 1998. Analisi strutturale di una porzione del margine pedeappenninico padano tra il Fiume Panaro e il torrente Idice. Tesi di dottorato. Consorzio Università Ferrara, Firenze e Parma.

Guglielmino, F., Anzidei, M., Briole, P., Elias, P., Puglisi, G., 2013. 3D displacement maps of the 2009 L’Aquila earthquake (Italy) by applying the SISTEM method to GPS and DInSAR data. Terra Nova, 25, 79–85, doi: 10.1111/ter.12008.

Guidoboni, E., Ferrari, G., Mariotti, D., Comastri, A., Tarabusi, G., Valensise, G., 2007. CFTI4Med, Catalogue of Strong Earthquakes in Italy (461 B.C.-1997) and Mediterranean Area (760 B.C.-1500). http://storing.ingv.it/cfti4med/.

Guliyev, I.I., Feizullayev, A.A., 1997. All about mud volcanoes, Bafta Press, Baku, Azerbaijan.

Gunderson, K.L., Pazzaglia, F.J., Picotti, V., Anastasio, D. A., Kodama, K. P., Rittenour, T., Frankel, K. F., Ponza A., Berti C., Negri A., Sabbatini A., 2014. Unraveling tectonic and climatic controls on synorogenic growth strata (Northern Appennines, Italy). GSA Bulletin, 126- 3/4, 532-552.

Gupta, R.P., 2003. Remote Sensing Geology, Springer, Berlin, p. 629.

Hanssen, R.F., Caro Cuenca, M., Klees, R., van der Marel, H., 2012. Decadal vertical deformation of the Netherlands via the geodetic integration of gravimetry, GNSS, leveling and SAR interferometry. In AGU Fall Meeting Abstracts.

Hanssen, R.F., 1998. Atmospheric heterogeneities in ERS tandem SAR interferometry. The Netherlands: Delft University Press.

Hanssen, R.F., 2001. Radar interferometry: data interpretation and error analysis. Kluwer Academic Publishers, Dordrecht, The Netherlands.

133

Herrera, G., Notti, D., García-Davalillo, J.C., Mora, O., Cooksley, G., Sánchez, M., Arnaud, A., Crosetto, M., 2011. Analysis with C-and X-band satellite SAR data of the Portalet landslide area. Landslides, 8(2), 195 -206.

Higgins, G., Saunders, J.B., 1974. Mud volcanoes e their nature and origin. In: Jung, P., et al. (Eds.), Contributions to the Geology and Paleobiology of the Caribbean and Adjacent Areas. Verh. Aturforsch. Ges., vol. 84(1), pp. 101-152.

Hilley, G.E, Bürgmann, R., Ferretti, A., Novali, F., Rocca, F., 2004. Dynamics of slow-moving landslides from Permanent Scatterer analysis. Science, 304 (5679), 1952 - 1955.

Hommels, A., Scholte, K.H., Munoz-Sabater, J., Hanssen R.F., Van der Meer, F.D., Kroonenberg S.B., Aliyeva, E., Huseynov, D., Guliev, I., 2003. Preliminary Aster and InSAR imagery combination for mud volcano dynamics, Azerbaijan. International Geoscience and Remote Sensing Symposium 3, III: 1573- 1575.

Hooper, A., Zebker, H.A., Segall, P., Kampes, B.M., 2004. A new method for measuring deformation on volcanoes and other natural terrains using InSAR persistent scatterers. Geophysical Research Letters, 31(23), doi: 10.1029/2004GL021737.

Hooper, A., Zebker, H.A., 2007. Phase unwrapping in three dimensions with application to InSAR time series. JOSA A, 24(9), 2737-2747.

Hooper, A., Bekaert, D., Spaans, K., Arikan, M., 2012. Recent advances in SAR interferometry time series analysis for measuring crustal deformation. Tectonophysics, 514-517, pp.1-13. doi: 10.1016/j.tecto.2011.10.013.

Hreinsdóttir, S., Sigmundsson, F., Roberts, M.J., Björnsson, H., Grapenthin, R., Arason, P., Árnadóttir, T., Hólmjárn, J., Geirsson, H., Bennett, R. A., Gudmundsson, M.T., Oddsson, B., Ófeigsson, B.G., Villemin, T., Jónsson, T., Sturkell, E., Höskuldsson, Á., Larsen, G., Thordarson, T., Óladóttir, B.A., 2014. Volcanic plume height correlated with magma pressure change at Grímsvötn Volcano, Iceland. Nature Geoscience, Vol. 7, Doi: 10.1038/Ngeo2044.

Hudson, S.M., Johnson, C.L., Efendiyeva, M.A., Rowe, H.D., Feyzullayev, A.A., Aliyev, C.S., 2008. Stratigraphy and geochemical characterization of the Oligocene–Miocene Maikop series: implications for the paleogeography of Eastern Azerbaijan. Tectonophysics, 451, 40–55.

Hung, W.C., Hwang, C., Chen, Y.A., Chang, C.P., Yen, J.Y., Hooper, A., Yang, C.Y., 2011. Surface deformation from persistent scatterers SAR interferometry and fusion with leveling data: A case study over the Choushui River Alluvial Fan, Taiwan. Remote Sensing of Environment, 115(4), 957-967.

Hutnak, M., Hurwitz, S., Ingebritsen, S.E., and Hsieh, P.A., 2009. Numerical models of caldera deformation: Effects of multiphase and multicomponent hydrothermal fluid flow: Journal of Geophysical Research, v. 114, p. B04411, doi: 10.1029/2008JB006151.

Inan, S., Yaclin, M.N., Guliev, I.S., Kuliev, K., Feizullayev, A.A., 1997. Deep petroleum occurrences in the Lower Kura Depression, South Caspian Basin, Azerbaijan: an organic geochemical and basin modelling study. Marine and Petroleum Geology 14, 731– 762.

134

ISIDe Working Group, 2010. Italian Seismological Instrumental and parametric database: http://iside.rm.ingv.it.

Jackson, J., 1992. Partitioning of strike-slip convergent motion between Eurasia and Arabia in eastern Turkey and the Caucasus: Journal of Geophysical Research, v. 97, p. 12,471–12,479, doi: 10.1029/92JB00944.

Jackson, J., Priestley, K., Allen, M., Berberian, M., 2002. Active tectonics of the South Caspian Basin. Geophysical Journal International, 148, 214– 245.

Jakubov, A.A., Ali-Zade, A.A., Zeinalov, M.M., 1971. Mud Volcanoes of the Azerbaijan SSR. Atlas (in Russian), Azerbaijan Academy of Sciences, Baku.

Jay J., Costa F., Pritchard M., Larac L., Singer B., Herrin J., 2014. Locating magma reservoirs using InSAR and petrology before and during the 2011–2012 Cordón Caulle silicic eruption. Earth and Planetary Science Letters 395, 254–266.

Kampes, B.M., Hanssen, R.F., 2004. Ambiguity resolution for permanent scatterer interferometry. IEEE TGARS, 42, 2446 –2453.

Kampes, B.M., 2006. Radar Interferometry: Persistent Scatterer Technique. Springer, Dordrecht, The Netherlands.

Khain, V.Y., Grigoryants, B.V., Isayev, B.M., 1966. The West Caspian Fault and factors governing the formation of transverse faults in geocynclinal fold belts. MOIP Bulletin 41, 5–23.

Kimura, H., Yamaguchi, Y., 2000. Detection of landslide areas using radar interferometry. Phot. Eng. & Remote Sensing, 66 (3), 337 - 344.

Koçyiğit, A., Ylmaz, A., Adamia, S., Kuloshvili, S., 2001. Neotectonics of East Anatolian Plateau (Turkey) and Lesser Caucasus: Implication for transition from thrusting to strike-slip faulting. Geodin. Acta, 14, 177–195.

Kholodov V. N., 2002. Mud Volcanoes, Their Distribution Regularities and Genesis: Communication 1. Mud Volcanic Provinces and Morphology of Mud Volcanoes. Lithology and mineral resources. Vol. 37, No. 3.

Kopf, A.J., 2002. Significance of mud volcanism. Reviews of Geophysics 40 (2), 1005.

Kusumoto, S., Sudo, K., Kawabata, M., Uda, T., Fukuda, Y., 2014.Vertical movement during the quiescent phase of the Murono mud volcano, Niigata, Japan. Earth, Planets and Space 2014, 66:14.

Lagios, E., Sakkas, V., Novali, F., Bellotti, F., Ferretti, A., Vlachou, K., Dietrich, V., 2013. SqueeSAR™ and GPS ground deformation monitoring of Santorini Volcano (1992–2012): Tectonic implications. Tectonophysics 594, 38–59. http://dx.doi.org/10.1016/j.tecto.2013.03.012.

135

Lanari, R., Mora, O., Manunta, M., Mallorqui, J.J., Berardino, P., Sansosti, E., 2004. A small-baseline approach for investigating deformations on full-resolution differential SAR interferograms. IEEE Transactions on Geoscience and Remote Sensing, 42(7), 1377-1386.

Lanari, R., Casu, F., Manzo, M., Zeni, G., Berardino, P., Manunta, M., Pepe, A., 2007. An overview of the small baseline subset algorithm: A DInSAR technique for surface deformation analysis. Pure and Applied Geophysics, 164(4), 637 -661.

Lanari, R., Berardino, P., Bonano, M., Casu, F., Manconi, A., Manunta, M., Manzo, M., Pepe, A., Pepe, S., Sansosti, E., Solaro, G., Tizzoni, P., Zeni, G., 2010. Surface displacements associated with the L’Aquila 2009 Mw 6.3 earthquake (Central Italy): new evidence from DInSAR time series analysis, Geophys. Res. Lett., 37, L20309, doi: 10.1029/2010GL044780.

Lindquist, S.J., 1999. Petroleum systems of the Po Basin Province of Northern Italy and the Northern Adriatic Sea: Porto Garibaldi (Biogenic), Meride/Riva di Solto (Thermal), and Marnoso Arenacea (Thermal). USGS Open-File Report, 99-50-M.

Lundgren, P., Usai, S., Sansosti, E., Lanari, R., Tesauro, M., Fornaro, G., Berardino, P., 2001. Modelling surface deformation observed with SAR interferometry at Campi Flegrei caldera. Journal of Geophysical Research – Solid Earth, Vol. 106, No. B9, pp. 19, 355- 366.

Lyons, S., Sandwell, D., 2003. Fault creep along the southern San Andreas from interferometric synthetic aperture radar, permanent scatterers, and stacking. Journal Of Geophysical Research, Vol. 108, No. B1, 2047, doi:10.1029/2002JB001831.

Maesano, F., Toscani, G., Burrato, P., Mirabella, F., D’ambrogi, C., Basili, R., 2013. Deriving thrust fault slip rates from geological modeling: examples from the Marche coastal and offshore contraction belt, Northern Apennines, Italy, Mar. Petrol. Geol. 42, 122–134, doi:10.1016/j.marpetgeo.2012.10.008.

Maesano, F., D’Ambrogi, C., Burrato, P., Toscani, G., 2015. Slip-rates of blind thrusts in the Po sedimentary basin (Northern Apennines, Italy). Tectonophysics, 643, 8–25, doi:10.1016/j.tecto.2014.12.007.

Manga, M., Brumm, M., Rudolph, M. L., 2009. Earthquake triggering of mud volcanoes. Mar. Petrol. Geol., 26, 1785–1798.

Manga, M., Bonini, M., 2012. Large historical eruptions at subaerial mud volcanoes, Italy. Nat. Hazards Earth Syst. Sci., 12, 3377–3386, doi:10.5194/nhess-12-3377-2012.

Mangino, S., Priestly, K., 1998. The crustal structure of the Southern Caspian region. Geophysical Journal International, v. 133, p. 630–648, doi: 10.1046/j.1365-246X.1998.00520.x.

Marabini S., Baldi P., Benini A., Mulargia F., Vai G.B., Valensise G., 1986. Strutture tettoniche da monitorare nei dintorni di S. Sofia (Appennino Forlivese). Atti 4° Conv. G.N.G.T.S., Roma 1985, 569- 580.

Mariotti, G., Doglioni, C., 2000. The dip of the foreland monocline in the Alps and Apennines, Earth Planet. Sc. Lett. 181, 191–202, doi:10.1016/S0012-821X(00)00192-8.

136

Martinelli, G., Albarello, D., Mucciarelli, M., 1995. Radon emissions from mud volcanoes in northern Italy: Possible connection with local seismicity. Geophys. Res. Lett., 22, 1989 – 1992.

Martinelli, G., Judd, A., 2004. Mud volcanoes of Italy, Geol. J., 39, 49 – 61.

Massironi, M., Zampieri, D., Bianchi, M., Schiavo, A., Franceschini, A., 2009. Use of PSInSAR™ data to infer active tectonics: Clues on the differential uplift across the Giudicarie belt (Central-Eastern Alps, Italy). Tectonophysics, 476, 297–303, doi:10.1016/j.tecto.2009.05.025.

Massonnet, D., Rossi, M., Carmona, C., Adragna, F., Peltzer, G., Feigl, K., et al., 1993. The displacement field of the Landers earthquake mapped by radar interferometry. Nature, 364, 138−142.

Massonnet, D., Holzer, T., & Vadon, H., 1997. Land subsidence caused by the East Mesa geothermal field, California, observed using SAR interferometry. Earth and Planetary Science Letters, 24(8), 901 – 904.

Massonnet, D., Feigl, K.L., 1998. Radar interferometry and its applications to changes in the Earth's surface, Rev. Geophys., 36, 441–500.

Mazzini, A., Svensen, H., Planke, S., Guliyev, I., Akhmanov, G.G., Fallik, T., Banks, D., 2009. When mud volcanoes sleep: insight from seep geochemistry at the Dashgil mud volcano, Azerbaijan. Marine and Petroleum Geology 26, 1704–1715.

Meisina, C., Zucca, F., Notti, D., Colombo, A., Cucchi, G., Giannico, C., Bianchi, M., 2008. Geological Interpretation of PSInSAR Data at Regional Scale. Sensors, 8(11), 7469-7492; doi:10.3390/s8117469.

Mellors, R.J., Bunyapanasarn, T., Panahi, B., 2005. InSAR Analysis of the Absheron Peninsula and Nearby Areas, Azerbaijan. Mud Volcanoes, Geodynamics and Seismicity, NATO Science Series, Volume 51, pp 201-209.

Mellors, R.J., Kilb, D., Aliyev, A., Gasanov, A., Yetirmishli, G., 2007. Correlations between earthquakes and largemud volcano eruptions. J. Geophys. Res.112, B04304, doi: 10.1029/2006JB004489.

Michetti, A.M., Giardina, F., Livio, F., Mueller, K., Serva, L., Sileo, G., Vittori, E., Devoti, R., Riguzzi, F., Carcano, C., Rogledi, S., Bonadeo, L., Brunamonte, F., Fioraso, G., 2012. Active compressional tectonics, Quaternary capable faults, and the seismic landscape of the Po Plain (northern Italy). Ann. Geophys. 55 (5), 969–1001. http://dx.doi.org/10.4401/ag-5462.

Minissale, A., Magro, G., Martinelli, G., Vaselli, O, Tassi, F., 2000. Fluid geochemical transect in the northern Apennines (central-northern Italy): Fluid genesis and migration and tectonic implications, Tectonophysics, 319, 199 – 222.

Monserrat, O., Crosetto, M., Cuevas M., Crippa B., 2011. The thermal expansion component of Persistent Scatterer Interferometry observations. IEEE Geoscience and Remote Sensing Letters, 8, 864-868.

Monserrat, O., 2012. Deformation measurement and monitoring with ground-based SAR. PhD thesis, Technical University of Catalonia.

137

Montone, P., Mariucci, M. T., Pierdominici S., 2012. The Italian present-day stress map. Geophysical Journal International, 1-12, doi: 10.1111/j.1365-246X.2012.05391.

Mora, O., Mallorquì, J.J., Broquetas, A., 2003. Linear and nonlinear terrain deformation maps from a reduced set of interferometric SAR images. IEEE Trans Geosci. Remote Sens. 41; 2243-2253.

Morley, C.K., Guerin, G., 1996. Comparison of gravity-driven deformation styles and behavior associated with mobile shales and salt. Tectonics 15, 1154–1170.

Morley, C.K., Van Rensbergen, P., Hillis, R.R., Maltman, A.J., Morley, C.K., 2003. Outcrop examples of mudstone intrusions from the Jerudong anticline, Brunei Darussalam and inferences for hydrocarbon reservoirs. Subsurface Sediment Mobilization: Geological Society, London, Special Publications, 216, pp. 381–394.

Mueller, T.G., Pusuluri, N.B., Mathias, K.K., Cornelius, P.L., Barnhisel, R.I., Shearer S.A., 2004. Map quality for ordinary Kriging and inverse distance weighted interpolation. Soil Science, Society of America Journal, 68, 2042–2047.

Musson, R.M., Haynes, M., Ferretti, A., 2004. Space-based tectonic modeling in subduction areas using PSInSAR. Seismological Research Letters, 75(5), 598-606.

Pagli, C., Sigmundsson, F., Arnadòttir, T., Einarsson, P., Sturkell, E., 2006. Deflation of the Askja volcanic system: Constraints on the deformation source from combined inversion of satellite radar interferograms and GPS measurements. Journal of Volcanology and Geothermal Research 152, 97–108.

Pagli, C., Wright, T.J., Ebinger, C.J., Yun, S. H., Cann, J. R., Barnie, T., Ayele, A., 2012. Shallow axial magma chamber at the slow-spreading Erta Ale Ridge. Nature Geoscience, doi: 10.1038/NGEO1414.

Pedersen, R., Jónsson, S., Arnadóttir, T., Sigmundsson, F., Feigle, K.L., 2003. Fault slip distribution of two June 2000 MW6.5 earthquakes in South Iceland estimated from joint inversion of InSAR and GPS measurements. Earth and Planetary Science Letters 213, 487–502.

Pedersen, R., Sigmundsson, F., 2006. Temporal development of the 1999 intrusive episode in the Eyjafjallajökull volcano, Iceland, derived from InSAR images. Bull. Volcanol., 68, 377–393.

Peltzer, G., Rosen, P., Rogez, F., Hudnut, K.,1996. Postseismic rebound in fault step-overs caused by pore fluid flow. Science, 273, 1202-1204.

Pepe, A., Sansosti, E., Berardino, P., Lanari, R., 2005. On the generation of ERS/ENVISAT DInSAR time-series via the SBAS technique. IEEE Geoscience and Remote Sensing Letters, 2(3), 265-269.

Pepe, A., Manunta, M., Mazzarella, G., Lanari, R., 2007. A space-time minimum cost flow phase unwrapping algorithm for the generation of persistent scatterers deformation time-series. Proc. of International Geoscience and Remote Sensing Symposium (IGARSS), Barcelona (Spain).

138

Pepe, A., Berardino, P., Bonano, M., Euillades, L.D., Lanari, R., Sansosti, E., 2011. SBAS-based satellite orbit correction for the generation of DInSAR time-series: Application to RADARSAT-1 data. IEEE Transactions on Geoscience and Remote Sensing, 49(12), 5150-5165.

Perrone, G., Morelli, M., Piana, F., Fioraso, G., Nicolò, G., Mallen, L., Cadoppi, P., Balestro, G., Tallone, S., 2013. Current tectonic activity and differential uplift along the Cottian Alps/Po Plain boundary (NW Italy) as derived by PS-InSAR data. Journal of Geodynamics 66, 65– 78, doi:10.1016/j.jog.2013.02.004.

Petrucci, F., Bigi, B., Cavazzini, R., Morestori, L., Panicieri, E., Pecorari, M., Rossetti, G., 1975. Ricerche sulle acque sotterranee della pianura parmense e piacentina: sezioni interpretative degli acquiferi (parte prima). Pubblicazioni Istituto di Geologia Università di Parma, Grafiche STEP Editrice Parma, pp. 14.

Peyret, M., Masson, F., Yavasoglu, H., Ergintav, S., Reilinger, R., 2013. Present-day strain distribution across a segment of the central bend of the North Anatolian Fault Zone from a Persistent-Scatterers InSAR analysis of the ERS and Envisat archives. Geophys. J. Int., 192, 929–945, doi:10.1093/gji/ggs085.

Pezzo, G., Merryman Boncori, J.P., Tolomei, C., Salvi, S., Atzori, S., Antonioli, A., Trasatti, E., Novali, F., Serpelloni, E., Candela, L., Giuliani, R., 2013. Coseismic deformation and source modeling of the May 2012 Emilia (Northern Italy) earthquakes. Seismological Research Letters, v. 84 n. 4 p. 645-655, doi: 10.1785/0220120171.

Pezzo, G., Tolomei, C., Atzori, S., Salvi, S., Merryman Boncori, J.P., 2014. Seismic cycle measurements from SAR interferometry: inter-, co- and post-seismic deformation associated to the 2012 Emilia seismic sequence. Atti del 33° Convegno Nazionale GNGTS. 001-260 volume 1 pp.107-113.

Picotti, V., Pazzaglia F.J., 2008. A new active tectonic model for the construction of the Northern Apennines mountain front near Bologna (Italy). J. Geophys. Res. 113, B8, doi:10.1029/ 2007JB005307.

Pieri, M., Groppi, G., 1981. Subsurface geological structure of the Po Plain, Italy, In Progetto Finalizzato Geodinamica, C.N.R., Publ. n. 414.

Pieri, M., 1983. Three seismic profiles through the Po Plain. Seismic Expression of Structural Styles. A Picture and Work Atlas. In A. W. Bally (Ed.), Am. Assoc. Pet. Geol. Stud. Geol., Vol. 15., pp. 3.4.1/8−3.4.1/26.

Pieri, M., 2001. Italian petroleum geology, in Anatomy of an Orogen: The Apennines and Adjacent Mediterranean Basins, edited by C.G.B. Vai and I.P. Martini, pp. 533 – 550, Springer, New York.

Planke, S., Svensen, H., Hovland, M., Banks, D.A., Jamtveit, B., 2003. Mud and fluid migration in active mud volcanoes in Azerbaijan. Geo-Marine Letters 23, 258–268.

Pondrelli, S., Salimbeni, S., Ekstrom, G., Morelli, A., Gasperini, P., Vannucci, G., 2006. The Italian CMT dataset from 1977 to the present, Phys. Earth Planet. In. 159, 3–4, 286–303, doi:10.1016/j.pepi.2006.07.008.

139

Pondrelli, S., Salimbeni, S., Perfetti, P., Danecek, P., 2012. Quick regional centroid moment tensor solutions for the Emilia 2012 (northern Italy) seismic sequence. Annals of Geophysics 55, 615-621, doi: 10.4401/ag-6146.

Ponza, A., Pazzaglia, F.J., Picotti, V., 2010. Thrust-fold activity at the mountain front of the Northern Apennines (Italy) from quantitative landscape analysis. Geomorphology 123, 211–231, doi:10.1016/j.geomorph.2010.06.008.

Price, E., Sandwell, D. T., 1998. Small-scale deformations associated with the 1992 Landers, California Earthquake mapped by synthetic aperture radar interferometry phase gradients, J. Geophys. Res., 103, 27001-27016.

Principi, G. Treves, B., 1984. Il sistema corso-appenninico come prisma d’accrezione. Riflessi sul problema generale del limite Alpi-Appennini. Memorie della Società Geologica Italiana, 28, 549– 576.

Pritchard, M., Simons, M., 2004. Surveying Volcanic Arcs with Satellite Radar Interferometry: The Central Andes, Kamchatka, and Beyond. GSA Today; v. 14; no. 8, doi: 10.1130/1052-5173.

Ramsay, J.G., 1967. Folding and Fracturing of Rocks. McGraw Hill, New York, NY, USA.

RER and ENI Agip, 1998. Riserve idriche sotterranee della Regione Emilia-Romagna. Società Elaborazioni Cartografiche, Florence, Italy, 119 pp.

Reilinger, R., McClusky, S., Vernant, P., Lawrence, S., Ergintav, S., Cakmak, R., Ozener, H., Kadirov, F., Guliev, I., Stepanyan, R., Nadariya, M., Hahubia, G., Mahmoud, S., Sakr, K., ArRajehi, A., Paradissis, D., Al-Aydrus, A., Prilepin, M., Guseva, T., Evren, E., Dmitrotsa, A., 2006. GPS constraints on continental deformation in the Africa-Arabia-Eurasia continental collision zone and implications for the dynamics of plate interactions, J. Geophys. Res., 111, B05411, doi:10.1029/2005JB004051.

Reynolds, A.D., Simmons, M.D., Bowman, M.B.J., Henton, J., Brayshaw, A.C., Ali-Zade, A.A., Guliyev, I.S., Suleymanova, S.F., Ateava, E.Z., Mamedova, D.N., Koshkarly, R.O., 1998. Implications of outcrop geology for reservoirs in the Neogene productive series; Apsheron Peninsula, Azerbaijan. AAPG Bull. 82, 25e49.

Ricci Lucchi, F., Colalongo, M.L., Cremonini, G., Gasperi, G., Iaccarino, S., Papani, G., Raffi, I., Rio, D., 1982. Evoluzione sedimentaria e paleogeografica del margine appenninico. Guide Geologiche Regionali Soc. Geol. Ital. (1982), pp. 17–46.

Ricci Lucchi, F., 1986. The Oligocene to recent foreland basins of the Northern Apennines. Spec. Publs. Int. Ass. Sediment. 8, 105–139.

Ricci Lucchi, F., 1987. Semi-allochthonous sedimentation in the Apenninic thrust belt. Sediment. Geol. 50, 119–134.

Righini, G., Raspini, F., Moretti, S., Cigna, F., 2011. Unsustainable use of groundwater resources in agricultural and urban areas: a persistent scatterer study of land subsidence at the basin scale. WIT Transactions on Ecology and the Environment, Vol 144 pp. 81-92, WIT Press 2011.

140

Righini, G., Pancioli, V., Casagli, N., 2012. Updating landslide inventory maps using Persistent Scatterers Interferometry (PSI). International Journal of Remote Sensing, 33 (7), pp. 2068-2096.

Roberts, K. S., Davies, R.J., Stewart, S. A., 2010. Structure of exhumed mud volcano feeder complexes, Azerbaijan. Basin Research 22, 439–451.

Roberts, K.S., Stewart, S.A., Davies, R.J., Evans, R.J., 2011a. Sector collapse of mud volcanoes, Azerbaijan. Journal of the Geological Society 168, 49–60.

Roberts, K.S., Davies, R.J., Stewart, S.A., Tingay, M., 2011b. Structural controls on mud volcano vent distributions: examples from Azerbaijan and Lusi, east Java. Journal of the Geological Society 168, 1013–1030.

Rosen, P.A., Hensley, S., Joughin, I.R., Li, F.K., Madsen, S.N., Rodriguez, E., Goldstein, R.M., 2000. Synthetic aperture radar interferometry. Proceedings of the IEEE, 88 (3), 333-382.

Rossi, M., Rogledi, S., Barbacini, G., Casadei, D., Iaccarino, S., Papani, G., 2002. Tectono-stratigraphic architecture of Messinian piggyback basins of Norther Apennines: the Emilia folds in the Reggio-Modena area and comparison with the Lombardia and Romagna sector. Bollettino Società Geologica Italiana, Volume speciale n. 1, 437-447.

Rott, H., Scheuchel, B., Siegel A., 1999. Monitoring very slow slope movements by means of SAR interferometry: A case study from mass waste above a reservoir in the Otztal Alps, Austria. Geophysical Research Letters, 26(11), 1629 - 1632.

Rovida, A., Camassi, R., Gasperini, P., And Stucchi, M., 2011. CPTI11, the 2011 version of the Parametric Catalogue of Italian Earthquakes, Milano, Bologna, http://emidius.mi.ingv.it/ CPTI/, doi: 10.6092/INGV.IT-CPTI1.

Rudolph, M.L., Shirzaei, M., Manga, M., Fukushima, Y., 2013. Evolution and future of the Lusi mud eruption inferred from ground deformation. Geophysical Research Letters, Vol. 40, 1–4, Doi:10.1002/Grl.50189.

Salvi, S., Atzori, S., Tolomei, C., Allievi, J., Ferretti, A., Rocca, F., Prati, C., Stramondo, S., Feuillet, N., 2004. Inflation rate of the Colli Albani volcanic complex retrieved by the permanent scatterers SAR interferometry technique. Geophysical Research Letters, 31(12).

Sani F., Bonini M., Piccardi L., Vannucci G.; Delle Donne D.; Benvenuti M; Moratti G.; Corti G.; Montanari D.; Sedda L.; Tanini C., 2009. Late Pliocene Quaternary evolution of outermost 145 hinterland basins of the orthern Apennines (Italy), and their relevance to active tectonics. Tectonophysics, 476, 336- 356.

Schmidt, D. A., Burgmann, R., 2003. Time dependent land uplift and subsidence in the Santa Clara Valley, California, from a large InSAR data set. J. Geophys. Res., 108(B9), 2416, doi:10.1029/2002JB002267.

Scognamiglio, L., Margheriti, L., Mele, F.M., Tinti, E., Bono, A., De Gori, P., Lauciani, V., Lucente, F.P., Mandiello, A.G., Marcocci, C., Mazza, S., Pintore, S., Quintiliani, M., 2012. The 2012 Pianura

141

Padana Emiliana seimic sequence: locations, moment tensors and magnitudes. Ann. Geophys. 55, 4, doi:10.4401/ag-6159.

Scrocca, D., E. Carminati, C. Doglioni, D. Marcantoni, 2007. Slab retreat and active shortening along the Central-Northern Apennines, Book Section: Thrust Belts and Foreland Basins Frontiers in Earth Sciences, 471–487.

Selvaggi, G., Ferulano, F., Di Bona, M., Frepoli, A., Azzara, R., Basili, A., Chiarabba, C., Ciaccio, M.G., Di Luccio, F., Lucente, F.P., Margheriti, L., Nostro, C., 2001. The Mw 5.4 Reggio Emilia 1996 earthquake: active compressional tectonics in the Po Plain, Italy. Geophysical Journal International 144, 1–13.

Serpelloni, E., Anzidei, M., Baldi, P., Casula, G., Galvani, A., 2005. Crustal velocity and strain-rate fields in Italy and surrounding regions: new results from the analysis of permanent and non-permanent GPS networks. Geophy. J. Int., 161(3), 861–880. doi:10.1111/j.1365–246X.2005.02618.x.

Sigmundsson, F., Vadon, H., Massonnet, D., 1997. Readjustment of the Krafla spreading segment to crustal rifling measured by Satellite Radar Interferometry. Geophysical Research Letters, Vol. 24, No.15, 1843-1846.

Sigmundsson, F., Hreinsdóttir, S., Hooper, A., Arnadottir, T., Pedersen, R., Roberts, M. J., Oskarsson, N., Auriac, A., Decriem, J., Einarsson, P., Geirsson, H., Hensch, M., Ofeigsson, B. G., Sturkell, E., Sveinbjornsson, H., Feigl, K. L., 2010. Intrusion triggering of the 2010 Eyjafjallajokull explosive eruption. Nature, Vol. 468, (7322), 426-430, doi: 10.1038/nature 09558.

Sobornov, K. O., 1994. Structure and petroleum potential of the Dagestan thrust belt, northeastern Caucasus, Russia. Bull. Can. Petrol. Geol., 42(3), 352–364.

Sobornov, K. O., 1996. Lateral variations in structural styles of tectonic wedging in the northeastern Caucasus. Bull. Can. Petrol. Geol., 44(2), 385–399.

Soto, J.I., Santos-Betancor, I., Sánchez Borrego, I., Macellari, C.E., 2011. Shale Diapirism and Associated Folding History in the South Caspian Basin (Offshore Azerbaijan). Search and Discovery Article n.30162.

Stewart, S. A., Davies, R.J., 2006. Structure and emplacement of mud volcano systems in the South Caspian Basin. AAPG Bulletin, 90, 771–786.

Stramondo, S., Bignami, C., Chini, N., Pierdicca, N., Tertulliani, A., 2006. Satellite radar and optical remote sensing for earthquake damage detection: results from different case studies. International Journal of Remote Sensing, 27, pp. 4433–4447.

Stramondo, S., Saroli, M., Tolomei, C., Moro, M., Doumaz, F., Pesci, A., Loddo, F., Baldi, P., Boschi, E., 2007. Surface movements in Bologna (Po Plain, Italy) detected by multitemporal DInSAR, Remote Sens. Environ., 110, 304-316.

142

Strozzi, T., Tosi, L., Wegmüller, U., Teatini, P., Carbognin, L., 2001. Thematic and land subsidence maps of the Lagoon of Venice from ERS SAR interferometry. CoRiLa Research Program 2001 Results (ed. by P. Campostrini), IVSLA Venezia, 345-355.

Strozzi, T., Farina, P., Corsini, A., Ambrosi, C., Thüring, M., Zilger, J., Wiesmann, A., Wegmüller, U., Werner, C., 2005. Survey and monitoring of landslide displacements by means of L-band satellite SAR interferometry. Landslides, 2(3), 193-201.

Sultan, M., Arvidson, R. E, Sturchio, N. C., Guinness A. E., 1987. Lithologic mapping in arid regions with Landsat thematic mapper data: Meatiq dome, Egypt. Geological Society of America Bulletin, 99, 6, 748-762.

Teatini, P., Tosi, L., Strozzi, T., Carbognin, L., Wegmüller, U., Rizzetto, F., 2005. Mapping regional land displacements in the Venice coastland by an integrated monitoring system. Remote Sensing of Environment, 98(4), 403-413.

Tevelev, A.V., Blyumkin, E.I., 1990. Comparative tectonics from space images of the Kura and Surkhandar’ya depressions. Soviet Journal of Remote Sensing, v. 7, no. 3, p. 509–516.

Tizzani, R., Castaldo, R., Solaro, G., Pepe, S., Bonano, M., Casu, F., Manunta, M., Manzo, M., Pepe, A., Samsonov, S., Lanari, R., Sansosti, E., 2013. New insights into the 2012 Emilia (Italy) Seismic sequence through advanced numerical modeling of ground deformation InSAR measurements. Geophys. Res. Lett. 40, 10, 1971–1977, Doi:10.1002/Grl.50290.

Tolomei, C., Salvi, S., Merryman Boncori, J.P., Pezzo, G., 2015. InSAR measurement of crustal deformation transients during the earthquake preparation processes: a review. Bollettino di Geofisica Teorica ed Applicata Vol. 56, n. 2, pp. 151-166; doi 10.4430/bgta0143.

Toscani, G., Burrato, P., Di Bucci, D., Seno, S., Valensise, G., 2009. Plio-Quaternary tectonic evolution of the northern Apennines thrust fronts (Bologna-Ferrara section, Italy): seismotectonic implications, It. J. Geosc. 128, 605–613, doi:10. 3301/IJG.2009.128.2.605.

Triep, E.G., Abers, G.A., Lerner-Lam, A.L., Mishatkin, V., Zakharchenko, N., Starovoit, O., 1995. Active thrust front of the Greater Caucasus: The April 29,1991, Racha earthquake sequence and its tectonic implications. Journal of Geophysical Research, v. 100, no. B3, p. 4011–4033, doi: 10.1029/94JB02597.

Vai, G.B., 2001. Structure and stratigraphy: an overview. In: Vai, G.B., Martini, I.P. (Eds.), Anatomy of an Orogen: The Apennines and Adjacent Mediterranean Basins. Kluwer Academic Publisher, pp. 15–32.

Valensise, G., Pantosti, D., 2001. Database of potential sources for earthquakes larger than M 5.5 in Italy, Ann. Geofis. 44, 4, 180 pp., doi:10.4401/ag-3562. van Leijen, F., 2014. Persistent Scatterer Interferometry based on geodetic estimation theory. Doctoral dissertation, TU Delft, Delft University of Technology.

Vannoli, P., Basili, R., Valensise, G., 2004. New geomorphic evidence for anticlinal growth driven by blind-thrust faulting along the northern Marche coastal belt (central Italy). J Seism 8:297–314

143

Vannoli, P., Burrato, P., Valensise, G., 2014. The seismotectonics of the Po Plain (northern Italy): tectonic diversity in a blind faulting domain. Pure Appl. Geophys. 172 (2015), 1105–1142_ 2014 Springer Basel. DOI 10.1007/s00024-014-0873-0

ViDEPI database; http://unmig.sviluppoeconomico.gov.it/videpi/en/

Vilardo, G., Ventura, G., Terranova, C., Matano, F., Nardò, S., 2009. Ground deformation due to tectonic, hydrothermal, gravity, hydrogeological, and anthropic processes in the Campania Region (Southern Italy) from Permanent Scatterers Synthetic Aperture Radar Interferometry. Remote Sensing of Environment 113, 197–212.

Vincent, S.J., Carter, A., Lavrishchev, A., Rice, S.P., Barabadze, T.G., Hovius, N., 2011. The exhumation of the western Greater Caucasus: A thermochronometric study. Geol. Mag., 148,1–21.

Waltham, T., 2002. Sinking cities—feature. Geology Today 18 (3), 95–100.

Wang, C. Y., Manga, M., 2010. Hydrologic responses to earthquakes and a general metric. Geofluids 10, 206–216.

Wang, R., Lorenzo-Martin, F., Roth, F., 2006. PSGRN/PSCMP - A new code for calculating co- and post-seismic deformation, geoid and gravity changes based on the viscoelastic-gravitational dislocation theory. Comput. Geosci., 32, 527–541.

Werner, C., Wegmüller, U., Strozzi, T., Wiesmann, A., 2003. Interferometric point target analysis for deformation mapping. Proc. of International Geoscience and Remote Sensing Symposium (IGARSS), Toulouse (France).

Westaway, R., 1994. Present-day kinematics of the Middle East and eastern Mediterranean. J. Geophys. Res., 99 (B6), 12,071–12,090, doi:10.1029/94JB00335.

Wicks, C., Thatcher, W., Dzurisin, D., 1998. Migration of fluids beneath Yellowstone caldera inferred from satellite radar interferometry: Science, v. 282, p. 458–462, doi: 10.1126/science.282.5388.458.

Wilson, L.F., Pazzaglia, F.J., Anastasio, D.J., 2009. A fluvial record of active fault propagation folding, Salsomaggiore anticline, northern Apennines, Italy. Journal of Geophysical Research 114 (B8). doi.10.1029/2008JB005984.

Wright, T.J., Parsons, B., England, P.C., Fielding, E. J., 2004. InSAR Observations of Low Slip Rates on the Major Faults of Western Tibet Science 305, 236; doi: 10.1126/science.1096388.

Wright, T.J., Sigmundsson, F., Pagli, C., Belachew, M., Hamling, I.J., Brandsdóttir, B., Keir, D., Pedersen, R., Ayele, A., Ebinger, C., Einarsson, P., Lewi, E., Calais, E., 2012. Geophysical constraints on the dynamics of spreading centers from rifting episodes on land. Nature Geoscience, Review Article, doi: 10.1038/NGEO1428.

Yassir, N.A., Bell, J.S., 1996. Abnormally high fluid pressure and associated porosities and stress regimes in sedimentary basins. SPE Formation Evaluation 48, 5–10. http://dx.doi.org/10.2118/28139-PA.

144

Yusifov, M., Rabinowitz, P. D., 2004. Classification of mud volcanoes in the South Caspian Basin, offshore Azerbaijan. Marine and Petroleum Geology, 21, 965–975.

Zebker, H.A., Villasenor, J., 1992. Decorrelation in interferometric radar echoes. IEEE Transactions on Geoscience and Remote Sensing, 30, 950−959.

Zebker, H.A., Rosen, P.A., Hensley, S., 1997. Atmospheric effects in interferometric synthetic aperture radar surface deformation and topographic maps. Journal of Geophysical Research: Solid Earth (1978– 2012), 102(B4), 7547-7563.

Zebker, H.A., Rosen, P.A., Goldstein, R.M., Gabriel, A., Werner, C. L., 1994. On the derivation of coseismic displacement fields using differential radar interferometry: The Landers earthquake: Journal of Geophysical Research, v. 99, p. 19,617–19,634.

Zeinalov, G. A., 2000. Importance of remote-sensing data in structural geologic Analysis of oil- and gas- bearing regions of Azerbaijan. Natural Resources Research, Vol. 9, No. 4.

Zerbini, S., Richter, B., Rocca, F., van Dam, T., Matonti, F., 2007. A Combination of Space and Terrestrial Geodetic Techniques to Monitor Land Subsidence: Case Study, the Southeastern Po Plain, Italy. Journal of Geophysical Research, Vol. 112, B05401, doi:10.1029/2006jb004338.

Zonenshain, L.P., Le Pichon, X., 1986. Deep basins of the Black Sea and Caspian Sea as remnants of Mesozoic back-arc basins. Tectonophysics, 123, 181–211.

145