SCUOLA DOTTORALE IN
GEOLOGIA DELL'AMBIENTE E DELLE RISORSE (SDiGAR)
XVII CICLO
Surface deformation and magma intrusion
along divergent plate boundaries
Dottorando:
Daniele Trippanera
__________________
Docente Guida:
Valerio Acocella
__________________
Coordinatore:
Claudio Faccenna
__________________
firma
firma
firma
Co-Tutor:
Dr. Joel Ruch: ricercatore presso King Abdullah University of Science and
Technology (KAUST), Thuwal, Saudi Arabia.
Revisori:
Dott. Marco Neri: primo ricercatore presso l’Istituto Nazionale di Geofisica e
Vulcanologia, Osservatorio Etneo, Sezione di Catania, Unità Funzionale
Gravimetria e Magnetismo.
Dott.ssa Eleonora Rivalta: PI of an ERC Starting grant research group, project
CCMP-POMPEI presso il GeoForschungsZentrum GFZ German Research
Centre for Geosciences.
Commissari:
Prof.ssa Maria Rita Palombo: professore Associato SSD
Dipartimento di Scienze della Terra Università Roma la Sapienza.
GEO/01
Prof. Orlando Vaselli: professore Associato SSD GEO/08 Dipartimento di
Scienze della Terra Università di Firenze.
Dott. Marco Bonini: professore Associato SSD GEO/03, ricercatore CNR,
Istituto di Geoscienze e Georisorse Dipartimento di Scienze della Terra,
Università di Firenze.
Table of Contents
Abstract (in English) ........................................................................................................................................... 1
Riassunto (in Italian) .......................................................................................................................................... 2
Introduction ....................................................................................................................................................... 4
1.2 Methods.................................................................................................................................................... 5
1.2.1 Field surveys ...................................................................................................................................... 5
1.2.2 Analogue models ............................................................................................................................... 8
Chapter 2: Surface deformation along the divergent plate boundaries: understanding magmatic vs.
tectonic processes............................................................................................................................................11
Abstract ........................................................................................................................................................11
1. Introduction .............................................................................................................................................12
2. Tectonic setting ........................................................................................................................................14
3. Methods ...................................................................................................................................................16
4. Results ......................................................................................................................................................17
4.1. Eruptive fissures ................................................................................................................................17
4.1.1 Lakagigar .....................................................................................................................................17
4.1.2 Eldgjá ..........................................................................................................................................18
4.1.3 Bardarbunga ..............................................................................................................................19
4.1.4 Sveinagja ....................................................................................................................................20
4.1.5 Sveinar .......................................................................................................................................21
4.2. Rift segments .....................................................................................................................................21
4.2.1 The Krafla magmatic system ......................................................................................................21
4.2.2 Vogar rift zone ...........................................................................................................................23
4.2.3 Thingvellir rift zone ....................................................................................................................24
4.2.4 Fantale magmatic system ..........................................................................................................24
5. Discussion ................................................................................................................................................25
5.1. Interpreting the collected data..........................................................................................................25
5.2. A general model.................................................................................................................................30
6. Conclusions ..............................................................................................................................................32
References ..........................................................................................................................................33
Figure and tables ................................................................................................................................39
Chapter 3: Experiments of dike-induced deformation: an application to divergent plate boundaries .......57
Abstract ........................................................................................................................................................57
1. Introduction .............................................................................................................................................58
2. Experimental setup and scaling ................................................................................................................59
2.1. Setup and methods............................................................................................................................59
2.2. Scaling and materials .........................................................................................................................61
2.3. Assumptions and limitations .............................................................................................................62
3. Results ......................................................................................................................................................63
3.1. Setup A: upward propagating intrusions. Experiment A2 .................................................................63
3.2. Setup B and C: thickening intrusions at constant depth ...................................................................64
3.2.1 Experiment B1: shallow rectangular intrusive complex .............................................................64
3.2.2 Experiment B4: medium depth rectangular intrusive complex .................................................66
3.2.3 Experiment B6: deep rectangular intrusive complex .................................................................67
3.2.4 Experiment C1: shallow triangular intrusive complex ................................................................68
3.2.5 Experiment C2: medium depth triangular intrusive complex ....................................................69
4. Discussions ...............................................................................................................................................69
4.1. Overall deformation pattern and setup relevance ............................................................................69
4.2. Effect of the intrusions depth and geometry on the deformation....................................................72
4.3. Fault propagation ..............................................................................................................................73
4.3.1 Normal faulting ...........................................................................................................................73
4.3.2 Arcuate normal/reverse faults ...................................................................................................74
4.4. Comparison to rift zones ...................................................................................................................75
4.4.1 Single rifting episodes .................................................................................................................76
4.4.2 Multiple rifting episodes .............................................................................................................77
4.4.3 Intrusion depth vs. thickness ......................................................................................................79
5. Conclusions ..............................................................................................................................................81
References ..........................................................................................................................................82
Figure and tables ................................................................................................................................87
Chapter 4: Experiments of dike-induced deformation: an application to divergent plate boundaries .....105
Abstract ......................................................................................................................................................105
1. Introduction ...........................................................................................................................................105
2. Field analysis ...........................................................................................................................................105
3. Analogue models ....................................................................................................................................107
4. Discussion and conclusions.....................................................................................................................109
References ........................................................................................................................................110
Chapter 5: General conclusions .....................................................................................................................112
Abstract
The interest in the role of magma in splitting plates at divergent plate boundaries has
been recently re-enhanced. However, the peculiar mechanism by which the magma affects
the geometry, the kinematics, and the temporal evolution of a rift is still poorly understood,
especially in a long time perspective. Moreover, it is also necessary to better define how and
to what extend the regional tectonics - by means of plate pull mechanism - affects the rift
structures formation along these margins. The aim of this work is to address these issues
through field survey and analogue modeling.
Field survey consists of studying the surface deformation along different fissural portions
of the magmatic systems at the Neovolcanic Zone of Iceland and the Main Ethiopian Rift,
focusing mainly on: 1) single eruptive fissures (Laki, Eldgjá and Bardarbunga in Iceland) or
narrow fissure zones hosting recent eruptive fissures (i.e., Sveinagjá and Sveinar in
Iceland), 2) wider fissure zones where several rifting episodes occurred (i.e. Krafla, Vogar
and Thingvellir in Iceland and Fantale in Ethiopia). In all these areas, fault and extension
fracture geometries and kinematics have been systematically characterized, including the
analysis of the structure of the fault’s lateral terminations, conceived as possible indicators
of their propagation direction.
In addition, these data have been integrated with the study of the roots (depth of about
1.5 km) of the fossil Alftafjordur magmatic system (Eastern Iceland).
Analogue models have been used to test the effect of repeated dike intrusions at the
surface, characterizing the geometry and the kinematics of dike-induced structures. In order
to quantify and reconstruct the temporal evolution of the surface deformation, laser-scanner
and Particle Image Velocimetry (PIV) techniques have been applied to the models.
Field analysis show that at the surface, the eruptive fissures are bounded by normal faults
forming a graben, thus suggesting a clear relationship between diking and surface
deformation. Grabens, normal faults and extension fractures also characterize the surface
of wider fissure zones. The faults usually exhibit an open structure, with locally tilted hanging
wall and possible contraction at its base. However, the study of the fissure zones roots
reveals that at depth the extension is accommodated by dikes, with almost no faulting.
Analogue models results show that a graben forms above a dike complex gradually
thickening at depth, whose structures geometry and kinematic depend on ratio between the
intrusion depth and its cumulative thickness. This suggests, again, that diking may play a
fundamental role in rift formation.
1
Models and nature share several common features (i.e., graben with downward
propagating normal faults, contraction at the base of the tilted hanging wall, subsidence
above the dike complex and uplift to its sides), suggesting that most deformation along
divergent plate boundaries may be acquired through repeated dike injection, requiring no
direct tectonic contribution. However, the extension due to the regional tectonics remains
still important for the long-term evolution of a divergent plate boundary: on one hand, it
provides the required conditions for the rise and focusing of magma along the axial zone of
the margin, on the other, it enhances the fault activity in more distal areas from the axial
zones and during the inter-rifting periods.
Riassunto
L’interesse rispetto al ruolo del magma nella separazione delle placche lungo i margini
divergenti si è recentemente rinvigorito. Tuttavia, il meccanismo preciso con il quale il
magma influenza la geometria, la cinematica e l’evoluzione temporale di un rift è ancora
poco chiaro, soprattutto nel lungo termine. Inoltre, resta da definire quanto e come la
tettonica regionale, attraverso il meccanismo di plate pull, influisce nella formazione delle
strutture di rift lungo tali margini. Questo lavoro ha lo scopo di far luce su tali problematiche,
attraverso analisi di terreno e modellazione analogica.
L’analisi di terreno comprende lo studio della deformazione in superficie lungo diverse
porzioni fissurali dei sistemi magmatici della Neovolcanic Zone in Islanda e del Main
Ethiopian Rift, in particolare focalizzandosi su: 1) singole fessure eruttive (Laki, Eldgjá e
Bardarbunga in Islanda) o strette zone fissurali che ospitano recenti fessure eruttive
(Sveinagjá e Sveinar in Islanda), 2) ampie zone fissurali dove sono avvenuti numerosi
episodi di rifting (Krafla, Vogar e Thingvellir in Islanda e Fantale in Etiopia). In queste
aree, la geometria e la cinematica di faglie e fratture estensionali sono state caratterizzate
in maniera sistematica, includendo l’analisi della struttura delle terminazioni laterali delle
faglie come possibile indicatore del loro senso di propagazione.
Inoltre, tali analisi sono state integrate con lo studio delle radici (paleo-profondità di circa
1.5 km) del sistema magmatico fossile di Alftafjordur (Islanda orientale).
Attraverso i modelli analogici, invece, si è testato l’effetto dell’intrusione ripetuta di dicchi
sulla superficie, caratterizzando la geometria e la cinematica delle strutture da essi indotte.
Per quantificare la deformazione e ricostruirne l’evoluzione nel tempo sono state applicate
ai modelli lo scansionatore laser e la tecnica di Particle Image Velocimetry (PIV).
2
Le analisi di terreno mostrano che in superficie le fessure eruttive sono bordate da faglie
normali che formano un graben, suggerendo così una chiara relazione tra dicchi e
deformazione in superficie. Graben, faglie normali e fratture estensionali caratterizzano
anche la superficie delle zone fissurali più ampie e mature dove più episodi di rifting sono
avvenuti. Le faglie mostrano generalmente una struttura aperta, con il tetto della faglia
localmente basculato e con la possibile presenza di contrazione alla sua base. Tuttavia, lo
studio delle radici dei rift rivela che in profondità l’estensione è accomodata quasi
esclusivamente da dicchi, con una ridotta fagliazione.
I risultati dei modelli analogici mostrano che al di sopra di un complesso di dicchi che si
inspessisce gradualmente, si ha la formazione di un graben, le cui geometria e cinematica
delle strutture interne dipendono dal rapporto tra la profondità di intrusione ed il suo
spessore cumulato. Questo suggerisce, ancora una volta, che l’intrusione di dicchi può
avere un effetto diretto sulla superficie e quindi un ruolo fondamentale nella formazione dei
rift.
Modelli e natura mostrano molte caratteristiche comuni (ad es.: graben con faglie normali
che si propagano verso il basso, contrazione alla base del tetto di faglia basculato,
subsidenza al di sopra del complesso di dicchi e sollevamento ai suoi lati), suggerendo che
la maggior parte della deformazione lungo i margini divergenti di placca può essere acquisita
attraverso ripetute iniezioni di dicchi, non richiedendo un contributo diretto della tettonica
regionale. L’estensione dovuta alla tettonica regionale rimane tuttavia importante per
l’evoluzione a lungo termine di un margine divergente, sia perché essa fornisce le condizioni
necessarie per la risalita e la focalizzazione del magma lungo la zona assiale del margine,
sia perché promuove l’attività di fagliazione nelle aree più distali dalle zone assiali e durante
i periodi di inter-rifting.
3
1. Introduction
Most seismic and volcanic activity on Earth is present along the divergent plate
boundaries. A peculiar aspect concerning these boundaries provides a definition of the
mechanism which characterizes the separation process of plates. According to traditional
theories, the phenomenon of separation occurs through tectonic processes at large scale
(i.e., plates pull), linked to extensional faults’ activity. However, the recent rifting episodes
which occurred in the last decades, highlighted the crucial role the magma can play in
shaping the divergent boundaries. These episodes occurred in the oceanic, as well as in the
continental and transitional crusts, allowing to undertake an accurate research on surface
deformation directly inducted by magma emplacement, through dikes intrusion. In particular,
geodetic measurements (mainly InSAR and GPS) show that during intrusive episodes a
characteristic deformation occurs at the surface. It is composed of a narrow (few km large)
strip above the dike/dikes that is subject to subsidence, while an uplift occurs at both sides;
both subsidence and uplift may reach up to few meters [i.e., Rubin, 1992; Sigmundsson,
2006; Wright et al., 2006]. Moreover, perpendicularly to the dike, a relevant horizontal
opening occurs, up to several meters [i.e., Wright et al., 2006; Biggs et al., 2009]. Such a
deformation is also obtained by means of numerical and analytical modeling of a dike that
opens at depth [i.e. Dieterich and Decker, 1975; Mastin and Pollard, 1988; Rubin and
Pollard, 1988; Rubin, 1992; Dvorak and Dzurisin, 1997]. Observations in the field prove that,
together with these kind of episodes or intrusive events, several fractures generate at the
surface. This phenomenon is also associated with the formation or re-activation of normal
faults (with metric throws), generating a graben centered and parallel respect to the
dike/dikes [i.e., Bjornsson, 1977; Acocella and Neri, 2003; Rowland et al., 2007; Pallister et
al., 2010]. Moreover, the generation of these grabens with inward dipping normal faults,
related to dike propagations, has been also demonstrated through analytical, numerical and
analogue models [Mastin and Pollard, 1988; Rubin and Pollard; 1988; Rubin, 1992;
Gudmundsson, 2003; Gudmundsson, 2005].
Thus, all these observations have newly enhanced the interest towards divergent plate
boundaries, reopening the whole question on traditional theories regarding plates’
separation process.
In particular, it has been observed that the opening during rifting episodes can reach
some meters of width across a short time period (less than few tens of years). These rates
are definitely abundant in comparison with regional extension rate, which is at the size of
mm/yr normally. Such evidences clearly show that in the short time period magma plays a
4
primary role in generating rift structures and extensional processes in general, by means of
episodic and rapid dike intrusions. On the contrary, regional tectonic seems to play a very
restricted role [Sigmundsson, 2006; Wright et al., 2006; Ayele et al., 2007; Ebinger et al.,
2010]. However, data directly concerning the effect of the magma intrusion along divergent
boundaries is limited to the last decades. Therefore, the relative influence of magma and
tectonics in controlling the whole extensional processes and the formation of rift structures
along divergent boundaries is still under debate, especially on a longer term view.
1.2 Methods
In order to better investigate on the arguments previously introduced, two different
methods have been applied: field survey and analogue modeling.
1.2.1 Field surveys
Rift generation along divergent plate boundaries occurs both in continental crust (where the
separation process among plates is at its first stage) and oceanic crust (where the process
is at its advanced level). One of the most representative divergent plate boundary in the
continental crust is the East African Rift System (EARS), which develops from Afar
depression to Monzambique, separating the Nubia plate (to the West) from the Somalia
plate (to the East) with an extension rate of 4-7 mm/yr [DeMets et al., 2010, and references
therein]. In particular, in this work, a field survey has been carried out in the portion of the
EARS coinciding with the Main Ethiopian Rift (MER), representing an excellent site for
investigating on the present structure of a continental rift.
Divergent plate boundaries at their mature stage are mainly associated with oceanic ridges.
In particular, this work focuses on the Middle Atlantic Ridge (MAR). The most accessible
portion of MAR geographically corresponds to Iceland, where the ridge arises at the surface.
Here, the active portion of the ridge is several tens of kilometres wide and separates the
North American plate (to the West) from the Eurasian one (to the Est), with a rate of ~2 cm/yr
[DeMets et al., 2010, and references therein].
The axial active zones of the divergent boundaries are characterized by the presence of
several magmatic systems (20-100 km long), which are their fundamental elements.
Magmatic systems are composed of a polygenic central volcano, generally with a summit
caldera, and a relevant fissural zone, characterized by open normal faults, extensional
fractures, eruptive fissures and monogenic aligned cones [Fig. 1; Einarsson e
Saemundsson, 1987; Gudmundsson, 1987, 1995, 2000]. At depth, they are mainly formed
by two types of dike swarms: local, limited to the area involving the central volcano, and
5
regional, which spread outside the area of the volcano and are parallel to the fissural zone
of the system. Local dikes are characterized by circular shapes, depending on the stress of
the volcanic structure and its magmatic chamber [Gudmundsson, 1995; Tibaldi, 2008], while
regional dikes concentrate mostly in swarms 5-10 km wide and ~ 50 km long, whose
direction is mainly dependent on regional stress [i.e., Walker, 1959; 1960; Gudmundsson,
1995; Paquet et al., 2007].
The field surveys carried out in this work aim at analyzing the fissural portions of different
magmatic systems distinguishing two main domains. The first is related to simple eruptive
fissures with a narrow deformation zone (hundreds of m wide, as Eldgjá, Laki, Sveinagja,
Sveinar and Bardarbunga, in Iceland), related to one or very few diking events. Since the
eruptive fissures are generated by feeder dikes, the study of geometry and kinematics of
deformation pattern associated with them is focused on understanding the direct impact of
the magma on the surface deformation, with a negligible contribute of the regional tectonics.
In particular, the eruptive fissures of Edgjà (933-941 A.C.) and Lakagigar, (1783-1784) are
the two greatest rifting episodes in Iceland history, with an eruptive volume of 18.3 km3 and
14.7 km3 [Thordarson et al., 1993; Thordarson, et al., 2001]. The Svenagià and Sveinar
grabens have been recently reactivated by a rifting episode (1875), associated with a
relevant caldera collapse of the Askja central volcano [i.e., Gudmundsson and Backstrom,
1991] and are characterized by a narrow deformed zone (<1.5 km). Indeed, the most recent
Bardarbunga rifting episode started back in August 2014 and still on-going: it represents a
unique opportunity for investigating on deformation related to active dike intrusions. This
event is also one of the biggest occurred in Iceland across the last 100 years and may be
directly compared with the oldest Edgjá and Lakagigar fissures.
The second domain concerns the geometric and kinematics investigation of grabens,
normal faults and extensional fractures, characterizing more complex and mature rift
segment, hosting several eruptive fissures within a broader deformed area (several km wide,
as Krafla, Vogar, Thingvellir in Iceland and Fantale in Ethiopia) and in which magma and
tectonic induced structures may be mixed. In particular, the fissural areas of the Krafla
magmatic system is well known for a recent rifting episode, during which ~20 dike intrusions
occurred in ~20 years (1975-1984) [i.e. Opheim and Gudmundsson, 1989 and references
therein].
For further details, Chapter n. 2 provides more information on geology, structure and
eruptive activity of each single area.
6
The origin of the faults and the way they develop, along such divergent plate boundaries
are still under debate. Actually, according to several authors, the ways they can develop are
several:
1) from the generation of extensional fractures developing at surface through a regional
stress field, then propagating downwards. When they reach a critical depth dependent on
lithostatic pressure they become faults [Gudmundsson, 1992];
2) from the coalescence of fractures inside inclined lava flows [Gudmundsson, 1992];
3) at depth, from the upper dikes’ tips, then propagating upwards [Grant and Kattenhorn,
2004; Tentler, 2005];
4) through subsidence caused by deflection, inducted by sills placed under the rifts [Tentler
e Temperley, 2007].
Therefore, understanding the direction of faults propagation is of crucial importance, since
it can provide unique information on their origin. In fact, generally, faults propagating upward
can be directly related to a dike emplacement, while downward propagating faults are mainly
due to the effect of tectonic or other minor reasons.
In this work, in order to understand the sense of fault propagation in the field, both along
the eruptive fissures and mature rift segments, the structure of the lateral tip of the fault at
the surface (where the faults displacement becomes zero) has been considered. The lateral
termination of the normal faults is the fault’s youngest portion, being at its incipient formation
stage. The fault tips have been classified according the two following schematic structures
[Cartwright and Mansfield, 1997]:
a) the fault tip displays a flat footwall and a flat hanging wall, interrupted by extension
fractures departing from the main fault; this suggests that the fault formed from the
coalescence of open fractures and propagated downward (OPF type);
b) the fault tip displays a flat footwall and a tilted hanging wall resembling an overall
monocline; this structure suggests a fault propagation folding mechanism associated
with an upward or lateral propagation of the fault (FPF type).
This analysis allows to better understand the geometry and kinematics of the rift
structures that have been investigated by creating detailed field maps, in which the values
of strikes, throws, openings, amount of hanging walls and footwalls tilts have been reported,
as well as the tips typologies of some faults within the magmatic systems.
Last, but not least, is the complementary analysis of the rifts structure at depth. Iceland
represent a unique setting for investigating on this subject, since at the sides (~80 km) of
the active axial portion of the Neovolcanic Zone the magmatic systems’ roots arise.
7
For instance, in Eastern Iceland, thanks to erosion and regional uplift, the outcropping
rocks correspond to a paleo depth of 1.5 Km [Walker, 1960]. By studying these magmatic
systems portions, it is possible to define the relationship between dikes and faults at such
depth, providing an estimation of extension percentage, respectively linked to the one or the
second. In order to achieve this aim, a field survey has been carried out by realizing some
transects perpendicular to the fissural zone of Alftafjordur magmatic system, where data of
directions and thickness of dikes have been reported, together with the geometry and
kinematics data of faults.
Fig. 1: Block diagram (a) and section view (b) of the main structural elements of a volcanic
system. The numbers in b indicate the growth sequence of the vertical dyke from a magma
reservoir. C = magma chamber; DS = dike swarm; CV = central volcano; EF = eruptive
fissure; FS = fissure swarm zone [from Thordarson and Larsen, 2007 and references
therein].
1.2.2 Analogue models
Although the studies of recent rifting episodes through geodesy and field surveys are
crucial for understanding the genesis and development of the structures along divergent
plate boundaries, their application is mostly valid if referred to a limited time period. In fact,
they usually focus on deformation and development of a single rifting event or episode
8
lasting some decades at maximum (i.e., ~20 years at Krafla 1975-1984; ~5 years at
Dabbahu 2005-2010, etc.) and related to some tens of dike intrusions (i.e., 20 at Krafla; 13
at Dabbahu) [i.e., Buck et al., 2006; Ebinger et al., 2010]. Moreover, also the geodetic
analysis of the deformation during the inter or post rifting episodes is generally limited to
about a time period of some tens [i.e. Hofton and Fougler, 1996]. However, deformation
along the rifts - especially in the mature ones (i.e., Krafla, Thingvellir, Vogar) - regards
several rifting episodes occurring across thousands of years, with some hundreds dike
intrusions. This is also evident at the roots of inactive rifts, in the E coast of Iceland, where
their fissural portions at 1-2 km depth are composed of one hundred parallel dikes at least
[Walker, 1959; 1960; Gudmundsson, 1984; Gudmundsson, 1990; Paquet et al., 2007].
Same as for geodesy, field investigations on eruptive fissures allows us to highlight the
deformation related to one single episode or event, thus providing again a short time period
(<102 years) view of the deformation. On the contrary, field surveys on mature rifts allow us
to carry out an analysis of deformation concerning a longer time period (>102 years).
However, in these areas, the results of magmatic and tectonic activity are definitely mixed,
interconnected and interdependent. In such a situation, it is complicated to physically
distinguish magmatic from tectonic induced structures. Thus, understanding the structures
evolution across years becomes very hard.
To sum up, while geodesy can provide “dynamic” information of rifting episodes, these
refer to a quite limited time period (< 102 years). On the contrary, while field analysis
represents an opportunity for studying both short and long term time period deformation (>
102 years), it provides with a nearly “static picture” of magmatic and tectonic deformation
cumulated across time.
The analogue modeling technique has been used to overcome these issues and try to
fully understand the rifting processes. Through this, it has been possible to realize a scaled
model of a precise natural process and studying its development from the initial to the final
stage. In addition, through analogue modeling it is possible to control and vary the different
parameters controlling that natural process. In this way, it becomes clear to what extent that
parameter may influence the whole process.
On the assumption that magma plays a fundamental role in defining the topography and
development of rift structures in the short term - as highlighted by the studies of recent rifting
episodes – the analogue models realized in this work aim at testing the effects of repeated
dike intrusions at the surface, achieving also a longer term perspective. This can help in
knowing if and how the surface deformation observed along the divergent plate boundaries
9
is linked to magma emplacement through dikes, consequently better understanding which
role tectonic could play in this.
Despite similar analysis of surface deformation induced by dike injections have been
previously carried out, they sometimes show an unrealistic deformation [i.e., Abdelmalak et
al., 2012] or limitations, due to the lack of advanced techniques in monitoring the models
[Mastin and Pollard, 1988] (more details are also reported in Chapter 3). Using modern
techniques aiming to overcome that limitation, thanks to analogue modeling, we may attempt
to describe how rifting structures generate and develop. This allows us to get a no longer
“static” vision of the rifting process, but a “dinamic” one - conceived in one whole rift livingprocess.
The final comparison among analogue modeling results, numerical models, field
observations and geodesy - also present in literature - has been crucial in deeply
understanding the overall rift structures generation and development along divergent plate
boundaries.
10
Chapter 2
(Under review in Tectonics)
Surface deformation along the divergent plate boundaries of Iceland and Ethiopia:
understanding magmatic vs. tectonic processes
Trippanera D.1*, Acocella V.1, Ruch J.1°, Abebe B.2
1 Roma
Tre University, Rome, Italy
°Now at King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia.
2University
of Addis Ababa, Addis Ababa, Ethiopia
* Corresponding author: [email protected]
Abstract
Recent studies suggest that diking controls the structure and development of divergent
plate boundaries; however, the exact role of diking on surface deformation is poorly known.
Here we analyze the deformation along oceanic (Iceland) and, subordinately, continental
(Main Ethiopian Rift) divergent plate boundaries. We focus on dike-fed eruptive fissures and
rift segments with multiple fissures. Eruptive fissures are systematically accompanied by
grabens with open normal faults with vertical displacement reaching up to ~10 m; this
structure is found, with asymmetries and repetitions, also within the rift segments. Most
faults along eruptive fissures terminate as open fractures on a flat surface, suggesting
downward propagating after surface nucleation; this is observed also in fastly extending rift
segments (>1 cm/yr). Conversely, in slowly extending rift segments (<1 cm/yr) normal faults
predominantly terminate as open fractures on a monocline, suggesting upward propagation
after deep nucleation. Recent experimental and numerical models show that dike-induced
normal faults propagate downward from the surface, suggesting that any upward
propagation of normal faults largely results from fault reactivation due to dike emplacement
after burial by volcanic deposits. The overall deformation pattern of eruptive fissures and rift
segments may be qualitatively explained only by dike emplacement. Simple calculations
also suggest that all the fault slip in the rift segments may result from diking. Therefore, in a
magmatic rift, the regional tectonic stress may rarely be high enough to be released through
11
regional faulting, suggesting that regional tectonics has negligible direct impact compared
to diking in shaping the studied plate boundaries.
Key words: divergent plate boundaries, normal faults, extension fractures, dikes, fault tip
1. Introduction
Divergent plate boundaries are characterized by the drifting of the interacting plates from
each other and may be found on oceanic, transitional or continental lithosphere. Divergent
boundaries consist of several magmatic segments where tectonic and volcanic activity
focus. Each segment is composed of a dominant polygenic volcano and one or more rift
segments with monogenic fissural volcanism [Gudmundsson, 1995a, Ebinger and Casey,
2001; Acocella, 2014]. The active tectonic features along the rift axis at the surface consist
of iso-oriented eruptive fissures, extension fractures and normal faults. Eruptive fissures
consist of the alignment of dike-fed monogenetic vents, and may reach along-rift lengths of
several tens of km. Extension fractures are ~102 m long and a very few m wide. Normal
faults, often forming grabens or half-grabens, are ~103 m long; the shallowest part of these
faults along the rift axis commonly displays a dilational (tensile) component, of a very few
m, between the foot-wall and the hanging-wall; the latter is often tilted outward; such a fault
structure, uncommon in sedimentary sequences, may result from the strength of the basaltic
lavas and welded tuffs in the rift axis (Hardy, 2013). The lateral termination of the open
normal faults is characterized by the progressive decrease in the vertical displacement,
which reaches zero and leaves only the open fracture separating the two portions of the fault
[Gudmundsson, 1992; Angelier et al., 1997; Acocella et al., 2003; Acocella, 2014, and
references therein]. The development of the normal faults and extension fractures along
divergent plate boundaries may directly result from regional extension. In this case, plate
pull develops tension fractures at the surface; when growing, these reach a critical depth
where the growing lithostatic stress imposes a shear component, developing normal faults
[Gudmundsson and Backstrom 1991; Forslund and Gudmundsson, 1992; Gudmundsson,
1992; Acocella et al., 2003]. According to this model, the normal faults start to nucleate
immediately below the surface, at a very few hundreds of m of depth, and propagate
downward. Modelling, geodetic and geophysical studies from the last decade have been
highlighting the importance of magma, in the form of dikes, in separating the plates and
controlling the structure and development of divergent plate boundaries, particularly when
activating the normal faults [e.g. Buck, 2006; Sigmundsson, 2006; Wright et al., 2006].
12
According to these models, dike-induced faults nucleate at depth, in correspondence with
the upper dike tip, and then propagate upward. At the surface, the upward propagating
normal faults first generate a monocline due to the bending above their upper tip; this is
subsequently displaced by the propagated fault, accordingly to established fault propagation
folding models: the final result is a faulted fold, or monocline [Grant and Kattenhorn, 2004;
Tentler, 2005; Rowland et al., 2007]. The dikes inferred to be responsible for plate spreading
and for shaping the shallow structure of divergent plate boundaries are usually tens of km
long, several m wide and propagate within a few days or weeks; if reaching the surface, they
feed the eruptive fissures. Multiple dikes may be injected along the same rift portion within
the same rifting episode, which may last up to several years, with a recurrence interval of a
few centuries [Ebinger et al., 2010, and references therein]. Important rifting episodes or
events include Krafla 1975-1984 (Iceland), Asal 1978 (Afar), Dallol 2004 (Afar), Dabbahu
2005-2009 (Afar), Natron 2007 (Tanzania), Harrat Lunayyir 2009 (Saudi Arabia),
Bardarbunga 2014 (Iceland). These episodes have been geodetically and seismically
constrained, showing that the deformation pattern at the surface commonly results in a
graben structure [e.g. Bjornsson, 1977; Ruegg, 1979; Tarantola et al., 1980; Tryggvason,
1984; Rubin and Pollard, 1988; Wright et al., 2006; Biggs et al., 2009; Pallister et al., 2010;
Nobile et al., 2012; Gudmundsson et al., 2014]. However, direct field investigations on the
structural features developed at the surface have been limited [i.e., Sigurdsson, 1980;
Rowland et al., 2007; Pallister et al., 2012]: despite the evidence that single normal faults
may be effectively reactivated with slip of few meters during dike emplacement, as at
Dabbahu [Rowland et al., 2007], little is known on the effect of diking at the surface, also on
the longer-term (hundreds of years or more) rift structure. In addition, the relative importance
of the magmatic and tectonic contributes on normal faulting has been poorly investigated on
the field, so that it is still uncertain how much of the structure of a given rift segment may be
ascribed to diking and/or regional processes.
In order to better define a) the structural features of divergent plate boundaries, b) their
relationships to diking and c) the possible contribute of magmatic and tectonic processes,
we performed a field survey at selected spots along the oceanic and, subordinately,
continental divergent plate boundaries of Iceland and Ethiopia, respectively. We first focused
on single eruptive fissures, to define simple deformation patterns related to diking, and then
on more complex rift segments, also trying to relate these to the observations along the
eruptive fissures. Moreover, we focused our attention to the lateral termination of the normal
faults, in order to suggest a possible sense or propagation (upward or downward) and relate
13
this to tectonic or magmatic processes (see methods section below). Our results suggest
that the deformation pattern observed along divergent plate boundaries may be entirely
explained by shallow dike emplacement.
2. Tectonic setting
Iceland marks the subaerial portion of the Mid-Atlantic Ridge, separating the North
American and Eurasian plates, spreading at ~2 cm/yr [Fig. 1a; e.g. DeMets et al., 2010, and
references therein]. Two main neovolcanic zones are found: to the west, is the ReykjanesLangjokull Volcanic Zone (RLVZ), including the obliquely spreading Reykjanes Peninsula
Rift, to the SW, and the Western Volcanic Zone (WVZ), to the NE, comprising the HengillLangjokull systems. The WVZ has been the main locus for crustal spreading in South
Iceland for the last 6-7 Ma. To the East, the eastern branch of the neovolcanic zone
comprises the north Volcanic Zone (NVZ), active for 6-7 Ma, to the north of the Vatnajokull
icecap, and the Eastern Volcanic Zone (EVZ), active for 2-3 Ma, to the south of it (Fig. 1a).
The EVZ is propagating southwards, and significant crustal spreading in the EVZ has only
developed north of the Torfajokull volcano. The EVZ and WVZ have an overlapping
configuration in southern Iceland. Here the spreading rate of the EVZ decreases southward,
while it increases linearly southwards along the WVZ, maintaining an overall constant total
velocity [LaFemina et al., 2005; Perlt et al., 2008]. The area enclosed by the EVZ and the
WVZ forms the South Iceland Seismic Zone, producing >M7 earthquakes, whose tectonic
frame is consistent with the early stages of development of a transform-like feature
[Gudmundsson and Brynjolfsson, 1993; Gudmundsson, 1995b; Gudmundsson, 2007;
Angelier et al., 2008; Clifton and Kattenhorn, 2006]. Volcanism focuses along the rift axis, in
40-150 km long and 5-20 km wide magmatic systems, each with a dominant volcano, with
silicic rocks, geothermal areas and often summit calderas, accompanied by tension
fractures, normal faults and volcanic fissures [Gudmundsson, 1987b, 1995b, 2000]. Usually,
each magmatic system is activated by along-rift diking episodes with a frequency of few
hundreds of years [Sigmundsson, 2006]. The eroded Late Tertiary and Pleistocene lava pile
at the eastern and western sides of Iceland allows also studying the deeper portion of extinct
magmatic systems, represented by local sheet swarms and regional dike swarms. The sheet
swarms are confined to extinct volcanoes and are usually 45° to 65° inward dipping, circular
to elliptical, several km in radius. Most sheet swarms are associated with large plutons at 12 km of paleodepth. The regional dikes occur outside the central volcanoes in 50 km long
and 5-10 km wide swarms; these dikes are usually subparallel and vertical and thicker than
14
the sheets. At depth of 1-2 km, these segments are composed of hundreds of parallel dikes
having a mean thickness of 3-4 m; in these <10 km wide and tens of km long swarms, the
dikes accommodate ~5-6% of crustal dilation, whereas very few and small faults are
observed [Walker, 1958; 1960; Gudmundsson, 1984; Helgason and Zentilli, 1985;
Gudmundsson, 1995a; Paquet et al., 2007].
The Main Ethiopian Rift (MER) separates a portion of the Nubia and Somalia plates at 47 mm/yr [Fig. 1b; e.g. DeMets et al., 2010, and references therein]. The NE-SW trending
MER developed in two main stages controlled by oblique opening [Corti, 2008; 2009]. An
early (Mio-Pliocene) continental rifting stage activated the NE-SW striking rift border faults,
creating up to 5 km of subsidence, and was associated with diffuse magmatic activity, also
off-axis. In this period, the rift development was primarily controlled by the lithospheric
structure. In the later, Quaternary stage magmatic processes become dominant, focusing
the tectonic, seismic and magmatic activity in segments along the rift axis, or “Wonji Fault
Belt” (WFB) and deactivating the main border faults [Woldegabriel et al., 1990; Ebinger,
2005; Keir et al., 2006; Keranen and Klemperer, 2008; Corti, 2009]. The WFB consists of
several NNE-SSW striking right-stepping en-echelon magmatic segments, ~20 km wide
and ~60 km long, where tectonic activity focuses [Mohr, 1967, 1987], developing m-wide
active extension fractures and open normal faults [Acocella and Korme, 2002; Acocella et
al., 2003]. These magmatically supplied WFB segments accommodate >80% of the strain
across the rift at depths <10 km and are considered as the loci responsible for the opening
of the rift, similarly to what observed along oceanic ridges; the magma accumulated in the
main reservoir of the segment may be intruded laterally through dikes, feeding monogenic
fissural volcanism, or be erupted in the volcano above [Ebinger and Casey, 2001; Casey
et al., 2006]. Seismicity is also focused along the magmatic segments, providing a deeper
evidence for faulting or diking [Keir et al., 2006]. The shallowest crustal portion beneath the
MER, between 7-25 km, is characterized by high P-wave velocities coinciding with gravity
maxima, suggesting pervasive cooled magmatic intrusions, or dikes, that accommodate a
large part of extensional strain at depth; these intrusions are segmented, ~20 km wide and
~50 km long bodies in a right stepping en-echelon pattern, approximately mimicking surface
segmentation of the Quaternary magmatic centres [Keranen et al., 2004; Mickus et al.,
2007; Daly et al., 2008]. In the upper mantle, shear-wave splitting suggets a strong
component of melt-induced anisotropy, supporting a magma-assisted rifting model in
continental lithosphere [Kendall et al., 2005]. GPS measurements between 1992-2010
highlight an extension of 8 mm/yr, distributed across the northern MER and focused in the
15
southern MER [Kogan et al., 2012]. The extension rate between 1992-2010 is consistent
with velocity field models from GPS stations, suggesting that the Nubia and Somalia plates
on the flanks of the MER separate at ~7 mm/yr [Fernandes et al., 2004]. These geodetic
estimates are at least one order of magnitude larger than the geological ones obtained on
the field on a longer term (~7 ka), suggesting that, along the rift axis, most of extension may
be accommodated at depth or that spreading does not occur at a uniform rate [Williams et
al., 2004]. The last eruption within the MER occurred at Fantale, probably in 1820 [Siebert
et al., 2010].
More details on the structural features of portions of Iceland and the MER are given at
the beginning of each studied area, in the Result section.
3. Methods
The field survey focused on the surface deformation along selected portions of oceanic
and, subordinately, continental divergent plate boundaries in Iceland and Ethiopia,
respectively. We distinguish between simple eruptive fissures with a narrow deformation
zone (hundreds of m wide, as Eldgjá, Laki, Sveinagja, Sveinar and Bardarbunga, in Iceland),
that are related to few diking episodes, and more complex and mature rift segments hosting
several eruptive fissures within a broader deformed area (km wide, as Krafla, Vogar,
Thingvellir in Iceland and Fantale in Ethiopia).
At each study area, we first considered the overall structure of the eruptive fissure or the
rift segment, as given by the dominant extension fractures and fault systems, which have
been geometrically and kinematically characterized in our study. In particular, we
systematically measured the opening of the extension fractures and the faults, as well as
the vertical throw and hanging wall tilt of the normal faults, if present (Fig. 2 a, b). We then
tried to evaluate the sense of fault propagation (i.e., upward or downward). Understanding
the propagation direction of the faults may in principle allow us to address whether a fault
may develop from regional plate pull (downward propagation, accordingly to section 1) or
from an intruded dike (upward propagation, accordingly to section 1). In order to try to
reconstruct the sense of fault propagation on the field, we considered the structure of the
lateral tip of the fault at the surface. The lateral termination of the normal faults is the least
portion of the fault to be propagated and thus formed at the surface, especially if the fault,
as in our studied cases, is relatively small, only a very few km long. Therefore, the lateral
fault tip can provide important information on the mode of propagation of the fault. The
structure of the lateral tip of the normal faults can be schematized in two main types
16
[Cartwright and Mansfield, 1998]. a) The fault tip displays a flat footwall and a flat hanging
wall, interrupted by extension fractures departing from the main fault (open fracture or OPF
type; Fig. 2c); this suggests that the fault formed from the coalescence of open fractures
and propagated downward. b) The fault tip displays a flat footwall and an outward tilted
hanging wall interrupted by an extension fracture, resembling an overall fractured
monocline; this structure suggests a fault propagation folding mechanism associated with
an upward or lateral propagation of the fault (FPF type; Fig. 2d). An important part of our
field survey consisted of a detailed mapping of the lateral tips of the normal faults, where
the faults displacement becomes zero, in order to identify the type of lateral termination: in
particular, we distinguished between a flat or tilted hanging wall to the side of the extension
fracture at the fault tip. Most fault tips were measured in areas of general flat topography. In
a few cases, the normal fault tips were on a gentle slope, extending both on the foot wall
and the hanging wall over a distance much larger (hundreds of m or more) than that which
may be related to the activity of a fault propagation fold (less than several tens of meters).
In these cases, the tilt due to this broader slope was distinguished from that eventually
present in the hanging wall.
4. Results
Here we present the results about the deformation pattern observed along five eruptive
fissures in Iceland (section 4.1) and four rift segments in Iceland and the MER (section 4.2).
All the presented data are original, except for a part of those from the areas of Vogar and
Thingvellir, Iceland, presented in Norini et al., 2009.
4.1 Eruptive fissures
4.1.1 Lakagigar
The 27 km long Lakagigar eruptive fissure is located in SE Iceland, in an area undergoing
an overall extension rate of 10 mm/yr (Table 1) [Perlt et al., 2008]. The fissure belongs to
the Grimsvotn volcanic system (East Volcanic Zone) and formed in 1783-1784, producing
one of the largest historical basaltic lava flows (~15 km 3); the feeder dike(s) propagated
towards NE, forming ten en-echelon fissures [Thordarson and Self, 1993]. Our field survey
focused on the ~4 km long central portion of the fissure in a nearly flat area, except for the
~200 m high and older Laki hill (Fig. 3). The N45°±5° striking fissure (Fig. 3d) is mostly
centered within a graben bordered by N55°±5° trending normal faults with mean dip of 75°
(Fig. 3a, b, c). In the flat areas north and south of Laki, the graben is 150-450 m wide. In
17
correspondence to Laki, the eruptive fissure becomes interrupted, but the border faults climb
the hill, enlarging up to ~700-800 m distant; the deformation is accompanied by diffuse
fracturing, given by additional normal faults and also extension fractures.
The mean vertical throw of the border normal faults along the fissure is ~3 m, locally reaching
10 m ~1.5 km north of Laki (Fig. 3e; Table 1). In addition, a mean tensile component of ~1.4
m is found between the footwall and the hanging wall of the faults (Fig. 3f). The hanging wall
and the footwall are usually outward tilted of 5°-10° (Fig. 3c). The deformation on the border
faults may be locally partitioned into several minor parallel normal faults, resulting in a
widespread and less defined border (Section A-A’; Fig. 3g). The eruptive fissure, usually
lying in the middle of the graben, may also locally coincide with the border fault (Section BB’; Fig. 3g). This suggests that the border fault formed before the dike reached the surface
and was subsequently reactivated by the dike to rise. Finally, the lateral termination of 8
normal faults bordering the graben shows open fracture structures (OPF type), suggesting
that the normal faults nucleated at the surface and propagated downward (Table 1).
4.1.2 Eldgjá
The 75 km long and N45°E trending Eldgjá fissure belongs to the Katla Magmatic System
(Iceland), undergoing an overall local extension rate is 9 mm/yr (Table 1) [Perlt et al., 2008].
The fissure results from at least two distinct rifting episodes, the last one in 933-941 AD
[Hammer, 1984; Zielinsky et al., 1995], producing a total erupted volume of ~20 km3. The
fissure can be divided in 3 main segments: a SW, a Central and a NE segments [Thordarson
et al., 2001 and references therein]. Our field survey focused along 1 km on the SW and
Central segments and along 4.5 km on the NE segment. All the segments consist of grabens
and eruptive fissures. The graben is generally defined by discontinuous (200-300 m long)
inward dipping normal faults and, at times, by minor parallel faults, also defining minor
nested grabens (Fig. 4a). The fissure, when located within the graben, is partly bounded by
the normal faults: otherwise, it may coincide with the border faults (Fig. 4b). In the
investigated area, the N35°±5° to N65°±5° oriented eruptive fissure mainly crops out in the
Central segment. The mean width of the graben containing the fissure is 230 m in the SW,
160 m in the Central and enlarges to 650 m in the NE segments (Fig. 4a, b, c; Table 1).
However, the overall mean strike and throw of the border faults, of N40°±5° and 5±0.5 m
respectively, remain consistent throughout the segments (Fig. 4d, e; Table 1). In each
segment, the highest fault throw of the border faults reaches 10±2 m (Table 1). The mean
fault dip, estimated in four well-exposed sections, is 78°±4° (i.e. Fig. 4h). Locally, the footwall
18
and the hanging wall of the border faults are separated by a tensile fracture 0.5 to 4 m wide
(mean value of 1.5±0.2 m; Fig. 4f; Table 1). In correspondence with the SW and Central
segments, the topography shows a regional tilt of ~10° toward SE, observed also on the
footwall and hanging wall of the border faults (Fig. 4a). Along these two segments, the
hanging wall tilt of the border faults towards the graben center is rare (<20% of the
observations in the SW and central segment, respectively; Table 1). In the NE segment, the
border faults display a dilational component; the footwall is flat or 10°±5° tilted outward with
respect to the graben axis, whereas the hanging wall is commonly (76% of the observation)
tilted toward the graben axis, with mean dip of 20° (Fig. 4g; Table 1). In between the hanging
and footwall, the tensile fracture is 2±0.5 m wide (Fig. 4i). To the northern termination of the
graben, the tips of two border faults show fault propagation folds (FPF type; Fig. 4c).
4.1.3 Bardarbunga
The deformation pattern associated with the 2014 rifting episode along the Bardarbunga
volcanic system, Iceland, has been studied in October 2014 in the area of the eruptive
fissure. This case provides an active example of dike-induced deformation at the surface
and may help better understanding also the non active rifting episodes previously described.
Seismicity, GPS and InSAR data suggest the lateral propagation of a dike, with maximum
opening of 5 m close to the surface, from the Bardarbunga Central Volcano to 50 km
northeastward in 2 weeks, reaching the surface at Holuhraun, to the south of Askja [Icelandic
Met Office IMO, en.vedur.is; Gudmundsson et al., 2014 BV]. The dike caused at least 30
cm of outward horizontal displacement at the surface before erupting, in a few days; the
resulting extension rate is much higher than the mean one, of 16 mm/yr (Table 1) [Perlt et
al., 2008]. Two fissure eruptions formed in a nearly flat area on August 29 th and on August
31st. The latter, larger, is still erupting as of December 2014, producing a lava field >75 km2.
Both fissures lie close to older fissures related to the 1910 rifting episode, when a lava flow
likely buried a pre-existing graben and the associated border faults, forming a flat area.
Satellite and aerial imageries show that during the 2014 diking episode a NNE-SSW
trending, 0.8 to 1 km wide graben containing the eruptive fissure appeared at the surface
[IMO, en.verdur.is]. The graben reaches the northern edge of the Vatnajokull Ice cap, with
a length of ~6 km (Fig. 5a).
We carried out a field survey to better characterize the structure of the graben, focusing
on the two border faults immediately to the north to the August 29 th eruptive fissure, where
the graben is 0.8 km wide. Both normal faults bordering the graben show a dilational
19
component, with flat footwall and 20°±5° inward tilted hanging wall (Fig. 5b, c, d). The faults
strike N25 ±5° and have a mean vertical throw of 6±1 m. The tensile component between
the footwall and the hanging wall is ~3 m on each fault (Fig. 5d). However, the hanging wall
is locally separated in two or more blocks, with a cumulative opening reaching up to 7 m.
Minor extension fractures (mean opening 0.5±0.2 m) are also present on the footwall, in a
30-40 m wide zone parallel to the fault. These fractures, mainly striking N15°±5°, merge
along strike together and with the main fault (Fig. 5c).
4.1.4 Sveinagja
The 34 km long and ~1.5 km wide Sveinagja graben belongs to the Askja magmatic
system of Iceland, undergoing an overall extension of 19 mm/yr (Table 1) [Perlt et al., 2008].
In the northern portion of the graben, a fissural eruption related to the 1872-1875 rifting
episode produced a volume of 0.3 km3 of lava (Fig. 6a). This episode is also coeval to the
formation of the Oskjuvatn caldera, in the 70 km distant Askja central volcano [Sigurdsson
and Sparks, 1978]. In 1872-1875, the Sveinagja graben has been probably reactivated, as
it formed during older diking episodes [Gudmundsson and Backstrom, 1991; Tentler and
Mazzoli, 2005].
Our field survey focused along 4 km of the northern portion of the graben, where the
eruptive fissure is centered within (Fig. 6a). Here the graben is ~1 km wide, with border
faults striking N10°±5° (Fig. 6b). The highest measured throw of the border faults is ~10 m,
whereas the average displacement is ~4 m (Fig. 6c). These data are consistent with
previous studies [Gudmundsson and Backstrom, 1991; Tentler, 2005; Tentler and Mazzoli,
2005]. The faults are subvertical at the surface, usually showing a dilational component
between the footwall and the hanging wall. The tensile area is usually ~2 m wide, locally
reaching up to >10 m. The geometry of the faults also varies along strike. Both the footwall
and the hanging wall are usually subhorizontal, but locally (in 40% of the observations) the
hanging wall may be inward tilted up to 20° (Fig. 6d; Table 1). Each border fault system may
locally consist of a smaller graben-like structure, 20-40 m wide, with a subsided block
between the footwall and the hanging wall. The termination of 5 border faults segments, all
grading in a series of open fractures on a flat topography, suggests an OPF type of lateral
termination of the normal faults.
20
4.1.5 Sveinar
The 0.5 to 0.8 km wide and 20 km long Sveinar graben lies ~30 km north to Sveinagja.
The Sveinar graben belongs to the Fremri-Namur volcanic system and formed mainly during
the eruption of the Sveinar Lava, 6000-8000 years ago [Gudmundsson and Backstrom,
1991]. This eruption developed the NNE-SSW trending Rauduborgir-Randarholar crater
row, partially contained within the graben (Fig. 7a). The graben continued to develop during
the 1875 rifting episode at Askja-Sveinagja [Sigurdsson and Sparks, 1978; Tentler and
Mazzoli, 2005 and references therein]. As a detailed analysis of the overall graben structure
has been already provided [Tentler and Mazzoli, 2005], here we focused on the
northernmost 1.5 km of the eastern border of the graben (Fig. 7b). This is segmented in 3
main normal faults striking almost N-S (mean strike N6°). The faults are subvertical at the
surface and show a mean vertical throw of ~4 m, locally reaching 8 m, much smaller than
the maximum throw of 23 m in central portion of the graben, out of our covered area [Tentler
and Mazzoli, 2005]. Our investigated fault segment shows an open fault with dilational
component between the footwall and the hanging wall up to 2 m wide. In general, the footwall
and the hanging wall are flat, consisting of horizontal layers. Locally, the border fault is also
accompanied by antithetic structures, forming minor and narrow grabens (Fig. 7b). Along
~200 m of the northern fault segment, the hanging wall is usually tilted of 20-25° (Fig. 7c),
at times becoming subhorizontal or, conversely, reaching 65° and becoming subvertical at
the hanging wall base (Fig. 7d, e). The throw and amount of the hanging wall tilt along the
southern segment gradually decreases southwards and the fault grades into a series of open
fractures with a minor graben like structure, within an almost flat area (Profile D-D’; Fig. 7b).
This portion of the Sveinar graben highlights the coexistence of dramatic variations in the
dip of the hanging wall of central portion of a border fault within an extremely limited area, a
few hundreds of m long.
4.2 Rift segments
4.2.1 The Krafla magmatic system
The 80 km long and up to 10 km wide Krafla magmatic system of Iceland strikes NNESSW from Lake Myvatn to the north coast of Axarfjordur, undergoing a mean extension rate
of 23 mm/yr (Fig. 8a) [Bjornsson et al., 1977; Perlt et al., 2008]. In its central portion, it hosts
the ~9 km wide active Krafla caldera, where most of the 35 Holocene eruptions occurred
(Fig. 8a) [Bjornsson et al., 1977]. These eruptions included several rifting episodes, as the
21
1.5-2 ka old basaltic fissure eruptions in Kelduhverfi and spatter cones at Gjastykki, the
1724-1729 Myvatn fires and the most recent rifting episode in 1975-1984 [Opheim and
Gudmundsson, 1989 and references therein]. During the latter episode, approximately 20
successive dike intrusions occurred, producing and reactivating fractures and normal faults
[Bjornsson et al., 1977; Tryggvason, 1980; Einarsson, 1991]. Aerial imagery analysis and
GPS networks related to this episode indicate a widening in the area of the eruptive fissure
of ~9 m focused 10 to 12 km north to the caldera and decreasing northward (reaching 2-5
m) [Tryggvason, 1994; Buck et al., 2006, Hollingsworth et al., 2012]. Leveling data focused
on two diking events also indicate that each caused 0.5-1 m of subsidence above the dike
and ≥0.5 m of uplift to its sides, decreasing with distance from the dike [Rubin and Pollard,
1988; Rubin, 1992]. About ten years after the end of the rifting episode, the post rifting
deformation still consists of several cm of horizontal opening and >10 cm of regional vertical
uplift, much larger than the inter-rifting values [Hofton and Foulger, 1996].
Here we focus on the cumulative surface displacement of the magmatic system, as
produced on the longer-term (hundreds of years or more) by normal faults and extension
fractures dissecting the lava flows. In the the central portion, the magmatic system does not
consist of a simple and single graben; rather, there is a 2-3 km wider graben, partly
asymmetric, with several extension fractures and normal faults within, defining 30-300 m
wide minor nested grabens (Fig. 8f). The mean strike of the faults is N8°, similar to the N12°
of the extension fractures (Fig. 8b). The graben border faults are usually vertical at the
surface with a tensile area between the footwall and the hanging wall; the largest fault,
defining the western rim of the 2-3 km wide central graben, has a maximum throw of 42 m
and maximum width of 28 [Opheim and Gudmundsson, 1989; Angelier et al., 1997]. The
average fault throw of the border faults in our studied area is 5 m; the width of the tensile
area between the footwall and the hanging wall of the faults is commonly ~3 m, locally
reaching 5 m (Fig. 8e). The opening of the extension fractures is usually <1 m (Fig. 8e),
smaller than the mean width of the tensile area of the normal faults. Most extension fractures
at Krafla, even with opening of a few m, are found on a flat surface. The footwall of the
normal faults is usually subhorizontal, whereas the hanging wall may be subhorizontal or
20°-30° inward tilted and locally separated in different collapsed blocks (Fig. 8d, h, i).
Contractional structures are found at the base of the tilted hanging wall of the normal faults
in two places: ~8 km north to the caldera rim and at the southern tip of the fissure swarm
(Fig. 8a). The latter is along the 2 km long Grjotagja fault, in the Myvatn area, on the western
boundary of a ~1 km wide graben on a ~2.7 ka lava flow. Here the fault striking N355°
22
displays 5-6 m of vertical throw, with a titled hanging wall of 30-35°, separated from the
footwall by a 2 m open fracture (Fig. 8i, l). The well-exposed base of the tilted hanging wall
consists of a strip parallel to the fault trace, 102 m long, of folded and at times chaotic lava
blocks, indicating localized contraction. The best exposures systematically reveal a tight
cylindrical fold, ~1 m wide and with an ~90° interlimb angle, whose axis is parallel to the
fault (Fig. 8 m). The associated contraction of the fold is 7–20% of the horizontal extension
[Trippanera et al., 2014a]. A similar contraction at the base of the tilted hanging wall is found
on a fault to the north of Grjotagja (Fig. 8a, 9) and on another fault north of Krafla caldera
(Fig. 8a). Along the Krafla magmatic system, 95% of the observed fault tips (that is 18 out
of 19) terminate as open fractures without any hanging wall tilt, or as OPF type (Table 1,
Fig. 8g; Fig. 9).
4.2.2 Vogar rift zone
The 10 km long Vogar fissure swarm lies on the northern portion of the 35 km long and
5-8 km wide Reykjanes fissure swarm, in southern Iceland (Fig. 10a, b) [i.e., Gudmundsson,
1980; Villemin and Bergerat, 2013; Clifton and Kattenhorn, 2006]. The amount of pure
extension of this highly oblique portion of rift is ~6 mm/yr [Keiding et al., 2008]. The Vogar
fissure swarm consists of a 3-4.5 km wide asymmetric deformation zone delimited on the
western edge by an east facing normal fault (Hrafnagjá) and on the eastern edge by a series
of open fissures (Grindavikurgjá) [Villemin and Bergerat, 2013]. An inner graben hosts five
major west facing normal faults, vertical at the surface and with a dilational component
between the footwall and hanging wall. The footwall is subhorizontal and the hanging wall
varies its dip from subhorizontal to tilted (~30°). The mean fault throw and width of the tensile
area are ~2 m and ~1 m respectively, both reaching a maximum >10 m. Similarly to what
observed at Krafla, at Vogar there is contraction at the base of the tilted hanging wall of
normal faults, as for example one with vertical throw of a few meters bordering a ~20 m wide
graben. Here the tilted lavas of the hanging wall abruptly terminate above the horizontal
lavas of the inner graben, indicating the presence of a thrust or reverse fault, rather than
folds, as in Krafla (Fig. 10e, f). Most extension fractures at Vogar, even with opening of a
few m, are found on a flat surface. Field measurements show that 35 out of 42 lateral tips of
the normal fault are tilted, forming a monocline and suggesting the presence of FPF types;
the remaining 7 faults terminate as open fractures in a flat topography area, suggesting OPF
types (Fig. 10c, d) [Norini et al., 2009]. These results are broadly consistent with previous
23
ones, highlighting that beyond their lateral tips, the normal faults continue as narrow
monoclines and/or linear clusters of tension fractures [Grant and Katternhorn, 2004].
4.2.3 Thingvellir rift zone
The 10-20 km wide NE-SW trending Thingvellir rift zone belongs to the >60 km long
Pleistocene Hengill Volcanic System and lies north to the Lake Thingvallavatn in Iceland
(Fig. 10a, 11a). Similarly to Vogar, here the extension rate is very low [5 mm/yr; Perlt et al.,
2008]. The inner part of the rift zone consists of a ~5 km wide graben delimited by several
inward dipping conjugate normal faults dissecting basaltic lavas flows, pillow lavas and lake
sediments >1.9 ka [Gudmundsson, 1987 and references therein; Bull et al., 2005]. The
graben has been affected by up to 40 m of subsidence and 70 m of horizontal extension in
the last 9 ka [Seamundsson, 1992].
Most of the border faults in Thingvellir show the typical dilational component, with flat
footwall and tilted hanging wall. The vertical throws of the major faults are in the order of
very few tens of meters, locally reaching 30-35 m. This is the case of the 7.7 km long
Almannagja fault, delimiting the eastern graben border, and the 11 km long Hrafnagja fault,
delimiting the western border (Fig. 11b, c) [Gudmundsson, 1987; Seamundsson, 1992; Bull
et al., 2005; Sonnette et al., 2010]. Thingvellir also shows smaller Holocene normal faults
with both flat footwall and hanging wall. Many extension fractures, even with opening of a
few m, are found on a flat surface. Field measurements at Thingvellir show that 22 out of
the 26 observed lateral terminations of the normal faults are tilted, suggesting a FPF type:
among these is the southern termination of the large Almannagjà Fault, bordering the graben
(Fig. 11d, e); conversely, only 4 terminations show a flat footwall and hanging wall,
resembling a OPF type [Norini et al., 2009].
4.2.4 Fantale magmatic system
The Fantale magmatic system lies in the central portion of the continental Main Ethiopian
Rift (MER), Ethiopia, where the mean extension rate is ~7 mm/yr [Fernandes et al., 2004].
From a structural point of view, this area is probably the most representative of the MER, as
it hosts recent and well-preserved fault and fracture systems (i.e., fissure swarm), well
expressed along the southern sector of the Fantale caldera. The last historical eruption in
the southern sector occurred in 1810 AD [Williams et al., 2004] evidenced by a lava field
and a series of N25° aligned cinder and splatter cones. The fissure a swarm follows an
overall NNE-SSW orientation and is characterized by extensional fractures and normal faults
24
that dissect the 168±38 ka welded tuff, originating from the Fantale caldera [Williams et al.,
2004]. The fissure swarm, developed during the last 7 ka, focuses in a 3 km wide and 15
km long area [Williams et al., 2004 and references therein].
Our field survey has been carried out across a principal graben structure, located between
the base of the volcano to the lake shoreline to the South. It has a central 2.5 km wide
depression, delimited by two main inward dipping normal border faults oriented N24°±12°
oriented (Fig. 12b). The faults show a constant dilational component of a few meters
between the footwall and the hanging wall (Fig. 12e, 12g). The throw on the border faults
reaches 10-15 m on the western and 15-20 m on the eastern border. The mean tilt of the
hanging wall tilt is of 30°10°, but may reach locally up to 88° (Fig. 12c). Similarly to Krafla
and Vogar, local contraction structures parallel to the fault strike have been observed in the
central graben (Fig. 12d) [Trippanera et al., 2014a]. At the graben center, several areas of
extension fractures, showing an opening of 1-2 m, grouped in different fracture swarms
(Fig. 12f). The analysis of the terminations of the principal faults reveals that 14 out of 26
tips have a tilted hanging wall, suggesting a FPF type, whereas 12 fault terminations show
an OPF type. Most of the FPF type of termination (10 cases) are located along the border
faults. Conversely, most of the OPF types of termination (11 cases) are located inside the
graben. To the North of Fantale, sparse border faults are observed with poor accessibility
due to dense vegetation coverage. We generally observed similar structures compared to
the South.
5 - Discussion
5.1 – Interpreting the collected data
All the considered studied areas, independently of the nature of the involved crust
(oceanic or continental) and the amount of extension, display an overall similar deformation
pattern, given by eruptive fissures, extension fractures and normal faults. Normal faults
usually form graben-like structures hosting eruptive fissures. In the case of simple eruptive
fissures (Laki, Eldgja, Sveinagja, Sveinar, Bardarbunga), the graben is narrow (1 km or less)
and usually symmetric. In the case of the rift segments (Krafla, Vogar, Thingvellir, Fantale),
the few km wide graben may also show some degree of asymmetry, hosting nested grabens
within and extended fracture swarms. The morphology and structure of the normal faults of
the graben(s) may vary along the eruptive fissures or the rift segments: the most common
types of fault morphology and structure, in section view, are summarized in Fig. 1SM
25
(Supplementary Material). The largest differences in the fault types are related to the
structure of the tensile portion between the footwall and the hanging wall, as well as the tilt
of the latter. In addition, where the material at the surface is poorly consolidated or weak, it
may be easily eroded and, as a result, the morphology of the fault may be not well defined,
the vertical throw may be underestimated and the dilational component may be partly buried
by the colluvium. This concerns for example the Lakagigar graben faults, not always well
marked on the flat areas filled by loose eruptive products, conversely to the well-exposed
faults and fractures on the consolidated hyaloclastite of Laki hill. Similar effects of the
strength of the rocks on the morphology and structure of the normal faults have been
highlighted by with numerical models [Hardy, 2013].
Even though with some scatter, there is an overall proportion between the vertical throw
of the normal faults and their dilational component, suggesting that the larger is the amount
of extension, the larger is also the vertical displacement (Fig. 13). The scatter effect may
partly be explained by the topography that induces strong along strike variations of the fault
throw. Conversely, the amount of tilt of the hanging wall does not seem to show any
relationship with the fault displacement or the dilational component (Table 1); in addition,
the fact that along the same fault segment, within a distance of very few tens of meters, the
tilt of the hanging wall varies dramatically, from 22° to 65° or more (as in Sveinar), suggests
that this feature is largely independent of the kinematic features of the normal fault. The tilt
of the hanging wall may be related to the variation in the friction along the fault plane at
depth, in turn also related to the presence of asperities; a higher friction at depth would result
in a tilted hanging wall at the surface, whereas a lower friction in a flat hanging wall. It is
expected that, decreasing the amount of friction along the fault through its activity, most
tilted hanging walls may evolve into flat hanging walls. Some recently formed and smaller
normal faults show an ubiquitous hanging wall tilt, as at Bardarbunga (Table 1), confirming
the possibility that the maturity of a fault decreases the hanging wall tilt. However, other
larger faults do still show a significant hanging wall tilt, as at Thingvellir, suggesting that fault
maturity alone cannot explain the tilt. If we consider the possibility that these larger faults
may be dike-induced (see below), recent analogue models of deformation due to dike
emplacement suggest that the depth of the dikes may partly explain the tilt of the hanging
wall of the large faults. In fact, in the experiments the hanging wall tilt is observed above
deeper intruded dikes, and is related to the presence of an upward propagating high angle
fault becoming reverse at the surface [Trippanera et al., 2014a]. Therefore, the higher depth
of the dike may explain the tilted hanging wall shown by some faults. A further possibility to
26
explain the tilt of the hanging wall of mature faults is their upward propagation during
reactivation, as explained further below.
The deformation pattern associated with the emplacement of the eruptive fissures (Laki,
Eldgja, Sveinagja, Sveinar, Bardarbunga) is consistently given by a set of conjugate normal
faults, usually to the sides of the fissure; the resulting graben is narrow, 0.16 to 1.7 km wide
(Table 1). Single rifting episodes may be associated with normal faults with mean vertical
displacement up to 6.5 m and dilational component up to 3 m, as at Bardarbunga in 2014;
single diking episodes may also generate up to 10 m of maximum vertical displacement, as
at Lakagigar in 1783, or even more, as at Sveinar. These values are significantly larger than
those previously measured, ≤ 2-3 m, in several major rifting episodes, as Krafla in 19751984, Dabbahu in 2005, and single dike intrusion such as at Dallol in 2004 and at Harrat
Lunayyir in 2009 [Sigurdsson, 1980; Rubin and Pollard, 1988; Rowland et al., 2007; Pallister
et al., 2010; Nobile et al., 2012].
In addition to these field evidences, experimental, analytical and numerical models have
repeatedly shown that the emplacement of a shallow dike feeding an eruptive fissure, may
generate a set of conjugate normal faults, resulting in a graben at the surface [Mastin and
Pollard, 1988; Rubin and Pollard, 1988; Rubin, 1992; Gudmundsson, 2003; Gudmundsson,
2005]. The studies linking normal faulting at the surface to dike emplacement at depth
assumed that the faults propagated upward from the upper tip of the dike [Grant and
Kattenhorn, 2004; Tentler, 2005, Rowland et al., 2007]. This implies that the lateral
terminations of the normal faults would show a fault propagation fold-like structure,
highlighting the upward propagation of the fault. Conversely to what expected, our field data
at the tips of the normal faults related to the development of three eruptive fissures
(Lakagigar, Sveinagja, Sveinar) show an OPF type of termination, suggesting downward
fault propagation during dike emplacement (Fig. 14a,b,c). In the remaining case (Eldgja NE),
the faults terminate laterally developing fault propagation folds-like structures, suggesting
upward propagation (Table 1).
Conversely to the poorly informative amount of tilt of the hanging wall in the central part
of the fault, the tilt of the hanging wall at the lateral termination of the fault is informative to
infer the possible sense of propagation of the fault [e.g. Cartwright and Mansfield, 1998].
This tilt, in fact, allows detecting the very early stage mechanism of development of the
structure, immediately before the fault forms, at the lateral tip of the precursory extension
fracture [Gudmundsson, 1992; Acocella et al., 2003; Grant and Kattenhorn, 2004]: at this
early stage, the presence or absence of the tilt of the hanging wall reflects the direction of
27
propagation of the fault. Therefore, the available data from the eruptive fissures suggest that
most faults related to the emplacement of the dikes have propagated from the surface
downwards. This is consistent with experiments of dike-induced faulting, where the normal
faults formed during the emplacement of a dike complex systematically develop at the
surface and propagate downward; these experiments, coupled with numerical models, show
that the nucleation of the normal faults and their downward propagation result from the
presence of a free-surface, where the lithostatic load is least [Trippanera et al., 2014b;].
These results are also consistent with previous models of fault formation along divergent
plate boundaries, where a normal fault results from the growth of an extension fracture at
the surface, once the latter reaches a critical depth when the lithostatic load imposes a shear
stress on the fracture plane [Forslund and Gudmundsson, 1992; Gudmundsson, 1992;
Acocella et al., 2003].
The narrow grabens, ~1 km wide, associated with the dikes feeding the studied eruptive
fissures also imply a very shallow depth of the dike at the moment of the nucleation of the
normal faults. Considering a commonly observed dip of ~70° of the normal faults in the
eroded portions of Iceland [Gudmundsson, 1992], these grabens very likely formed when
the upper tip of the dike was shallower than 1.5-2 km. However, in some cases (Lakagigar,
Eldgja Central) a mean width of 200-300 m of the graben suggests a much shallower depth
to the top of the dike for the nucleation of the normal faults, in the order of 500-700 m or
less.
The structure of the rift segments at Krafla, Vogar, Thingvellir and Fantale consists of an
overall repetition of the structural features found along the eruptive fissures. In fact, these
rift segments are characterized by a wider graben-like structure (as at Thingvellir), often
showing minor structures or grabens within (as at Fantale or Krafla). Alternatively, the rift
zone may consist of juxtaposed and nested graben systems along a much wider area than
that of the single eruptive fissures (as at Vogar). Despite repetitions and asymmetries, the
basic graben-like scheme found along the eruptive fissures is met also in the considered rift
segments; in addition, the structural features, size and displacements of the normal faults
and the extension fractures are remarkably similar in the eruptive fissures and rift segments.
All these similarities suggest overall common mechanisms in the development of the
eruptive fissures and the rift segments. A lateral termination of the normal faults as OPF
type is also supported by the data collected along the Krafla magmatic system and, to a
minor extent, along the rift segments of Fantale, Thingvellir and Vogar (Table 1). This
suggests that an important part of the normal faults along wider and more complex rift
28
segments may still form though downward propagation from the surface. However, the
measured FPF types of fault tips predominate along the rift segments, and near double the
OPF types, being 72 against 41 (Table 1). This FPF type of fault termination suggests an
upward propagation of the normal faults, towards the surface. Despite the limited amount of
data, on the rift zones there may be a proportion between the frequency of the OPF types
of fault termination and the extension rate of the rift segment (Table 1). This is also supported
by the fact that the only FPF type of fault tip at the faster extending rift of Krafla has been
found at the less active southern termination of the magmatic segment, where the extension
rate appears significantly lower [Brandsdóttir et al., 1997].
In synthesis, the collected field data along selected portions of divergent plate boundaries
show that eruptive fissures are systematically bordered by graben-like structures with open
normal faults with vertical displacement easily reaching up to ~10 m. This simple structural
scheme is found, with asymmetries and repetitions, also within the rift segments. In addition,
both the OPF and the FPF types of fault termination are present. While the OPF type
dominates in faults at the sides of eruptive fissures, the rift segments show a predominance
of FPF types, with the exception of the more extended Krafla portion, suggesting a possible
correlation between the mode of formation of the normal faults and the spreading rate. The
OPF type of fault tip, implying downward propagation from the surface, is well supported by
modeling evidence, suggesting that this is a feasible mechanism of development of normal
faults along divergent plate boundaries; these faults, despite their downward propagation,
result from dike emplacement. The FPF type of fault tip is not supported by modeling results
and appears more challenging to be explained. A possible solution is related to the
reactivation of pre-existing buried normal faults. Most faults along divergent plate boundaries
are reactivated during magmatic and/or tectonic events and fault reactivation is an important
process in the activity of a rift, often requiring the least energy for extension. Volcanic
products along rift zones may easily bury faults and extension fractures at the surface. The
possibility to bury a fault and the duration of the burial depend, in addition to the frequency
and type of volcanic activity, on the extension rate of the rift segment: the lower the extension
rate, the more effective is the possibility to bury a fault, and for a longer period. Any
reactivation induces slip along a pre-existing fault, or portion of; if the slip reaches the
periphery of the fault, then the fault propagates. Along buried normal faults, if the uppermost
portion of the reactivated fault carries most of the slip, the fault will propagate towards the
surface. Therefore, the reactivation of pre-existing normal faults may induce an upward
propagation, through the deposits above the fault, developing a FPF type of fault tip (Fig.
29
14d, e, f). A similar process, highlighting the relationships between fault propagation and
resurfacing by lava flows, has been previously described to explain the geometric and
kinematic features of the Koa’e Fault System on Kilauea volcano, Hawaii [Podolsky and
Roberts, 2008]. The proposed mechanism may also explain the higher frequency of the FPF
types of fault tip in the less extending rift segments (Table 1), where fault activity is slower
and the faults may be and remain more easily buried.
Even if a reactivated normal fault has been propagating upward towards the surface,
developing a fault propagation fold, our collected data, as well as modeling results and field
studies [Gudmundsson, 1992; Acocella et al., 2003; Trippanera et al., 2014b], suggest that
in general the normal faults primarily nucleate from the surface and propagate downwards.
Conversely to previous studies directly relating normal faulting to regional extension
(Gudmundsson, 1992; Acocella et al., 2003), our data suggest that, neglecting any
reactivation, any normal fault along divergent plate boundaries may propagate downward
during diking. This is evident in the case of the narrow grabens containing all the eruptive
fissures, where the common downward propagating normal faults result from the
emplacement of the dikes feeding the fissure. The possibility of a magmatic origin for the
normal faults is likely in the case of rift segments with higher spreading rate, consisting of a
major, more or less symmetric graben-like structure with nested grabens within, where OPF
types of fault tip predominate, as at Krafla. These OPF types may be well reconciled with
the downward propagation of normal faults induced by diking. The possibility of a magmatic
origin for the normal faults is also likely in the case of rift segments with lower spreading
rate, characterized by dominant FPF types of fault termination, interpreted as resulting from
fault reactivation; in this case, the same faults should have formed propagating downward
during diking episodes and reactivated during subsequent diking episodes.
5.2 - A general model for rift segments
The collected data a) indicate the importance of the dikes feeding eruptive fissures in
creating surface deformation and b) suggest that diking may also explain most of, perhaps
all, the structure of the rift segments. Here we try to further test the likely possibility that the
more elusive structure of the rift segments may be magma induced, similarly to that of the
eruptive fissures. To this aim, we consider available data from the deeper portions of the rift
zones, as outcropping on the eroded magmatic systems in eastern Iceland. Here, at a
paleodepth of 1–2 km, the magmatic segments consist of regional dike swarms 50 km long
and 5-10 km wide [Walker, 1958; 1960; Gudmundsson, 1983; 1995]. We consider the well30
exposed and well-studied Alftafjordur paleo-magmatic system to constrain the mean number
of dikes possibly constituting a dike swarm and any variation in their frequency with the
distance from the dominant volcano; to this aim, we trace five profiles across different
portions of the dike swarm, using both published and unpublished data (Fig. 15)
[Gudmundsson, 1990; Paquet et al., 2007]. Of the considered sections, A and B do not cover
the entire width of the swarm, that is partly below the sea level; therefore, here we
extrapolate the possible number of dikes across the entire width of the swarm from their
frequency in the inland portion of the profiles. The comparison among the 5 profiles shows
that there is not any significant variation in the dike frequency across the swarm with the
distance from the dominant volcano (Fig. 15b), with a mean frequency of 100±20 dikes at a
paleodepth of ~1.5 km.
In the eroded rift portions of Iceland, the frequency of the dikes is highest at the sea level
and decreases upwards, suggesting that most dikes do not propagate upward and taper
away towards their paleosurface [Walker, 1960; Gudmundsson, 1983]. In order to have a
rough estimate of the percentage of dikes reaching the paleosurface, we plot the variation
of the frequency of the dikes with their present altitude (the inverse of the paleodepth) using
data from Alftafjordur [Walker, 1960] and Reydarfjordur [Walker, 1958] in Eastern Iceland
and from NW Iceland [Gudmundsson, 1984]. Despite the considerable scatter of the data,
only a fraction (<20%) of the dikes found at ~1.5 km of depth within a magmatic system
reaches an altitude of 1 km above sea level, corresponding to a paleodepth of ~0.5 km (Fig.
16). Therefore, if the 100±20 dikes found at Alftafjordur are representative, as order of
magnitude, of the frequency of the dikes currently at a paleodepth of ~1.5 km below the
active rifts of Iceland, only a very few tens of dikes are expected to reach shallow levels
(~0.5 km depth) in a magmatic system.
Let’s now try to consider the possible effect of these very few tens of dikes on the surface
deformation pattern. Section 5.1 has shown that at a depth of ~0.5 km any propagating dike
has induced anelastic surface deformation (see section 5.1), activating or reactivating
normal faults at the surface. The mean vertical slip commonly observed at the surface on
normal faults during different diking episodes is shown in Fig. 17 as a function of the inferred
depth and thickness of the dike. Despite the scatter, there is an overall increase in the slip
of the fault with the thickness of the dike and its shallowness. This implies that shallower
and thicker dikes induce larger slip on the normal faults at the surface. In any case, the
observed slip on the faults is on the order of a very few meters, commonly between 1 and 3
31
m (Fig. 17). This confirms that each diking episode may be, in general, responsible for a
vertical slip of a very few meters along a pair of conjugate normal faults.
Assuming that 10-20 dikes may reach the uppermost 0.5 km of depth in the active portion
of a rift zone (Fig. 16), one may be expect a cumulative amount of vertical deformation at
the surface between 10 m (assuming the lower bound of the mean slip value, 1 m, for 10
dikes) and 60 m (assuming the upper bound of the mean slip value, 3 m, for 20 dikes). These
values may explain the common vertical displacements, carried either along a few major
faults or several minor faults, currently observed at the surface along the active portions of
rift segments, as at Krafla, Thingvellir, Vogar and Fantale. However, since single diking
episodes, as observed at Lakagigar, Sveinar or Bardarbunga, may induce surface faulting
with vertical slip of 10 meters or more (Fig. 3e), even very few shallow dikes may explain
the large surface deformation observed along active rift segments, as at Krafla. These
results suggests that shallow dike emplacement (<0.5 to 1 km) may explain all the observed
deformation also along the considered rift segments, so that the overall shape, structure and
development of the considered divergent plate boundaries may be in principle entirely
magma-induced. Regional tectonics, in the form of ridge push, certainly plays a fundamental
preparatory role in shaping these plate boundaries, influencing the direction of the dikes and
separating the plates apart on the longer-term and over longer distances. Regional tectonics
may also directly activate normal faults, seismically or not, during inter-diking episodes.
However, the collected data suggest that in these magmatically active rifts, where plate
separation is constantly maintained by diking, the regional tectonic stress may rarely reach
values high enough to be released and activate the normal faults. Therefore, any direct role
of regional tectonics in separating and shaping these plate boundaries appears negligible
compared to that of diking.
6 – Conclusions
1) Eruptive fissures along divergent plate boundaries in Iceland and at Fantale (Ethiopia)
are systematically bordered by grabens with open normal faults, with vertical displacement
reaching ~10 m. Most faults along eruptive fissures terminate as open fractures on a flat
surface, suggesting downward propagation after surface nucleation.
2) A consistent graben-like structure is found, with asymmetries and repetitions, also
within the rift segments of Iceland and the MER. Here the development of the normal faults
may be influenced by the spreading rate, even though this needs to be confirmed. In fastly
extending rift segments (>1 cm/yr), normal faults usually terminate as open fractures on a
32
flat surface, suggesting downward propagation after surface nucleation. In slowly extending
rift segments (<1 cm/yr), normal faults predominantly terminate as open fractures on a
monocline, suggesting upward propagation after deep nucleation.
3) Recent experimental and numerical models (Trippanera et al., 2014b) show that dikeinduced normal faults propagate downward from the surface, suggesting that any upward
propagation of normal faults observed on the field is largely related to fault reactivation due
to dike emplacement after burial by volcanic deposits. The fact that the latter type of fault
propagation is largely found on slowly extending rift segments suggests that the faults here
may be and remain more easily buried.
4) In general, the entire surface deformation pattern of eruptive fissures and rift segments
observed along the considered divergent plate boundaries may be qualitatively explained
by dike emplacement.
5) In addition, simple calculations, partly based on the structure of the eroded rift portions
of eastern and western Iceland, suggest that all the fault slip in the studied rift zones,
reaching several tens of meters, may result from repeated shallow dike intrusion.
6) The collected data suggest that, in a magmatically active rift with repeated dike
emplacement, the regional tectonic stress may be rarely high enough to be released, so that
any direct role of regional tectonics in shaping plate boundaries appears negligible
compared to diking.
Acknowledgements
Thor Thordarson and Gianluca Norini partly participated to the field analysis. Financed
with PRIN 2009 funds (2009H37M59, responsible V. Acocella). Any user can access the
data of this work by contacting the corresponding author.
References
Acocella, V. (2014). Structural control on magmatism along divergent and convergent
plate boundaries: overview, model, problems. Earth-Science Reviews, 136, 226–288.
Doi:10.1016/j.earscirev.2014.05.006
Acocella, V., and Korme, T. (2002). Holocene extension direction along the Main
Ethiopian Rift, East Africa. Terra Nova 14, 191–197
Acocella, V., Korme, T., and Salvini, F. (2003). Formation of normal faults along the axial
zone of the Ethiopian Rift. Journal of Structural Geology, 25, 503–513.
Acocella, V., and Neri, M. (2003). What makes flank eruptions? The 2001 Etna eruption
and its possible triggering mechanisms. Bulletin of Volcanology, 65, 517-529. Doi:
10.1007/s00445-003-0280-3
33
Angelier, J., Bergerat, F., Dauteuil, O., and Villemin, T. (1997). Effective tension-shear
relationships in extensional fissure swarms, axial rift zone of northeastern Iceland. Journal
of Structural Geology, 19(5), 673–685. Doi: 10.1016/S0191-8141(96)00106-X
Angelier, J., Bergerat, F., Stefansson, R., and Bellou, M. (2008). Seismotectonics of a
newly formed transform zone near a hotspot: earthquake mechanisms and regional stress
in the South Iceland Seismic Zone. Tectonophysics, 447, 95–116.
Biggs, J., Amelung, F., Gourmelen, N., Dixon, T. H., and Kim, S.W. (2009). InSAR
observations of 2007 Tanzania rifting episode reveal mixed fault and dyke extension in an
immature continental rift. Geophysical Journal International, 179(1), 549–558.
doi:10.1111/j.1365-246X.2009.04262.x
Bjornsson, A., Saemundsson, K., Einarsson, P., Tryggvason, E., and Gronvold, K. (1977).
Current rifting episode in North Iceland. Nature, 266, 318–322.
Brandsdóttir, B., Menke, W. H., Einarsson, P., White, R. S., and Staples, R. K. (1997).
Faroe-Iceland Ridge Experiment. 2. Crustal structure of the Krafla central volcano. Journal
of Geophysical Research, 102(B4), 7867–7886.
Buck, W.R. (2006). The role of magma in the development of the Afro-Arabian Rift
System. Geological Society of London, Special Publucations, 259, 43–54.
Buck, W.R., Einarsson, P., and Brandsdóttir, B. (2006). Tectonic stress and magma
chamber size as controls on dike propagation: constraints from the 1975–1984 Krafla rifting
episode.
Journal of
Geophysical
Research,
111(B12), B12,
404.
Doi:
10.1029/2005JB003879
Bull, J.M., Minshull, T.A., Mitchell, N.C., Dix, J.K., and Harnardottir, J. (2005). Magmatic
and tectonic history of Iceland's western rift zone at Lake Thingvallavatn. Geological Society
of America Bulletin, 117, 1451–1465.
Cartwright, J.A., and Mansfield, C.S. (1998). Lateral displacement variation and lateral tip
geometry of normal faults in the Canyonlands National Park, Utah. Journal of Structural
Geology, 20(1), 3-19.
Casey, M., Ebinger, C., Keir, D., Gloaguen, R., and Mohamed, F. (2006). Strain
accommodation in transitional rifts: extension by magma intrusion and faulting in Ethiopian
rift magmatic segments. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar
Volcanic Province within the East African Rift System. Geological Society of London, Special
Publications, 259, 143–163.
Clifton, A.E., and Kattenhorn, S.A. (2006). Structural architecture of a highly oblique
divergent plate boundary segment. Tectonophysics, 419, 27–40.
Corti, G. (2008). Control of rift obliquity on the evolution and segmentation of the main
Ethiopian rift. Nature Geoscience, 1, 258–262.
Corti, G. (2009). Continental rift evolution: from rift initiation to incipient break-up in the
Main Ethiopian Rift, East Africa. Earth-Science Reviews, 96, 1–53.
Daly, E., Keir, D., Ebinger, C.J., Stuart, G.W., Bastow, I.D., and Ayele, A. (2008). Crustal
tomographic imaging of a transitional continental rift: the Ethiopian rift. Geophysics Journal
International, 172, 1033–1048.
DeMets, C., Gordon, R., and Argus, D. (2010). Geologically current plate motions.
Geophysical Journal International, 181, 1–80, doi: 10.1111 /j.1365 -246X .2009 .04491.x
Ebinger, C., 2005. Continental breakup: the East African perspective. Astron. Geophys.
46, 2.16–2.21.
Ebinger, C. J., and Casey, M. (2001). Continental breakup in magmatic provinces : An
Ethiopian example. Geology, 29, 527–530. Doi: 10.1130/0091-7613(2001)029<0527
Ebinger, C., Ayele, A., Keir, D., Rowland, J. V, Yirgu, G., Wright, T., Hamling, I. (2010).
Length and Timescales of Rift Faulting and Magma Intrusion : The Afar Rifting Cycle from
2005 to Present. Annual Review of Planetary Science, 38, 439–466. Doi: 10.1146/annurevearth-040809-152333
34
Ebinger, C.J., Keir, D., Ayele, A., Calais, E., Wright, T.J., Belachew, M., Hammond,
J.O.S., Campbell, E., and Buck, W.R. (2008). Capturing magma intrusion and faulting
processes during continental rupture: seismicity of the Dabbahu (Afar) rift. Geophysics
Journal International, 174, 1138–1152.
Einarsson, P., (1991). Earthquakes and present-day tectonism in Iceland.
Tectonophysics, 189, 261–279.
Fernandes, R.M.S., Ambrosius, B.A.C., Noomen, R., Bastos, L., Combrinck, L., Miranda,
and J.M., Spakman, W. (2004). Angular velocities of Nubia and Somalia from continuous
GPS data: implications on present-day relative kinematics. Earth Planetary Science Letters,
222, 197–208.
Forslund, T., and Gudmundsson, A. (1992). Structure of Tertiary and Pleistocene normal
faults in Iceland. Tectonics, 11, 57–68.
Grant, J. V, and Kattenhorn, S. A. (2004). Evolution of vertical faults at an extensional
plate boundary, southwest Iceland. Journal of Structural Geology, 26, 537–557.
Doi:10.1016/j.jsg.2003.07.003
Gudmundsson, A. (1980). The Vogar fissure swarm, Reykjanes Peninsula, SW lceland.
Jokull, 30, 43-64.
Gudmundsson, A. (1983). Stress estimates from the length/width ratios of fractures.
Journal of Structural Geology, 5, 623-626.
Gudmundsson, A. (1984). Tectonic aspect of dykes in Northern western Iceland. Jokull,
34, 81-96.
Gudmundsson, A. (1987). Tectonics of the Thingvellir fissure swarm, SW Iceland. Journal
of Structural Geology, 9(1), 61–69.
Gudmundsson, A. (1987b). Geometry, formation and development of tectonic fractures
on the Reykjanes Peninsula, southwest Iceland. Tectonophysics, 139, 295–308.
Gudmundsson, A. (1990). Dyke emplacement at divergent plate boundaries. Mafic dykes
and emplacement mechanisms, 47-62.
Gudmundsson, A. (1992). Formation and growth of normal faults at the divergent plate
boundary in Iceland. Terra Nova, 4, 464-471.
Gudmundsson, A. (1995a). Infrastructure and mechanics of volcanic systems in Iceland.
Journal of Volcanology and Geothermal Research, 64, 1–22.
Gudmundsson, A. (1995b). Ocean-ridge discontinuities in Iceland. Journal of Geological
Society of Londondon, 152, 1011–1015.
Gudmundsson, A. (2000). Dynamics of volcanic systems in Iceland: example of tectonism
and volcanism at juxtaposed hot spot and Mid-Ocean Ridge system. Annual Review of Earth
and Planetary Science, 28, 107-140.
Gudmundsson, A. (2003). Surface stresses associated with arrested dykes in rift zones.
Bulletin Volcanology, 65, 606-619.
Gudmundsson, A. (2005). Effect of mechanical layering on the development of normal
faults and dykes in Iceland. Geodinamica Acta, 18(1), 11-30.
Gudmundsson, A. (2007). Infrastructure and evolution of ocean-ridge discontinuities in
Iceland. J. Geodyn. 43, 6–29.
Gudmundsson, A. and Backstrom, K. (1991). Structure and development of the Sveinagja
graben, Northeast Iceland. Tectonophysics, 200, 111–125.
Gudmundsson, A. and Brynjolfsson, S., (1993). Overlapping rift-zone segments and the
evolution of the South Iceland Seismic Zone. Geophysical Research Letters, 20, 1903–
1906.
Gudmundsson, A., Lecoeur, N., Mohajeri, N., and Thordarson, T. (2014). Dike
emplacement at Bardarbunga, Iceland, induces unusual stress changes, caldera
deformation, and earthquakes. Bulletin of Volcanology, 76, 1-7. Doi: 10.1007/s00445-0140869-8
35
Hammer, C.U. (1984). Traces of Icelandic eruptions in the Greenland ice sheet. Jokull,
34, 51-65.
Hardy, S. (2013). Propagation of blind normal faults to the surface in basaltic sequences:
Insights from 2D discrete element modelling. Marine and Petroleum Geology, 48, 149-159.
Helgason, J., and Zentilli, M. (1985). Field characteristics of laterally emplaced dikes:
anatomy of an exhumed Miocene dike swarm in Reydarfjordur, eastern Iceland.
Tectonophysics, 115, 247-274.
Hofton, M.A., and Fougler, G.R. (1996). Postrifting anelastic deformation around the
spreading plate boundary, north Iceland. 1. Modeling of the 1987-1992 deformation field
using viscoelastic Earth structure. Journal of Geophysical Research, 101(B11), 25,40325,421.
Hollingsworth, J., Leprince, S., Ayoub, F., and Avouac J.P. (2012). Deformation during
the 1975-1984 Krafla rifting crisis, NE Iceland, measured from historical optical imagery.
Journal of Geophysical Research, 117(B11407), 1-24. Doi: 10.1029/2012JB009140
Jonsson, S., Einarsson, P., and Sigmundsson, F. (1997). Extension across a divergent
plate boundary, the Eastern Volcanic Rift Zone, south Iceland, 1967-1994, observed with
GPS and electronic distance measurements. Journal of Geophysical Research, 102(B6),
11,913-11,929.
Keiding, M., Arnadottir, T., Sturkell, E., Geirsson, H., and Lund, B. (2008). Strain
accumulation along an oblique plate boundary: Reykjanes Peninsula, southwest Iceland.
Geophysics Journal International, 172, 861–872.
Keir, D., Hamling, I. J., Ayele, a., Calais, E., Ebinger, C., Wright, T. J., and Bennati, L.
(2009). Evidence for focused magmatic accretion at segment centers from lateral dike
injections captured beneath the Red Sea rift in Afar. Geology, 37(1), 59–62.
Doi:10.1130/G25147A.1
Kendall, J.M., Stuart, G.W., Ebinger, C.J., Bastow, I.D., and Keir, D. (2005). Magmaassisted rifting in Ethiopia. Nature, 433, 146–148.
Keranen, K., Klemperer, S.L., Gloaguen, R., and Eagle Working Group, (2004). Threedimensional seismic imaging of a protoridge axis in the Main Ethiopian Rift. Geology, 32,
949–952.
Keranen, K., and Klemperer, S.L. (2008). Discontinuous and diachronous evolution of the
Main Ethiopian Rift: implications for development of continental rifts. Earth and Planetary
Science Letters, 265, 96–111.
Kogan, L., Fisseha, S., Bendick, R., Reilinger, R., McClusky, S., King, R., and Solomon,
T. (2012). Lithospheric strength and strain localization in continental extension from
observations of the East African Rift. Journal of Geophysical Research, 117,
B03402.http://dx.doi.org/10. 1029/2011JB008516
LaFemina, P.C., Dixon, T.H., Malservisi, R., Arnadottir, T., Strukell, E., Sigmundsson, F.,
and Einarsson, P. (2005). Geodetic GPS measurements in south Iceland: strain
accumulation and partitioning in a propagating rift. Journal of Geophysical Research, 110,
B11405. Doi: http://dx.doi.org/ 10.1029/ 2005JB003675.
Mastin, L.G., and Pollard, D.D. (1988). Surface deformation and shallow dike intrusion
processes at Inyo Craters, Long Valley, California. Journal of Geophysical Research,
93(B11), 13,221–13,235.
Mickus, K., Tadesse, K., Keller, G.R., and Oluma, B. (2007). Gravity analysis of the main
Ethiopian rift. J. Afr. Earth Sci., 48, 59–69.
Mohr, P. (1967). Major volcano-tectonic lineament in the Ethiopian Rift System. Nature,
213, 664–665.
Mohr, P. (1987). Patterns of faulting in the Ethiopian rift Valley. Tectonophysics, 143,
169–179.
36
Nobile, A., Pagli, C., Keir, D., Wright, T. J., Ayele, A., Ruch, J., and Acocella, V. (2012).
Dike-fault interaction during the 2004 Dallol intrusion at the northern edge of the Erta Ale
Ridge
(Afar,
Ethiopia).
Geophysical
Research
Letters,
39,
1–6.
Doi:
10.1029/2012GL053152
Norini, G., Acocella, V., Gudmundsson, A., Lagmay, M., and Paguican, E. (2009). Oblique
spreading, extensional fractures, and fault growth on the rift zone of SW Iceland. Egu
abstract, Vienna 2009.
Opheim, J. A., and Gudmundsson, A. (1989). Formation and geometry of fractures, and
related volcanism, of the Krafla fissure swarm, northeast Iceland. Geological Society of
Aamerica Bulletin, 101(12), 1608–1622.
Pallister, J.S., McCausland, W.A., Jonsson, S., Lu, Z., Zahran, H.M., El Hadidy, S.,
Abrukbah, A., Stewart, I.C.F., Lundgren, P.R., White, R.A., and Moufti, M.R.H. (2010). Broad
accommodation of rift-related extension recorder by dyke intrusion in Saudi Arabia. Nature
Geoscience, 3(10), 708–712. Doi: 10.1038/ngeo966
Paquet, F., Dauteuil, O., Hallot, E., and Moreau, F. (2007). Tectonics and magma
dynamics coupling in a dyke swarm of Iceland. Journal of Structural Geology, 29(9), 1477–
1493. Doi:10.1016/j.jsg.2007.06.001
Perlt, J., Heinert, M., and Niemeier, W. (2008). The continental margin in Iceland — A
snapshot derived from combined GPS networks. Tectonophysics, 447(1-4), 155–166.
Doi:10.1016/j.tecto.2006.09.020
Podolsky, M.W. and Roberts, G.P. (2008). Growth of the volcano-flank Koa’e fault
system, Hawaii. Journal of Structural Geology, 30(10), 1254-1263.
Rowland, J.V., Baker, E., Ebinger, C.J., Keir, D., Kidane, T., Biggs, J., Wright, T.J. (2007).
Fault growth at a nascent slow-spreading ridge: 2005 Dabbahu rifting episode, Afar.
Geophysical Journal International, 171(3), 1226–1246. Doi: 10.1111/j.1365246X.2007.03584.x.
Rubin, A.M. (1992). Dike-induced faulting and graben subsidence in volcanic rift zones.
Journal of Geophysical Research, 97(B2), 1839–1858.
Rubin, A.M., and Pollard, D.D. (1988). Dike-induced faulting in rift zones of Iceland and
Afar. Geology, 16, 413–417.
Ruegg, J.C., Lépine, J.C., and Tarantola, A. (1979). Geodetic measurements of rifting
associated with a seismo-volcanic crisis in Afar. Geophysical Research Letters, 6(11), 817–
820.
Saemundsson, K. (1992). Geology of the Thingvallavatn area. Oikos, 64, 40–68.
Siebert, L., Simkin, T., and Kimberly, P. (2010). Volcanoes of the World, 3rd Ed.
University of California Press, Berkeley, 568 pp.
Sigmundsson, F. (2006). Iceland geodynamics. Crustal deformation and divergent plate
tectonics. Ed. Springer and Praxis Publishing, 247 pp.
Sigurdsson, O. (1980). Surface deformation of the Krafla fissure swarm in two rifting
events. Journal of Geophysical Researches, 47, 154–159.
Sigurdsson, H., and Sparks, S. (1978). Rifting episode in North Iceland in 1874-1875 and
the eruptions of Askja and Sveinagja. Bulletin of Volcanology, 41, 149-167.
Sonnette, L., Angelier, J., Villemin, and T., Bergerat, F. (2010). Faulting and fissuring in
active oceanic rift: surface expression, distribution and tectonic–volcanic interaction in the
Thingvellirfissure swarm, Iceland. Journal of Structural Geology, 32, 407–422.
Tarantola, A., Ruegg, J.C. and Lepine, J.P. (1980). Geodetic evidence for rifting in Afar,
2. Vertical displacement. Earth and Planetary Science Letters, 48, 363-370.
Tentler, T. (2005). Propagation of brittle failure triggered by magma in Iceland.
Tectonophysics, 406(1-2), 17–38. Doi:10.1016/j.tecto.2005.05.016
Tentler, T., and Mazzoli, S. (2005). Architecture of normal faults in the rift zone of central
north Iceland. Journal of Structural Geology, 27, 1721–1739. Doi: 10.1016/j.jsg.2005.05.018
37
Thordarson, T., Miller, D.J., Larsen, G., Self, S., and Sigurdsson, H., (2001). New
estimates of sulfur degassing and atmospheric mass-loading by the 934 AD Eldjá eruption,
Iceland. Journal of Volcanology and Geothermal Research, 108, 33-54. Doi:
10.1016/S0377-0273(00)00277-8.
Thordarson, T., and Self, S. (1993). The Laki (Skaftlir Fires) and Grimsvotn eruptions in
1783-1785. Bulletin of Volcanology, 55, 233–263.
Trippanera, D., Acocella, V., and Ruch, J. (2014a). Dike-induced contraction along
oceanic and continental divergent plate boundaries. Geophysical Research Letters, 40, 1–
7. Doi: 10.1002/2014GL061570
Trippanera, D., Acocella, V., Ruch, J., Abebe, B., Norini, G., Thordarson, T., Urbani, S.,
and Gudmundsson, A. (2014b). What controls the shallow structure of divergent plate
boundaries? Insight from field and modelling data, EGU abstract, Vienna 2014.
Tryggvason, E. (1980). Subsidence events in the Krafla area, North Iceland, 1975 – 1979,
Journal of Geophysics, 47, 141 – 153
Tryggvason, E. (1984). Widening of the Kraflafissure swarm during the 1975–1981
volcano-tectonic episode. Bulletin of Volcanology, 47, 47–69.
Tryggvason, E. (1994). Surface deformation at the Krafla volcano, North Iceland, 1982–
1992. Bulletin of Volcanology, 56, 98–107.
Villemin, T., and Bergerat, F. (2013). From surface fault traces to a fault growth model:
the Vogar Fissure Swarm of the Reykjanes Peninsula, Southwest Iceland. Journal of
Structural Geology, 51, 38-51. Doi: http://dx.doi.org/10.1016/j.jsg.2013.03.010.
Walker, G.P.L. (1958). Geology of the Reydarfjordur Area, Eastern Iceland. Quarterly
Journal of the Geological Society, 114(1-4), 367–391. Doi:10.1144/gsjgs.114.1.0367
Walker, G.P.L. (1960). Zeolite zones and dike distribution in relation to the structure of
the basalts of Eastern Iceland. Journal of the Geological Society of London, 68, 515–527.
Williams, F.M., Williams, M.A.J., and Aumento, F. (2004). Tensional fissures and crustal
extension rates in the northern part of the Main Ethiopian Rift. J. Afr. Earth Sci. 38, 183-197.
Woldegabriel, G., Aronson, J.L., and Walter, R.C. (1990). Geology, geochronology and
rift basin development in the central sector of the Main Ethiopian Rift. Geological Society of
America Bulletin, 102, 439–458.
Wright, T. J., Ebinger, C., Biggs, J., Ayele, A., Yirgu, G., Keir, D., and Stork, A. (2006).
Magma-maintained rift segmentation at continental rupture in the 2005 Afar dyking episode.
Nature, 1–5. Doi: 10.1038/nature04978
Zielinsky, G.A., Germani, M.S., Larsen, G., Baillie, M.G.L., Whitlow, S., Twicker, M.S.,
and Taylor, K. (1995). Evidence of the Eldgjá (Iceland) eruption in the GISP2 Greenland ice
core: relationship to eruption processes and climatic conditions in the tenth century. The
Holocene, 5, 129-140.
38
Figures and tables
LOCATION
FAULT TIP
GRABEN
WIDTH
(km)
MEAN
MAX
MEAN
THROW THROW OPENING MAX opening
(±0.5 m) (± 0.5 m) (±0.5 m)
EF
MEAN
TILT
(±2°)
MAX
TILT
76%
16°
30°
9
9
Inward
Hw TILT
C
Ext rate
(mm/yr)
FPF
OPF
Lakagigar
0
8
0.15-0.40
3.3 m
10 m
1.4 m
3.5 m
Eldgjà NE
2
0
0.65
5.3 m
10 m
1.6 m
3.3 m
Eldgjà C
0.16
4.6 m
10 m
1.5 m
4.0 m
x
10%
24°
24°
Eldgjà SW
0.23
4.8 m
10 m
1.7 m
3.0 m
x
20%
32°
35°
9
Bardarbunga
0.80
6m
7m
3.0 m
7.0 m
x
100%
20°
25°
16
5
1.00-1.70
3.7 m
10 m
2.1 m
13.0 m
x
33%
13°
17°
19
2.0 m
x
43%
32°
65°
x
19
28 m
x
63%
17°
48°
x
23
40°
Sveinagja
0
Sveinar
0
1
0.50-0.80
3.8 m
8m
Krafla
1
18
3.00-9.00
5.5 m
42 m
Thingvellir
22
4
5.00
Vogar
35
7
3.00-4.50
Fantale
14
12
2.50-2.70
9.0 m
2.7 m
x
0%
10
11 m
68 m
15°
?
8m
30°
20 m
5m
x
47%
30°
90°
5
x
6
x
7
Table 1: Summary of the mean structural features of the investigated areas; EF = Eruptive
fissures within the graben; C= Contraction at the base of the tilted hanging wall.
Fig.1: Map of Iceland (a) and the Main Ethiopian Rift [after Ebinger et al., 2008 and
references therein] (b), with the relative magmatic systems. Red squares show the studied
areas. WVZ = Western Volcanic Zone; EVZ = Eastern Volcanic Zone; NVZ = North Volcanic
Zone; RLVZ = Reykjanes-Langjokull Volcanic Zone; RR = Reykjanes Ridge; SISZ = South
Iceland Seismic Zone; TFZ = Tjörnes Fracture Zone; KR = Kolbeinsey Ridge.
39
Fig. 2: Scheme of open normal fault with (a) tilted hanging wall and (b) sub-horizontal
hanging wall, with definitions. Structure of the lateral tip of the normal faults observed on the
field, showing (c) both a flat footwall and hanging wall separated by extension fractures
departing (OPF type) or (d) a flat footwall and tilted hanging wall forming a monocline (fault
propagation fold type, FPF) [Cartwright and Mansfield, 1998]. Fw = Footwall; Hw = Hanging
wall.
40
Fig. 3: (a, b) Structural map of the Lakagigar eruptive fissure, Iceland, from South (1) to
North (3). (c) View of the central fissure southwards from the Laki hill (location in a; white
arrow indicates the viewpoint); the faults bounding the fissure (black lines) and the tilt of both
foot wall and hanging wall (white lines) are highlighted; (d) rose diagram of the eruptive
41
fissures and normal faults strikes; (e, f) frequency histogram of the fault throw and opening;
(g) profiles along the lines A-A’ (above) and B-B’ (below), locations in a and b respectively.
Fig. 4: (a, b, c) Structural maps of the investigated portions of the Eldgjá eruptive fissure,
Iceland. The base maps are extracted from Google Earth. (d) Rose diagram of the faults
strikes; (e, f, g) frequency histogram of the fault throw, opening and amount of tilt of the
hanging wall. (h) View of the graben on the SW segment (location in a). Here the border
42
faults are exposed on a cliff, highlighting a dip of ~80°. (i) Section A-A’ (at the NW segment;
location in c): open border fault with inward tilted hanging wall. A scheme of the fault
geometry is given in the inset.
Fig. 5: (a) Structural map of the Holuhraun area, along the northern 2014 Bardarbunga
fissure, Iceland. (b) Schematic structure of the new formed graben border faults, along the
line A-A’ (not to scale; location in a). Detailed field map (d) and photo (d) of portion of the
western boundary normal fault displaying a flat foot wall, a tilted hanging wall and a tensile
area in between (location in a and b).
43
Fig. 6: (a) Structural map of the Sveinagja graben and eruptive fissure, Iceland [from
Gudmundsson and Backstrom, 1991], in which the studied area is highlighted (red
rectangle); (b) Rose diagram of the normal faults strike; (c) frequency histogram of the
normal faults throw; (d) example of open normal fault with tilted hanging wall along the
eastern border of the graben (location in a).
44
Fig. 7 (a) Structural map of the Sveinar graben and eruptive fissure, Iceland [from Tentler
and Mazzoli, 2005]; the studied area is shown by the red rectangle); (b) detailed field map
of the studied area with relative profiles perpendicular to the fault strike, highlighting the
variations of the fault structure in a few hundreds of meters; (c, d) photos of the border faults,
showing an open structure with flat footwall and nearly subvertical hanging wall (location in
b). In (d) the base of the tilted hanging wall is almost overturned; (e) detailed scheme of d),
highlighting the very steep base of tilted hanging wall.
45
46
Fig. 8: (a) Structural map of Krafla rift segment, Iceland [after Opheim and Gudmundsson,
1989 and references therein]; (b) rose diagram of the normal faults and extension fractures
strikes; (c, d) frequency histogram of the faults throw and amount of hanging wall tilt; (e)
frequency histogram of the normal faults (black) and extension fractures (gray) opening. (f)
Minor inner graben with several extension fractures located in the central portion of the
Krafla magmatic system (within the Krafla caldera); (g) example of OPF lateral termination
at Krafla (location in a); the fault increases its displacement to the back of the photographer;
(h) high throw (>20 m) fault with pervasive fracturing of the hanging wall, close to Mofell area
in central-northern Krafla. (i) Grjotagja open normal fault showing a tilted hanging wall and
contraction at its base (details in m); the inset in f, g, h and l schematically show the
geometrical structure of the fault in the relative photo.
47
Fig. 9: a) Detailed field map of the lateral termination of a N40° oriented open normal fault
having a tilted hanging wall and contraction at its base (section A-A’) south of Krafla caldera,
Iceland (location in Fig. 8a). The tip of the fault, enlarged in b, grades in a series of open
fractures in a flat area (also shown in section B-B’), indicating an OPF structure. In a the
numbers close to the faults indicate the amount of throw, in meters; in b the numbers close
the fractures indicate the amount of opening of the extension fractures, in meters.
48
Fig. 10: (a) Map of the SW Iceland, and the related magmatic systems [Villemin and
Bergerat, 2013 and references therein]; the studied systems are outlined in red; (b) structural
map of Vogar rift segment; light grey: Upper Pleistocene hyaloclastites and lavas; white:
Postglacial lavas, older than AD 871; dark grey: Postglacial lavas, younger than AD 871
[Villemin and Bergerat, 2013 and references therein]. (c) OPF and (d) FPF lateral fault
termination at Vogar (location in a); in both cases the fault increases its displacement to the
back of the photographer; (e) open normal fault with flat footwall, tilted hanging wall and
contraction at the hanging wall base (white arrows); (f) scheme of the fault in Fig. e.
49
Fig. 11: (a) Structural map of the Thingvellir rift segment, Iceland (location in Fig. 10a)
[after Bull et al., 2005 and references therein]. (b) Typical open border fault with tilted
hanging wall close to Thingvallavatn Lake (location in a); (c) schematic structure of the fault
in b; (d) lateral fault termination of FPF type, with diffuse open fractures over a monocline,
at the southernmost portion of Almannagjà Fault (location in a); (e) detailed map of the lateral
termination of Almannagjà Fault. The fault termination is characterized by a series of open
fractures within a broad monocline tilted up to ~10°, indicating an overall FPF structure.
50
Fig. 12: (a) Structural map of the Fantale rift segment, MER; (b) rose diagram of the fault
strike; (c) frequency histogram showing the amount of hanging wall tilt; (d) open normal fault
with a flat footwall (Fw), a tilted hanging wall (Hw) and contraction at its base (white arrows);
(e) open normal fault with highly tilted hanging wall; (f) example of OPF tip; the fault
increases its displacement to the back of the photographer; (g) example of FPF tip.
51
Fig. 13: Comparison between the dilation component and vertical displacement of the
normal fault measured on the field. Values for Main Ethiopian Rift (MER) are from Acocella
et al., (2003).
52
Fig. 14: (a-c) Block diagrams showing the formation of open fractures type (OPF) of fault
termination leading to the downward propagation of normal faults during dike emplacement;
(d-f) Block diagrams showing the formation of monocline type (FPF) of fault termination due
to a fault propagation folding mechanism induced by upward propagating pre-existing faults
buried by lavas and reactivated by dikes. Red arrows indicate direction of propagation of the
normal faults.
53
Fig. 15: (a) Map of the NNE trending Alftafjordur magmatic system in E Iceland [after
Paquet et al., 2007], including the position of the sections; (b) Dike frequency in the sections
across the Alftafjordur magmatic system, with respect to the distance from the dominant
volcano. The letters above each triangle refer to the sections in a. (c) Values of dike
frequency and distance from the dominant volcano and relative references.
54
Fig. 16: Percentage of dikes with altitude above sea level (a.s.l.) at Alftafjordur [Walker,
1960] and Reydarfjordur [Walker, 1958] in Eastern Iceland and in NW Iceland
[Gudmundsson, 1984].
Fig. 17: Amount of slip (numbers above the circles) on faults formed or reactivated during
dike intrusion as a function of the dike thickness and its depth. The dike depth (D) is inferred
from the graben width (W) formed above the dike and assuming a fault dip of 70°. D=W/2
55
tan70°. References: Acocella and Neri, 2003 (Etna); Rowland, 2007 (Dabbahu); Ruegg et
al., 1979 (Asal); Pallister et al., 2010 (Harrat); Rubin, 1992 (Namafjall, Krafla); Jonsson et
al., 1997 (Lakagigar); this study and IMO Office (Bardarbunga).
Fig. 1SM: Scheme showing the most common morphology and structure of the normal
faults to the sides of the eruptive fissures and in the rift segments. Below each case the
location where a structure type is most common is reported.
56
Chapter 3
(Under review in Journal of Geophysical Research)
Experiments of dike-induced deformation:
an application to divergent plate boundaries
Trippanera D.1 *, Ruch J.1°, Acocella V.1, Rivalta E.2
1 Roma
Tre University, Rome, Italy
°Now at King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia.
2 Deutsches
GeoForschungsZentrum Potsdam, Potsdam, Germany
* Corresponding author: [email protected]
Abstract
The shallow transport of magma occurs through dikes causing surface deformation. Our
understanding of the effects of diking at the surface is still limited, especially for repeated
intrusive episodes. We use analogue models to study the upper crustal deformation induced
by dikes. We insert metal plates within cohesive sand with three setups: in A, the intrusion
rises upward with constant thickness; in B and C, the intrusion thickens at a fixed depth, with
final rectangular (B) or triangular (C) shape in section. Experiments A create a doming
delimited by reverse faults, with secondary apical graben, without close correspondence in
nature. In experiments B and C, a depression flanked by two uplifted areas is bordered by
inward dipping normal faults propagating downward and, for deeper intrusions in B, also by
inner faults, reverse at the surface; this deformation is similar to what observed in nature,
suggesting a consistent physical behavior. These results suggest that dikes in nature initially
propagate developing a mode I fracture at the tip, subsequently thickened by magma
intrusion, without any direct upward push (as in A). The deformation pattern in B and C
depends on the intrusion depth and thickness, consistently to what observed along divergent
plate boundaries. The early deformation in B and C is similar to that from single diking events
(i.e. Laki, Iceland; Dabbahu, Afar), whereas the late stages resemble the structure of mature
rifts (i.e. Krafla, Iceland), confirming diking as a major process in shaping divergent plate
boundaries.
57
Keywords
dike swarm, surface deformation, divergent plate boundaries, analogue models
1. Introduction
The shallow transport of magma, including that feeding most eruptions, occurs by means
of dikes. Dikes play a fundamental role in controlling the magmatic, structural and
morphological evolution of volcanic edifices by feeding their rift zones [i.e., Swanson et al.,
1976; Pollard et al., 1983; Acocella and Neri, 2009]. At a larger scale, dikes have been
proposed to control the evolution of divergent plate boundaries, by directly generating rifting
episodes which represent the culmination of a regional tectonic process of plate spreading
[i.e., Bjornsson et al., 1977; Opheim and Gudmundsson, 1989; Rubin, 1992; Buck et al.,
2006; Wright et al., 2006; Ebinger et al., 2010, and references therein]. In the last decades,
geological, geodetic and geophysical observations detected repeated rifting episodes along
divergent plate boundaries in Iceland (Krafla, 1975-1984; Bjornsson et al., 1977;
Sigurdsson, 1979; Rubin and Pollard, 1988; Rubin, 1992), Tanzania [Calais et al., 2008] and
Afar (Asal-Ghoubet 1978; Dallol, 2004; Dabbahu, 2005-2009). In the Dabbahu case, the mscale reactivation of pre-existing normal faults during dike injection has been directly
witnessed [Rowland et al., 2007], confirming that normal faults may be generated by dikes,
possibly with an upward propagation [Grant and Kattenhorn, 2004; Tentler, 2005]. These
studies lead to a general consensus that magma injection may be a primary factor affecting
the activity and shape of the rift zones. At the surface, these rift zones usually consist of
grabens, eruptive fissures, normal faults and extension fractures. These are organized in
magmatic systems, with a dominant volcano connected to along-rift fissure swarms. At
depth, the analysis of the extinct and exhumed magmatic systems, as in Eastern Iceland,
shows that these almost exclusively consist of dike swarms, locally reaching 5-10% of the
crustal width. However, here the upper tips of the dikes are not connected to the bottom of
any normal fault; indeed, normal faulting appears restricted to the uppermost crust and not
geometrically connected to the underlying dike complex [i.e., Gudmundsson, 1983; Paquet
et al., 2007]. Regional tectonics and seismicity may also enhance fault slip at the surface on
the longer-term, between the main dike-induced rifting episodes [Calais et al., 2008; Ebinger
et al., 2010], possibly explaining a part of the observed deformation pattern at the surface
and the lack of connection between the normal faults and the dikes tips. Therefore, it is still
unclear how much of the cumulative surface deformation observed along divergent plate
boundaries or rift zones may be directly related to dike-induced rifting episodes or to
58
regionally induced plate spreading. A major limitation in understanding the longer-term plate
separation process is represented by the limited observation of its evolution, both at the
surface and at depth. Geodetic and geophysical data commonly capture a single rifting
episode or, seldom, inter-rifting deformation with a time-scale much shorter than that
responsible for regional plate spreading. Conversely, structural field data commonly show
the final, cumulative result of several spreading episodes. Therefore, any understanding of
the processes in between, potentially characterized by variable amounts of tectonic and
magmatic contributions, is poorly constrained.
In order to better understand the structural development of rift zones in general, and those
along divergent plate boundaries in particular, we experimentally simulate progressive dike
injections and analyze the resulting deformation. A few experimental studies have already
analyzed the deformation pattern due to dike intrusion. Some experiments have been
carried out intruding Golden-Syrup or vegetable oil, as magma analogue, into granular
materials, simulating the brittle crust. These models usually aim at studying the deformation
induced by a single dike, rather than a dike swarm. The results show a broad dome, bounded
by reverse faults, above the dike [Mathieu et al., 2008; Abdelmalak et al., 2012]. However,
this deformation pattern does not seem comparable with that observed at rift zones in
nature, where a major subsiding area, bordered by inward dipping faults, forms above the
dike [i.e., Rubin and Pollard, 1988; Wright et al., 2006]. The deformation due to dike
emplacement has also been studied intruding cards within a granular material, obtaining a
fault-bounded depression above the intrusion, more comparable to natural cases [Mastin
and Pollard, 1988]. In this study, we use a derivate of the Mastin and Pollard (1988) setup
as a proxy for our experiments. The main differences with our experiments are: a) we use
crushed silica sand (instead of a flour and sugar mixture) to simulate the host rock and metal
plates for dike analogues; b) we use image correlation techniques to monitor the deformation
of the experiments, obtaining detailed deformation map and time series; c) we test different
modalities and shapes of intrusions, studying the effect of the intrusion depth and its
geometry.
2. Experimental setup and scaling
2.1 Setup and methods
The repeated emplacement of dikes and the development of a dike complex may be
achieved in the experiments through two end-member processes: a) the upward rise of the
59
plate(s) inserted within the sand, maintaining the same thickness throughout the experiment;
b) the thickening of the intrusion complex due to incremental insertion of each plate within a
sleeve, maintaining a constant depth of insertion. In order to test both modes, we perform
two series of experiments in which we simulate a) upward propagating dikes with constant
thickness (setup A) and b) an incrementally thickening dike complex at a fixed depth (setup
B and C). We maintain the same boundary conditions for all the experiments, consisting of
a glass box (25x45x15 cm) hosting a central vertical slot (25x2 cm) where the metal plates
(dike analogues) are intruded. The box is filled with crushed silica sand, as upper crust
analogue (see section 2.2). The incremental dike intrusion is simulated by the progressive
vertical insertion of 20x20 cm iron plates, 0.5 mm thick, into the central slot (Fig. 1).
In the models with upward propagating intrusions (setup A), the iron plates are directly
inserted into the sand through the slot at the base of the sand box (Fig. 1a, c). Each plate is
pushed 2.5-3 cm upward into the sand by incremental steps of 0.5 cm. By using this setup,
the overall initial intrusion thickness during the experiment is kept constant, while the upper
tip of the intrusion propagates upward. In order to test how the intrusion thickness affects
the results, we consider a total thickness of the analogue intrusions of 0.5 mm in model A1
(1 plate) and 2.5 mm in model A2 (5 plates; Table 1). For brevity, here we show only
experiment A2.
In setups B and C, the intrusion depth is fixed and the intrusion thickens. These setups
consist of intrusions of the analogue dikes through a vertical slot at fixed depth (6 cm from
the base of the box). The slot is formed by two parallel and rigid plexiglas sheets (25x25 cm)
(Fig. 1b) jointed on their top and allowed to move horizontally, but not vertically. Therefore,
the intrusion apparatus can only dilate at a fixed depth under the incremental injection of up
to 20 metal plates (each 0.5 mm thick) into the slot. We also vary the final shape of the
intrusion in section view, testing its influence on the deformation pattern. The junction at the
top of the plexiglas plates allows us to modify the final intrusion geometry and also to control
the amount of opening at the top. In setup B, the top of the intrusion complex can dilate up
to 1 cm, similarly to its base (Table 1). In this case, the final shape of the intrusions has a
rectangular shape in section view (Fig. 1d). In setup C, the maximum dilation of the intrusion
top is 0.2 cm, whereas the base can dilate up to 1 cm, providing an overall triangular shape
of the intrusions in section view (Fig. 1e; Table 1). In order to assess the influence of the
intrusion depth, we perform 6 experiments with setup B and 3 with setup C, varying the
intrusion depth from 1 cm to 8 cm. Here we describe three experiments with setup B and
two with setup C (Table 1).
60
To detect the vertical displacement of the model at the surface, we use a laser-scanner
with sub-mm resolution. The output of the scanner acquisitions is an X, Y, Z matrix (where
X and Y are the map coordinates and Z is the vertical component), representing the
topography of the model at the moment of the scanning. We further compute and analyse
the surface deformation at different times, subtracting successive deformation maps. The
profiles along a section perpendicular to the intrusion obtained from the deformation maps
in different time frames allow us to appreciate the vertical displacement during the
experiment. In order to estimate the horizontal displacement (perpendicularly to the intrusion
strike) at the surface, we acquire map images with a digital camera at different periods of
intrusion (Fig. 1). With the Particle Image Velocimetry (PIV) technique, we extract the
displacement field through image cross correlation techniques (i.e. Adam, 2005, Ruch et al.,
2012). The displacement in successive digital images is first computed for small subsectors
(interrogation window); here, we use an interrogation window of 16X16 pixels,
corresponding to 2x2 mm on the model. The merging of the subsector displacements allows
appreciating the global displacement field. This technique is also applied in a few
experiments on images acquired by a second camera from the side, detecting potential
vertical and horizontal displacement fields in section view (Fig. 1). Master optical and laser
images are acquired before the experiments start, as reference for post-processing
deformation analysis. Slave images acquisitions are then systematically taken each 0.5 cm
of vertical intrusion, in setup A experiments, and after each incremental metal plate insertion
(0.5 mm) in setup B and C experiments. Finally, to fully exploit the 3D evolution of faulting
in the subsurface, we wet and cut the final models into slices of 0.5 cm, perpendicularly to
the intrusion axis. This allows us to better analyse the deformation pattern at depth.
2.2 Scaling and materials
Analogue modeling is the scaled reproduction of natural processes to understand the
controlling key mechanisms. We impose a length ratio between model and nature (L* =
Lmod/Lnat) on the order of 10-4 – 10-5 (1 cm in the model corresponds to hundreds of meters
in nature). The density ratio between the model and the nature is ρ*=ρmod/ρnat. The mean
density of the brittle upper crust in nature is 2700 Kg/m3. The mean density of the materials
used to simulate the upper brittle crust in the analogue models is 1500 Kg/m 3. Hence,
ρ*=ρmod/ρnat ~0.6. The gravity ratio between the model and nature is g*= gmod/gnat =1. The
product among ρ*, g* and L* has the dimension of a stress, allowing us to calculate the
stress ratio between the model and nature: σ* = ρ* g* L* = 0.6×10-4 – 0.6×10-5. Then, we
61
assume that natural rocks fail according to the Mohr-Coulomb failure criterion and have a
mean cohesion of 107 Pa, with an angle of internal friction Φ= 35°. Knowing the cohesion
value in nature Cnat and the ratio between the cohesion in the model and in nature C*, we
calculate the cohesion of the material Cmod to use in the model, that is Cmod = C* Cnat= 60 to
600 Pa. The values of cohesion and internal friction are characteristic of the crushed silica
sand (grain size= 40-200 µm) which we use to simulate the upper crust, similarly to Ruch et
al., (2012).
Granular materials are often used in analogue models to simulate the main properties of
the upper crust. Shearing tests on different granular materials demonstrate that they are
characterized by an initial elastic behavior, followed by plastic behavior and then strainhardening prior the onset of brittle failure at the peak strength. The elastic strain stage is
due to diffuse intergranular movements. Shear failure occurs when the grains displacement
focuses along a localized zone [Panien et al., 2006].
2.3 Assumptions and limitations
Our dike analogues (metal plates) are inserted without external or remote stresses acting
on the sand within the box, simulating any regional tectonic contribution. As the aim of our
work is to simulate the effect of discrete rifting episodes, each causing ~10 m of opening in
few months or years [i.e., Wright et al., 2006], we neglect the much slower regional stress
field, in the order of mm/yr.
In principle, the confined box hosting the sand may confine any deformation developed
during the insertion of the plates. However, in all the experiments, the measured surface
deformation diminishes away from the intrusion axis, reaching zero or anyway negligible
vales, within the noise range, towards the sandbox walls. This suggests that any boundary
effect due to the confinement of the sand within the box may be neglected.
In setups B and C, we place a thin silicone layer between the glass of the box and the
lateral termination of the intrusion apparatus, in order to avoid lateral sand dispersion. The
silicone layer slightly reduces the opening of the sides of the intrusion apparatus next to the
box walls. Even though this localized border effect may be considered a limitation of our
experiments, in a few cases we tried to exploit its potential advantage. In fact, any peripheral
cross section directly visible through the plexiglas provides a snapshot of a deformation
pattern less developed than that in the rest of the model, resembling a less advanced
evolutionary stage. This apparently younger stage at the periphery of the intrusions may be
compared to the more advanced stage visible in the cross sections throughout the rest of
62
the model, adding complementary information on the evolution of the experiment. This
complementary information is never used alone and has been always double-checked with
independent data, as for example with experiments ending with younger stages of
deformation in the central part of the model.
Another possible limitation concerns rare and minor asymmetries in the surface
deformation developed in setups B and C. These may be related to the adopted sequence
of manual insertion of the plates. However, any final asymmetry of the models is generally
small, except from experiment B5, and is not significantly altering the results.
A final potential limitation is represented by the difference in the spacing of the elements
of the intrusion complex in model and nature. In nature, the dikes may focus in a narrow
area, but still showing host rock in between [Walker 1958, 1960, 1963; Gudmundsson, 1983;
1995 and references therein; Paquet et al., 2007], whereas in the experiments the metal
plates are all next one another. This implies that the surface deformation in nature may
resemble more complex and asymmetric patterns than in the experiments.
3. Results
Eleven experiments have been performed (Table 1); here we present the results of six
representative models, with setup A (1 model with upward moving metal plates; section 3.1),
B (three models of gradually thickening intrusions with rectangular tip shape; section 3.2)
and C (one model of gradually thickening intrusions with triangular tip shape; section 3.2).
3.1 Setup A: upward propagating intrusions
Experiment A2
We anticipate that an overall consistent deformation pattern is found in setup A models.
This is given by a marked uplift above the intruded plates, independently of their number (1
to 5 plates; Table 1). As the deformation pattern is best resolved with thicker intrusions, here
we show model A2, in which 5 plates are intruded, with final thickness of 2.5 mm,
propagating upwards from the base of the table for 2.5 cm (Fig. 2). Once the intrusions rise
up of 1.5 cm, a 6 cm wide uplifted area (0.5 mm high) forms elongated above the intrusion,
slightly asymmetric with respect to the intrusion axis. A continuous normal fault appears at
the surface above the intrusion, slightly to the right (Fig. 2b). Further moving the plates
upward until the end of the experiment (2.5 cm of rise of the intrusion), the uplift above the
intrusion increases up to ~1.5 mm (Fig. 2c) and includes a discontinuous narrow apical
63
graben (0.5 cm wide and <1 mm deep) bounded by normal faults (Fig. 2c, d). In addition, 23 cm to the side of the graben, a continuous reverse fault striking parallel to the intrusion
develops (Fig. 2c). Section A-A’ (orthogonal to the strike of the intrusion), at the end of the
experiment, reveals that the uplifted area is confined by two mostly blind inward dipping
reverse faults (Fig. 2d) departing close to the intrusion tip. They form a triangular wedge
above the intrusion that moves upward (Fig. 2l). At the wedge center, a shallow, not rooted
and discontinuous apical graben (~0.5 mm deep) is bordered by high angle inward dipping
normal faults with minor displacement with respect to the outer reverse fault. At times, the
graben is replaced by a normal and reverse fault on one side of the intrusion (section B-B’;
Fig. 2e). The horizontal displacement map at the end of the experiment (Fig. 2h,i) shows
that the surface moves ~0.4 cm away orthogonally to the intrusion axis. The PIV analysis
on the side view of the experiment reveals that at depth this horizontal displacement focuses
exactly above the intrusion; this displacement seems a consequence of the doming induced
by the intrusion.
Very similar results, even though less pronounced, are obtained also for model A1,
intruding only 1 plate.
3.2 Setup B and C: thickening intrusions at constant depth
We anticipate that the final deformation pattern strongly depends on the intrusions depth
in setup B experiments (rectangular intrusive complex in section view), but not in setup C
(triangular). Therefore, among the nine experiments, in this section we present three
representative models of setup B and two of setup C (Table 1). For setup B, in order to
appreciate the difference of the fault pattern with the depth to the top of the intrusive
complex, we describe: 1) experiment B1, having the shallowest intrusions depth (1 cm from
the surface); 2) experiment B4, with medium intrusions depth (4 cm); 3) experiment B6, with
the largest intrusions depth (8 cm). For setup C, we describe: 1) experiment C1, with a small
intrusions depth (2 cm from the surface) and a sand cone above one side of the intrusions;
2) experiment C2, with a medium intrusions depth (4 cm).
3.2.1 Experiment B1: shallow rectangular intrusive complex
The map view of the experiment shows that once 2 metal plates are intruded at 1 cm of
depth a depression forms above, at the surface (Fig. 3b). Two parallel fractured areas 1.5
cm distant develop to the sides of the depression. These fractured areas consist of
64
discontinuities formed by coalescing shorter fracture segments parallel to the intrusion axis.
Our resolution does not allow us to define whether they formed as mode I (extension
fractures) or mode II (normal faults), or both. Intruding additional plates (between 2 and 10),
the fracture lengths increase and the fractures merge with neighboring ones. Once 10 plates
(thickening of 5 mm) are intruded, these fracture zones become normal fault segments,
dipping towards the center of the depression, or graben (Fig. 3c). Meanwhile, minor fractures
form within the graben. Inserting further plates, the graben progressively enlarges and the
distance between the border faults increases. The thickening of the intrusion complex
increases the vertical throw and along strike length of the normal faults and the fractures.
Therefore, the segments of normal faults and fractures merge along-strike, forming two
continuous conjugate border normal faults above the intrusion sides. These fault zones are
well developed at the end of the experiment (20 intruded plates, for a total thickening of 1
cm at the intrusions tip; Fig. 3d). At this final stage, secondary inner normal faults developed
from the coalescence of the fractures within the graben (Fig. 3d).
The cross section A-A’ at the end of the experiment shows that the depression is bordered
by two main inward dipping normal faults (conjugate faults), forming a graben-like structure
rooted to both sides of the intrusion’s top, with minor normal faulting within (Fig. 3e). In order
to better characterize the kinematics of the faults, their vertical throw has been measured at
various depths in the section. Both the border and secondary faults have the highest vertical
offset at the surface (2-3 mm), decreasing linearly with depth (Fig 3f).
The vertical deformation map and related time series profiles show that the incremental
thickening of the intrusions induces two uplifted areas to the sides of the intrusions and the
subsidence of the sand above the intrusions, delimiting a 6 cm-wide depression. At the end
of the experiment, the subsidence reaches ~2.5 mm and the uplift ~1.5 mm (Figs. 3g, h).
The horizontal displacement map and profiles show that the widening of the intrusion at
depth causes a progressive divergent displacement of the surface (up to ~2.8 mm) with
respect to the intrusion axis (Fig. 3i, l).
To better capture the cumulative displacement field, we use the section view PIV analysis
during the experiment (Figures 3m, n, o), plotting the vertical and horizontal components of
the displacement. The horizontal component, highest close to intrusions and decreasing
upward, shows a diverging motion of the sand to the sides of the intrusion (Fig. 3n).
However, in the wedge directly above the intrusions the horizontal deformation is negligible
(Fig. 3n, i), consistently with the fact that the maximum horizontal deformation at the surface
is ~5 cm aside from the intrusions axis (Fig. 3l). The vertical component of displacement
65
(Fig. 3m) shows that the host rock to the sides of intrusions moves upward, whereas the
wedge above the intrusions subsides (Fig. 3m, o), similarly to what observed at the surface
(Fig. 3g). These motions result from the lateral push of the sand by the intrusions thickening
at depth, enhancing outward and, in turn, upward displacement of the sand to the intrusions
sides and extension and thinning the wedge above the intrusions (Fig. 3o).
3.2.2 Experiment B4: medium depth rectangular intrusive complex
The map view of the experiment after the insertion of 3 plates at 4 cm of depth shows a
1 cm wide depression at the surface above the intrusions (Fig. 4b). With the insertion of 4
plates, two continuous and parallel faults develop symmetrically to the sides of the
depression (Fig. 4b). After the insertion of 6 plates, two distributed outer fracture zones begin
to form at a distance of 0.7 cm from the previous fault zones (Fig. 4c). After 6 plates, the
deformation at the surface is mostly accommodated by the outer fracture zones, while the
inner faults do not develop further. As in B1, the fracture zones are first formed by minor
segmented structures, progressively increasing their vertical throw and merging along strike
to form a single continuous border fault at each intrusions side (Fig. 4d). Once the border
faults are formed, they are repeatedly activated at each injection, increasing their throw. On
both sides of the intrusive complex, the surface of the model between each pair of internal
and outer faults is tilted inward up to 20-30° (Fig. 4d, l). The vertical displacement profiles
show that during the first stages of the experiment (5 plates) a 0.6 mm deep depression
forms above the intrusions and a 0.1-0.2 mm uplift develops to its sides (Fig. 4f). This early
stage corresponds to the formation of the inner set of faults. The subsidence within the
depression increases when 10 plates are inserted and the outer fracture zones develop. At
this time, the outer uplift becomes asymmetric and increases up to 0.3 mm only at one side
of the intrusions. Intruding more than 10 plates, the subsidence reaches the highest value
(~3.5 mm) above the intrusions (Fig. 4e, f). The maximum uplift reaches 0.6 mm on one side
of the intrusions, delimiting a final central 6 cm wide depression. Despite the uplift
asymmetry, the overall vertical displacements profiles are similar to B1. The horizontal
displacement map and the time series profiles (Fig. 4g, h) show an overall divergent,
extensional motion. However, within the depression there is also an inward horizontal
motion, as indicated by the two symmetric peaks on the profiles close to the central axis, in
correspondence with the tilted blocks enclosed between the outer and inner faults (Fig. 4h).
At the end of the experiment, in order to capture also an intermediate state of evolution,
we consider peripheral cross section A-A’ (Fig. 4d and 4i). Here, as anticipated in section
66
2.3, the deformation is smaller than that in the rest of the model; the faults form only on one
side of the intrusions and are not completely developed. The inner high-angle fault is better
defined at depth, close to the intrusions tip and disappears upward, where the marker layers
are bent, but not faulted (Fig. 4i). This fault dips inward at its base and then outward at its
top, showing an arcuate shape in section view. Thus, the fault is normal below the maximum
curvature and reverse above. The outer inward dipping fault is normal and well defined at
the surface, but gradually disappears with depth. This section shows that inner arcuate fault
and the outer normal fault are two distinct structures.
Section B-B’, along the center of the model (Fig. 4d, l), reveals the final stage of
deformation. Here the two parallel and continuous inner fault zones observed at the surface
correspond to two high-angle arcuate faults, normal at their base and reverse at their top.
The outer normal faults are inward dipping and connect to the reverse faults at their
maximum curvature points.
Fig. 4m shows the variation of the vertical throw of both the normal and arcuate faults in
the central section B-B’. To better understand the kinematics of the faults, we separate the
throws measured above and below their junction. We separately display the throws of the
inner arcuate faults (LR and RR) and those of the outer normal faults (LN and RN) above
the junction from those measured on the left and right arcuate faults below the junction (L
and R) (Fig. 4l, m). Above the junction, the vertical throw of the inner reverse faults (LR and
RR) increases downwards. Conversely, the vertical throw of the outer normal faults (LN and
LR) decreases downwards. This confirms that at each side of the intrusive complex above
the junction the deformation is partitioned on two faults, while below the junction it is
accommodated by a single fault with greater slip.
3.2.3 Experiment B6: deep rectangular intrusive complex
After the insertion of 10-12 plates at a depth of 8 cm, a depression forms at the surface
above the intrusions; this is better defined (0.3 mm of depth) when bordered by two
continuous faults, 1.5 cm distant (Fig. 5b). After the insertion of 14 plates, an outer fracture
zone develops to the outer side of the faults, at a distance of 1.5 cm (Fig. 5c); normal faults
start to form from the coalescence of this fracture zone. Increasing the intrusions thickness
up to 1 cm (20 plates inserted; Fig. 5d), the inner and outer faults at the sides of the intrusions
become better defined, even though still discontinuous. The outer fracture zones broaden
the depression, which becomes 5.5 cm wide and 2.2 mm deep (considering both the uplifted
and the subsided parts). However, here the outer faults are more distant and less defined
67
than in B4 experiment. The maximum absolute subsidence is 1.8 mm above the intrusions
and the uplift is <1 mm to the intrusions sides (Fig. 5e, f). The horizontal surface deformation
map displays a low signal/noise ratio and is not shown.
At the end of the experiment, to capture also an intermediate state of evolution of the
model, we consider peripheral cross section A-A’: this shows a depression with two blind
inner arcuate faults, without any fault at the surface (Fig. 5g), as observed in B4. Conversely,
in cross section B-B’, at the center of the model, the arcuate faults reach the surface (Fig.
5h). The outer normal faults are here absent, appearing only in other cross sections, and
the depression is bordered by the two inner arcuate faults propagating from the intrusions
top. The fault throw of the arcuate faults increases from the surface to 4 cm of depth (area
of maximum curvature); below the throw becomes constant or slightly decreases (Fig. 5i).
3.2.4 Experiment C1: shallow triangular intrusive complex
In this experiment, the intrusive complex has a triangular shape in section view and is
placed at 2 cm of depth. In addition, in order to test also the role of the topography, we have
placed a 3 cm high and 9 cm wide sand cone above the intrusions, towards one side of the
box (Fig. 6a). Intruding one plate, in the flat area away from the cone, a continuous normal
fault appears at the surface ~0.5 cm to one side of the intrusion axis; the fault terminates in
a series of fractures vanishing towards the cone. To the other side, a short fracture forms
on the flat area and a normal fault on the cone (Fig. 6b). Intruding more plates (2 to 5 plates)
shorter fault segments develop on both sides of the intrusions. These define a ~1.2 cm wide
and 0.5 cm deep depression (Fig. 6c, e); minor fractures form within. On the cone, the
previously formed fault grows (Fig. 6c). Intruding more than 5 plates, in the flat area the
border fault segments merge along strike, deepening the depression (Fig. 6e). At the end of
the experiment (20 plates intruded), in the flat area the depression is defined by one
continuous fault on one side and two fault segments on the other. Secondary normal faults
locally develop from the fractures within the depression. At this final stage, the faults are well
developed also on the cone, where they are more distant (~2.5 cm) than in the flat area
(~1.2 cm) (Fig. 6d).
The vertical displacement time series profiles across the graben highlight that the final
subsidence of the depression reaches 2.1 mm, with negligible uplift to the sides (Fig. 6e).
Cross section A-A’ reveals that the depression is a graben formed by two inward dipping
normal faults, geometrically connected to the intrusions top (Fig. 6f). The throw of the faults
is almost constant with depth (Fig. 6g).
68
3.2.5 Experiment C2: medium depth triangular intrusive complex
The map view of the experiment shows that, after the insertion of 2 plates at 4 cm of
depth, a fracture zone forms 2 cm to one side of the intrusions (Fig. 7b). With 8 inserted
plates, a continuous fault forms from the coalescence of these fractures. On the opposite
side, fractures and smaller normal faults start to develop (Fig. 7c), delimiting a <1 mm deep
depression above the intrusions (Fig. 7f). Inserting more than 8 plates, the faults and
fractures on both sides grow and the depression further subsides (Fig. 7f). At the end of the
experiment (20 plates; Fig. 7d) the main fault on the one side is partly flanked by an inner
secondary continuous fault at a distance of 0.8 cm, whereas on the other side the main fault
is flanked by outer shorter faults and fractures at a distance of 0.8 cm (Fig. 7d). The
subsidence of ~1 mm above the intrusion, without any uplift to the sides, delimits a 5.5 cm
wide depression. The subsidence is less than half of that observed in B4, having the same
intrusion depth, but different final intrusion shape (Figs. 4f, 7f).
Section A-A’ in the peripheral part of the depression, against the plexiglas wall, shows
that the faults are normal and inward dipping, disappearing downward (Fig. 7g). Central
section B-B’ confirms that the depression is bordered by two inward dipping normal faults
decreasing their vertical throw with depth (Fig. 7h, i). The vertical throw is here 0.5-1 mm
smaller than that of the normal faults in B4 (Fig. 4m, 7i).
The overall geometry of the graben is similar to that of C1 (shallow intrusions) and C3
(deep intrusions) experiments. The main difference among the three models with setup C
regards the amount of surface deformation, progressively decreasing with the intrusions
depth.
4. Discussion
4.1 Overall deformation pattern and setup relevance
A first significant difference in the deformation pattern has been observed between the
upward propagating intrusions (setup A) and the intrusions thickening at constant depth
(setups B and C). In setup A, the intrusions push the overburden up (Fig. 2l), resulting in a
doming delimited by major reverse faults with shallow and secondary apical graben (Fig. 2c,
d, f, g). Surface uplift bounded by reverse faults has been previously obtained in analogue
models simulating dike emplacement, by injecting vegetable oil in sand [i.e., Abdelmalak et
al., 2012] or viscous intrusions with silicone in sand [Acocella, et al., 2001]. However,
69
analytical, geological and geodetic data [i.e., Rubin and Pollard, 1988, Bonaccorso, 2003;
Buck et al., 2006; Dzurisin, 2006; Wright et al., 2006; Rowland et al., 2007] show that dike
emplacement mainly produces a depression within a broadly uplifted area, not confined by
any reverse fault. This deformation pattern is different from that observed in setup A, where
most of the deformation focuses on the uplifted area bordered by reverse faults. Therefore,
the deformation pattern of setup A is considered as poorly realistic for dike emplacement.
In setups B and C, a main depression, delimited by normal faults, typically forms at the
surface above the intrusions; the depression is bordered by slightly uplifted areas not
confined by any reverse fault (Figs. 3 to 7). The overall deformation pattern of these
experiments is comparable to that obtained by Mastin and Pollard, (1988), even though the
latter do not find any systematic variation on the graben structure with the intrusion depth.
In addition, the image processing results (PIV and laser scanner) provide detailed horizontal
and vertical displacement profiles, showing subsidence above the intrusions and uplift to the
sides, with a maximum horizontal displacement to the side of the intrusions and a minimum
above it (i.e., Fig. 3l). The overall vertical and horizontal deformation pattern of our
experiments is also consistent with that of numerical and analytical models [i.e., Dieterich
and Decker, 1975; Rubin, 1992; Dzurisin, 2006, and references therein]. In addition, our
results are consistent with those obtained by geologic and geodetic data during dike
emplacement [i.e., Pollard et al., 1983; Rubin and Pollard, 1988; Jonsson, 1999; Cervelli et
al., 2002; Wright et al., 2006; Rowland et al., 2007; Keir et al., 2009; Pallister et al., 2010;
Nobile et al., 2012]. For example, the overall horizontal and vertical displacement map and
profiles of experiment B1 (Figs. 3g, h, i, l) are similar to those of the Dabbahu (2005, Afar)
[Wright et al., 2006] and Krafla (1975-1984, Iceland) [Rubin, 1992] rifting episodes. For a
more quantitative study, we compare the measured vertical surface deformation of model
A2 and B2 with that from numerical models [Dieterich and Decker, 1975] and natural
examples of rifting episodes. The latter include the episodes in Afar (Asal-Ghoubbet in 1978
and Dabbahu in 2005) [Ruegg et al., 1979; Wright et al., 2006] and Iceland (at Kulduhverfi
in 1978-1979 and at Namafjall in 1975-1980, both in the Krafla magmatic system) [Rubin,
1992]. The vertical displacements are normalized to the total vertical displacement as a
function of the distance from the intrusion X normalized to the intrusion depth Y (Fig. 8).
The numerical model describes the elastic behavior of the host rock under dike injection,
without any faulting. The natural examples show both the elastic and anelastic behavior of
the crust during diking, where extension fractures and normal faults are also generated. In
order to better compare these behaviors with our data, we consider the vertical displacement
70
in model B2 after the insertion of 2 plates, when the faults are not yet formed (i.e. elastic
behavior, see section 2.2), and after the intrusion of 20 plates, when the faults are well
developed (anelastic behavior). The elastic deformation curve of our model shares an
overall similar pattern to that of the numerical model, with minor negative values (indicating
absolute subsidence) close to the intrusion axis and positive values (absolute uplift) reaching
a maximum followed by a decay away from the intrusions. However, the maximum
deformation in the experiments is closer to the intrusion axis than in the numerical model,
showing a more marked deformation curve. This suggests that the sand in the experiments
carries only a portion of the elastic behavior of the numerical model, due to the diffuse
intergranular slip in the sand particles before developing any shear zone [i.e., Panien et al.,
2006]. The anelastic curve of B2 after the intrusion of 20 plates shows much higher negative
displacements close to the intrusion axis, indicating greater subsidence above the intrusion
than in the numerical model. The subsidence above the intrusions in B2 results from the
thickening of the intrusions, which extends the sand above. The experimental displacement
curve is similar to that measured during diking events in nature with surface faulting and
developing a graben above the intrusions. Such a faulting explains the discrepancy between
the elastic numerical models, where the subsidence appears negligible, and the natural
cases, where the subsidence is larger and bounded by faults. The displacement curve of
experiment A2 does not show any negative value close to the intrusions axis, conversely to
B2, the numerical model and nature. This confirms that setup A is poorly representative to
study the deformation induced by dikes. Overall, the comparison among the displacement
curves suggests that the surface deformation observed in setup B and C is a good
approximation of nature. This indicates that, despite the general limited elasticity of sand,
the intergranular sliding of the sand grains in our experiments may satisfactorily approximate
the overall elastic behavior of the natural rocks for low stresses. Moreover, the use of sand
allows us to enter also the anelastic domain, where the conditions for the stress
concentration are met.
More in general, the different results obtained in our three experimental setups, as well
as this preliminary comparison to natural data, suggest a model for dike propagation
essentially governed by the horizontal opening of the dike walls (as in the setups B and C),
without upward push of the intrusions (as in the setup A). This implies that the end member
process described in experimental setup A has a negligible importance during diking in
nature. This also suggests a general model of dike propagation where the indefinite
repetition of two stages captures the essential steps of a continuum process: a) the dike first
71
focuses the stresses at its tip, inducing a gas-filled mode I fracture of negligible thickness
with regard to that of the dike (Fig. 9a); b) the magma progressively penetrates the fracture,
enlarging it and pushing the two walls aside, until the intruded fracture reaches the same
thickness as that of the main body of the dike (Fig. 9b). The continuous repetition of these
two stages allows the propagation of the dike only through the lateral expansion of its sides,
without any upward push. This model may be applied also to laterally propagating dikes.
4.2. Effect of the intrusions depth and geometry on the deformation
In setup B the deformation pattern varies with the intrusion depth (Fig. 10a). The graben
of the model with shallowest intrusions (1 cm in B1) is bordered by normal faults. For 4 cm
deep intrusions, the depression is bordered by inner arcuate faults and outer inward dipping
normal faults (B4). The deepest intrusions, at 8 cm, form almost exclusively arcuate faults
(B6). Despite the similar final vertical deformation pattern at the surface (a central
depression and uplifted areas to the sides), increasing the intrusion depth decreases the
subsidence and uplift at the surface (Fig. 10a, 11b), resulting in a progressively shallower
depression (Fig. 11c). Moreover, the deepest the intrusions, the highest the distance at the
surface between the peripheral normal faults, forming wider depressions (Fig. 10a, 11a).
In setup C the deformation does not vary with the injection depth (Fig. 10b). The graben
is always bordered by two conjugate normal faults, whose distance at the surface increases
with the intrusion depth (Fig. 11a). The final vertical displacement profiles also show a
central subsidence and minor lateral uplift, whose extent decreases with the intrusion depth
(Fig. 10b, 11b). However, differently from setup B, in C the subsidence is ~30-50% lower
and the uplift is negligible (Fig. 11b). This is also in accordance with the fact that, for an
equal depth of intrusion, the fault throws measured in setup B are 1-2 mm larger than in C
(Figs. 4m, 7i). All these differences suggest that the subsidence, and especially the uplift at
the surface, depend on the intrusion depth and overall configuration.
In order to better understand the effect of the thickening of the intrusions on the sand, we
focus on the cross correlation imagery from the side view of model B1 (PIV analysis; Fig.
3m, n, o). This shows how the thickening of the intrusive complex promotes subsidence
above and uplift to the sides by means of lateral compression of the material. The lateral
compression (or lateral displacement) of the sand and the related subsidence and uplift at
the surface are proportional to the intrusion thickening. Since the overall thickening at the
intrusive complex tip is higher in setup B than C, we expect higher surface deformation in
setup B. Hence, we suggest that the lack of uplift and the shallower depression in setup C
72
(Fig. 10, 11c) is caused by the lower final opening of the upper tip of the intrusion apparatus
in setup C (0.2 cm) than in B (1 cm). This indicates that the final shape of the dike complex
in section view (rectangular or triangular) may affect the deformation at the surface and
shape the topography. In addition, the increase in the intrusions depth increases the
lithostatic pressure above, which in turn limits the uplift of the surface to the sides of the
intrusions. Therefore, the higher the intrusion depth, the lower the uplift to the sides (Figs.
10, 11). These considerations suggest that the overall deformation pattern is controlled by
the interplay between the intrusion depth and its thickness.
4.3. Fault propagation
Two types of faults form in our experiments, differing in geometry and kinematics: normal
faults and arcuate normal/reverse faults (Fig. 12). The study of the cross sections and fault
throws allows us to better define their kinematics and development.
4.3.1. Normal faulting
At the surface, at the side of the dike the maximum compressive stress (σ1) is sub vertical
[Rubin and Pollard, 1988; Patton and Fletcher, 1995; Bonafede and Danesi, 1997;
Gudmundsson and Loetveit, 2005]. This enhances the formation of vertical extension
fractures and the high-angle normal faults observed in the experiments. However, one
debated aspect regarding faulting induced by diking is whether the faults nucleate at the
surface or at depth. In our models, the throw of the normal faults in the central portions of
the models is usually highest at the surface, decreasing downwards (Figs. 3f, 4m, 7i). In the
peripheral cross sections, where the deformation is smaller, the normal faults are still well
defined at the surface and gradually disappear at depth (Fig. 4i, 7g).
In order to better investigate the conditions leading to the nucleation of the normal faults,
we calculate the 2D Coulomb failure stress change on optimally oriented faults due to a
vertical opening fracture, simulating a dike, in a homogeneous elastic medium [Okada,
1992]. The fracture is composed of 10x6 rectangular dislocations at the center, on which a
uniform pressure of 5 MPa is applied. The Coulomb stress change is highest at the upper
tip of the opening fracture (Fig. 12g). However, due to the effect of the free surface, two
minor areas with relatively high Coulomb stress change (~5 MPa) are visible also close to
the surface, to the sides of the fracture (Fig. 12g). Failure nucleation at the surface in a
homogeneous medium may be favored by the least lithostatic pressure close to the surface.
73
In order to test this possibility, we compute the total Coulomb stress, where the total normal
stress is equal to the lithostatic pressure, plus the normal stress change induced by the
fracture (Fig. 12h). This shows that the likelihood of fault nucleation is highest at the surface
in two wide areas at the side of the dike and at the upper tip, reaching ~20 MPa (Fig. 12h).
Considering that the mathematical singularity at the fracture tip is not realistic, as non-elastic
processes will quickly dissipate such high stresses, the Coulomb stress models suggest that
normal faulting is enhanced at the surface, to the sides of the dike. Therefore, both the throw
distribution of the normal faults in the experiments and the Coulomb stress models suggest
that the dike-induced high-angle normal faults nucleate close to the surface and propagate
downwards.
Surface fracturing during a rifting episode have been often associated with upward
propagating normal faults that nucleate from the dike tip [Grant and Kattenhorn, 2004;
Tentler, 2005; Rowland et al., 2007]. Our results show that dike emplacement likely induces
downward propagating normal faults nucleating at the surface (Fig. 12a, b, c). This
downward propagation of the normal faults is also proposed by previous studies
[Gudmundsson and Backstrom 1991; Forslund and Gudmundsson, 1992; Acocella et al.,
2003]. Moreover, our modelling evidence is consistent with field evidence from eroded
portions of rift zones in Iceland, where any direct connection of normal faults from the upper
tip of dikes is lacking. Therefore, our experiments provide a possible solution to reconcile
these apparent contradictions in the geometric, kinematic and causal distribution of dikes
and normal faults below rift zones.
4.3.2. Arcuate normal/reverse faults
The inner arcuate faults are observed above rectangular intrusive complexes with depth
≥2 cm (models B3, B4, B5 and B6). The fault throws decreasing towards the surface and
the peripheral cross sections suggest that these faults propagate upward from the intrusions
top (Figs. 4i, m and 5g, i). These faults result from the differential vertical movement (uplift
to sides and subsidence above the intrusions) induced by the intrusions. For example, when
the faults form in experiment B4, this vertical movement is estimated as ~1 mm at the
surface, or ~10% of the thickness of the injected plates. The transition from subsidence to
uplift is first accommodated by a broad inward tilt of the surface, subsequently replaced by
anelastic deformation; when the arcuate high-angle faults propagate upward, they reach the
surface as reverse faults (Fig. 12d, e). High-angle reverse faults are commonly found with
differential vertical movements of adjacent crustal blocks, due to an upward local rotation of
74
the maximum compressive stress, potentially explained by the vault effect above a cavity
or, generally speaking, above a subsiding portion of crust. These structures have been
observed in sedimentary basins, basement faulting, calderas, and mining cavities [Wise,
1963; Prucha et al., 1965; Given, 1973; Vendeville, 1988; Roche et al., 2000; Merle et al.,
2001; Abe et al., 2011]. However, a novelty of our study is that the reverse faults may also
result from dike emplacement, not necessarily from the activity of any high-angle dip-slip
fault, caldera or mining collapse. While in the latter cases the maximum compressive stress
is sub vertical at the surface, in our experiments the maximum compressive stress at the
upper plate tip is sub-horizontal. While approaching the surface, the maximum stress
rotates, becoming inward dipping, subvertical, and finally outward dipping, thus explaining
the arcuate geometry of the faults (Fig. 12i; i.e., Patton and Fletcher, 1995; Bonafede and
Danesi, 1997]. The absolute (above the intrusions) and relative (on the uplifted portions to
the sides of the intrusions) subsidence induced by the intruded plates causes the inward
horizontal displacement of the overburden above the intrusions. This develops the outer
faults, acting as gravitational structures resulting from the activity of the reverse faults (Fig.
12e). The outer normal faults propagate downward, conversely to previous studies on highangle reverse faults [Roche et al., 2000, and references therein], where they propagate
upward. Once the normal outer faults form, they merge along dip with the bottom (inward
dipping) portion of the arcuate faults, forming a continuous structure with localized
displacement, locking the reverse faults (Fig. 12f).
4.4. Comparison to rift zones
Here we compare the experimental deformation pattern with that observed in nature along
rift zones. According to the scaling factor used in the experiments, a single plate insertion
corresponds to the emplacement of a dike ~101 m thick, while the final cumulative thickness
at the tip of the intrusive complex is of ~10 2 m in setup B and between ~101 to ~102 m in
setup C. This implies that the early stage of the models (1-2 inserted plates) can be
compared with nature on single rifting episodes, where one or more dikes, for a total
thickness of ~10 m, emplace over weeks to a very few years. Alternatively, the final stage
of the models can be compared with nature on several rifting episodes, where multiple dikes,
for a total thickness of ~100 m, emplace over centuries or more. The latter condition is for
example met in the eroded portions of rift zones, as in Eastern Iceland, where the cumulative
dike thickness of 102-103 m (5-10% of fissure swarm width) [Walker 1958, 1960, 1963;
Gudmundsson, 1983; 1995 and references therein; Paquet et al., 2007] is comparable with
75
the final opening of setup B. In comparing the early stages of setup B to single rifting
episodes we have to face a resolution problem: while any structure can be observed in detail
on the field, at the experimental scale it can only be appreciated in its very general features
above our limit of resolution. This discrepancy may lead to a limited comparison between
the details of our experiments and nature. Our early stage experiments will be mainly
compared to the 1783-1784 Lakagigar rifting episode, while the final stages will be
compared to the structure of the mature magmatic systems of Iceland and the Main
Ethiopian Rift, also testing the possibility that these may have been produced by several
rifting episodes.
4.4.1. Single rifting episodes
Here we compare the very early stage of deformation of our experiments, obtained
inserting 1-2 plates in setup B or more plates in setup C, with single rifting episodes. These
can cause an opening of up to ~10 m at the surface, as at Krafla in 1975-1984, AssalGhoubbet in 1973-1979; Dabbhau 2005-2010 [Ruegg et al., 1979 ; Tryggvason, 1980;
Rubin, 1992 and references therein; Wright et al., 2006; Ayele et al., 2007; Grandin et al.,
2009; Ebinger et al., 2010]. We mainly focus on the 1783-1784 Lakagigar rifting episode, as
this occurred in a previously undeformed area, so that the visible surface deformation can
be entirely related to the emplacement of the 1783-1784 dikes [Thordarson and Self, 1993].
The rifting episode along the 27 km long Lakagigar eruptive fissure consisted of ~10 intrusive
events. The NE-SW trending fissure comes out in a nearly flat area, locally intersecting the
~200 m higher and older Laki hyaloclastite hill [i.e., Thordardson and Self, 1993, and
references therein]. The fissure is partly bounded by conjugate normal faults with mean dip
of 70° and maximum throw up to 6-10 m, forming a 150 to 450 m wide graben. The fissure
is partly interrupted in correspondence to Laki. Here, even though the normal faults
decrease their vertical displacement, they climb the hill, becoming ~700-800 m distant.
Considering the mean fault dip and width of the graben north and south of Laki, we suggest
a mean depth d of the intrusion top of ~400 m to the NE and SW sides of the mountain (with
d=1/2w tanα; where d is the intrusion depth; w is the graben width, α is the fault dip). This
depth refers to the top of the dike at the moment of faulting the surface; the dike has then
subsequently risen, reaching the surface and erupting.
The overall faulting pattern is similar to that obtained in the early stage models of setup
B or final stage models of setup C. Similarly to Laki, the B1 and C1 models show a narrow
graben at the surface bordered by inward dipping normal faults (Fig. 3, 6, 13b). Model C1,
76
with the 3 cm high sand cone above the intrusion, also allows us to simulate the effect of
diking on a topography of 102 to 103 m, similar to Laki. In C1, the constant distance between
the faults in the flat area (1.2 cm) is more than doubled in correspondence with the cone
(2.8 cm), similarly to what observed at Laki (Fig. 6d and 13c). This comparison confirms that
a graben widens when developing on a relief; this is due to the locally increased distance
between the dike tip and the surface, so that the deeper the dike, the wider the graben.
While the graben structures are well developed on Laki, the fissure becomes interrupted,
with minor vents on the NW portion of the hill. Laki provides therefore an interesting example
of how dike-induced deformation may be influenced by a relief and how an eruptive fissure
may interrupt against a relief. Similarly to Lakagigar, graben formation has been also
observed during diking episodes at Dabbahu, Afar [i.e., Rowland et al., 2007], highlighting
an overall similar deformation pattern to our experiments.
4.4.2. Multiple rifting episodes
The final stages of setup B experiments (intrusion of 20 plates with rectangular intrusive
complex in section view) can be compared to magmatic systems along divergent plate
boundaries undergoing repeated rifting episodes. In particular, here we mainly refer to the
Krafla magmatic system, in the Northern Volcanic Zone of Iceland.
The experiments show an overall symmetric depression bounded by normal faults,
similarly to the graben-like structures commonly observed in nature. As anticipated, a minor
difference is related to the fact that rift zones in nature may not simply consist of a single
symmetric graben, but of asymmetric and/or multiple and nested grabens. This is for
example the case of the Krafla magmatic system: at the surface, its central portion consists
of a 2-3 km wide graben, with several normal faults within, defining 30-300 m wide minor
nested grabens (Fig. 14a) [Opheim and Gudmundsson, 1989]. The border faults are usually
vertical at the surface, showing the highest displacements of several tens of m and a tensile
area of several meters between the footwall and hanging wall; the footwall is usually sub
horizontal, whereas the hanging wall may be sub horizontal or 20°-30° inward tilted (Fig.
14b) [Opheim and Gudmundsson, 1989; Angelier et al., 1997]. Seismic tomography at Krafla
shows that the depth to the top of the high velocity layer varies along the strike of the
magmatic system, being shallower (~2-3 km deep) at the center and deepening (~6-7 km
deep) towards the northern and southern peripheries (Fig. 13 b) [Brandsdóttir et al., 1997].
This suggests a deeper location of the intrusive complex beneath the northern and southern
terminations of the magmatic system.
77
Our models suggest that the overall deformation pattern depends on the depth of the
intrusion with respect to the thickness of the dike complex. The shallower intrusion
experiment B1 shows that, similarly to the central portion of the Krafla magmatic system, the
graben is delimited by two inward dipping normal faults in which most of the deformation
focuses, with minor faults and fractures within (Fig. 14c, h). The deeper intrusion
experiments of setup B also show inner arcuate faults with a reverse kinematics in their
upper portion, leading to contraction close to the surface (i.e., experiment B5 in Fig. 14i). At
Krafla, similar contractional structures are found at the base of the tilted hanging wall of the
normal faults several km north of the caldera and at the southern tip of the magmatic system,
where the intrusive complex appears deeper (Fig. 14a) [Trippanera et al., 2014]. An example
of contraction at the surface is observed along the 2 km long Grjotagja fault, in the Myvatn
area (Fig. 14a, d, e, f). The fault displays 5-6 m of vertical throw with well-defined titled
hanging wall, separated from the footwall by a 2 m open fracture (Fig. 14d). The contraction
at the base of the titled hanging wall exhibits a 1 m wide cylindrical-like fold, whose axis is
parallel to the fault strike (Fig. 14e, f). Similar contractional structures have been also
observed at Vogar (Reykjanes Peninsula, Iceland) and Fantale magmatic systems (Main
Ethiopian Rift) [Trippanera et al., 2014]. Contractional structures are thus present in various
rift systems. The rift segments of Vogar and Fantale currently undergo a lower amount of
extension (~6 mm/yr), much lower than that along the central portion of Krafla, of 22-24
mm/yr [Perlt and Heinert, 2006]. As the depth to the axial melt zone along divergent plate
boundaries is usually proportional to the extension rate [Purdy et al., 1991; Biggs et al.,
2009], both the Vogar and Fantale rift segments probably have a deeper intrusive complex
than the central portion of the more active and extending Krafla system. At Vogar, the
possibility of a deeper intrusive complex is also supported by the dominant eruptive fissures
and absence of polygenic volcanoes, suggesting the lack of a proper shallow magma
chamber (Gee et al., 2000). In synthesis, it appears that the contractional structures found
at the surface in poorly extending (Fantale) and magmatically immature (Vogar) systems, or
at the periphery of more developed systems (Krafla), may be the expression of a deeper
dike complex, similarly to what observed in setup B experiments. This suggests that any
diffuse contraction along divergent plate boundaries may largely depend upon the mean
depth of the dike complex and/or the amount of opening of the rift segment.
These cases allow a qualitative estimate of the geometric and kinematic boundary
conditions for the formation of contractional structures along rifts. In order to be more
quantitative on the geometrical conditions, we attempt to estimate the minimum depth of the
78
dike complex to have contraction at the surface. In setup B, the normal faults reach the
surface after the intrusion of ~4 plates (intrusion thickness 2 mm). This indicates that, in the
shallower intrusion experiments (B1) with only normal faulting, the depth/thickness (D/T)
ratio of the top of the dikes at the moment of faulting is ≤10 1. Conversely, in the deeper
intrusion experiments, the D/T ratio at the moment of faulting is >101. Transferring this to
nature, assuming a cumulative mean thickness of the dikes during a rifting episode of 101
m (excluding any host rock in between the dikes), the minimum depth to develop normal
faults at the surface is 101 m; if arcuate faults also develop, the minimum depth becomes
102 m. However, our models show that asymmetries in the graben structure may occur, as
in B5 (Fig. 14i): here the border faults on one side of the graben consist of an inner arcuate
and an outer normal fault (as expected for this intrusion depth) and, on the other side, of a
single high angle normal fault with negligible hanging wall tilt (Fig. 14h). This indicates that
normal faults without any contraction may be also found where the dike complex is expected
to be relatively deep, as at Reykjanes (Fig. 14g). More in general, this suggests that,
depending on the overall rift structure, contractional structures and hanging wall tilt may not
be a rule in rifts with deeper intrusion.
4.4.3. Intrusion depth vs. thickness
To better understand how the interplay between the intrusions thickness and depth
influences the deformation, we compare the depth to thickness ratio (D/T) of the intrusion in
the models and in nature.
In the models, the estimate of the D/T ratio is based on the thickness values of the
intrusive complex reached when the faults appear at the surface. The D/T is overall constant
(always on the order of ~101), resulting in a linear trend for each setup (Fig. 15a). In
particular, the D/T value is ~20 in setup B and ~ 40 in setup C. These results are obtained
with similar boundary conditions in the experiments (including homogenous sand, without
layering and preexisting fractures), so that, the difference in D/T between the two setups is
only due to the different shape of the intrusive complex; this highlights that surface fracturing
occurs inserting fewer plates for setup C (rectangular intrusion) than in B (triangular
intrusion). This also suggests that the intrusion shape not only controls the pattern and
magnitude of the deformation at the surface (see section 4.2), but it also affects the
possibility to develop faults.
The mean depth to the top of the dikes and their thickness along divergent plate
boundaries and volcanic rift zones, where fault formation or reactivation have been
79
observed, are summarized in Fig. 15b [Tryggvason, 1980; Pollard, 1983; Stein et al., 1991;
Rubin, 1992; Jonsson et al., 1997; Dvorak and Dzurisin, 1997 and references therein;
Chadwick and Embley, 1998; Pallister et al., 2000; Acocella and Neri, 2003; Wright et al.,
2006]. The estimated D/T for these natural cases is highly variable (from ~10 1 to ~103), even
though clustering around ~102, with an overall proportion between the depth and thickness
of the dikes (Fig. 15b). This higher variability of D/T in nature may depend on the nonhomogenous boundary conditions, including variations in the elastic properties of the host
rock, layering (i.e., stress barriers due to stiffness contrast between layers) and pre-existing
fractures. The latter may be particularly important, since the presence of faults affects the
final surface deformation (Fig. 8). Moreover, fault reactivation requires lower energy than
fault formation, suggesting that thinner or deeper dikes may easily reactivate preexisting
structures. In addition, the variability of D/T may depend upon the uncertainty in estimating
the intrusion thickness and depth in nature, as for instance at Lakagigar and Trollagigar.
Despite the variability of the natural data, there is an overall agreement between models
and nature, highlighted by a proportion between the depth to the top of the dikes and their
thickness; this suggests that the deeper is a dike, the thicker this has to be in order to fracture
the surface. Moreover, in both experiments and nature dike-induced faulting at the surface
depends on D/T and, in particular, occurs when D/T~10 2.
This study has shown several consistencies between our dike experiments and the
overall structure of divergent plate boundaries, at various scales, from the single diking
episode to the entire portion of a rift zone. The fact that most of the structure of a rift, at
various scales, can be explained by repeated dike injections highlights the importance of
diking in constraining the surface deformation and thus in shaping the structure and
topography of divergent plate boundaries. Regional extension is certainly important for the
long-term evolution of a divergent plate boundary, providing the required conditions for the
rise and emplacement of magma at various levels; it probably also plays an important role,
through fault activity and seismicity, during inter-rifting events. However, our results suggest
that most, if not all, the deformation pattern along divergent plate boundaries may be
acquired through repeated dike injection, not requiring any direct tectonic contribute, so that
rifting may be approximated by multiple diking episodes.
80
5. Conclusions
1) Experiments from setup A create a doming delimited by reverse faults, with a secondary
apical graben. In experiments with setup B and C, a depression flanked by two uplifted
areas is bordered by inward dipping normal faults propagating downward and, for deeper
intrusions in B, also by inner faults, reverse at the surface.
2) The comparison between models and nature reveals that setups B and C realistically
simulate dike emplacement in the upper crust, conversely to the less realistic setup A.
The comparison between these setups suggests that the propagation of a dike consists
of the indefinite repetition of two essential steps: a) initial development of a mode I fracture
at the dike tip; b) lateral opening of the fracture by magma intrusion (as simulated in
setups B and C), without any direct upward push of the magma (as in setup A).
3) The elastic and anelastic components of surface deformation shown by experiments of
setup B and C are consistent with the elastic component suggested by numerical models
and the anelastic component shown by geodetic data during diking events in nature,
respectively.
4) The magnitude and pattern of the deformation in the experiments, both at depth and at
the surface, depend on depth to the top of the intrusion. The most striking difference is
that in setup B shallower intrusions promote normal faults propagating downward,
whereas deeper intrusions also promote faults propagating upwards becoming reverse
close to the surface.
5) While we confirm that dikes may generate normal faults, we also show that the latter
propagate from the surface downward. This explains the common lack of any geometric
connection between the base of normal faults and the upper tips of dikes in the eroded
portions of rift zones, as in Iceland.
6) There is a close similarity between our experiments and divergent plate boundaries, at
various scales. The deformation pattern observed in the early stages of our setups B and
C, where a graben formed by normal faults forms above the dike(s), is consistent with
that during single rifting episodes (i.e., Lakagigar, Iceland). In addition, the final stages of
setup B display a graben structure bounded by outer inward dipping normal faults (for
shallower dikes), and also with inner arcuate reverse faults (for deeper dikes), similarly to
what observed along rift segments in Iceland and the MER. In particular, the deeper
intrusions of setup B allow us to explain the formation of the contractional structures
observed at the base of the tilted hanging wall along poorly extending rift segments,
(Fantale, MER), at times lacking a proper shallow magma chamber (Vogar, Iceland) and
81
at the periphery of more active rift segments (as Grjotagja, Krafla, Iceland). Both nature
and models suggest that this contraction develops above a deeper dike complex (at least
~102 meters deep).
7) The onset of formation of the faults at the surface during dike injection depends upon the
depth to the top of the dike D and its thickness T. There is an overall proportion between
the depth to the top of the dike and its thickness in experiments and nature, suggesting
that the deeper is the dike, the thicker this has to be in order to fracture the surface. The
agreement between models and nature also suggests that dike-induced faulting at the
surface occurs when D/T~102.
8) The fact that most of the structure of a rift, at various scales, can be explained by repeated
dike injections highlights the importance of diking in constraining the surface deformation
and thus in shaping divergent plate boundaries. Despite the importance of regional
tectonics, our results suggest that most, if not all, of the deformation pattern along
divergent plate boundaries may be ultimately acquired through repeated dike injections,
so that rifting may be approximated by multiple diking episodes.
Acknowledgements
Andrea Giordano helped in the setting up of the experiments.
Financed with PRIN 2009 funds (2009H37M59, responsible V. Acocella). Any user can
access the data of this work by contacting the corresponding author.
References
Abdelmalak, M. M., Mourgues, R., Galland, O., & Bureau, D. (2012). Fracture mode
analysis and related surface deformation during dyke intrusion : Results from 2D
experimental modelling. Earth and Planetary Science Letters, 359-360, 93–105.
Doi:10.1016/j.epsl.2012.10.008
Abe, S., van Gent, H., & Urai, J. L. (2011). DEM simulation of normal faults in cohesive
materials. Tectonophysics, 512(1-4), 12–21. Doi:10.1016/j.tecto.2011.09.008
Acocella, V. (2007). Understanding caldera structure and development: An overview of
analogue models compared to natural calderas. Earth-Science Reviews, 85(3-4), 125–160.
Doi:10.1016/j.earscirev.2007.08.004
Acocella, V. (2014). Structural control on magmatism along divergent and convergent
plate boundaries: Overview, model, problems. Earth-Science Reviews, 136, 226–288.
Doi:10.1016/j.earscirev.2014.05.006
Acocella, V., & Neri, M. (2009). Dike propagation in volcanic edifices: Overview and
possible developments. Tectonophysics, 471(1-2), 67–77. Doi:10.1016/j.tecto.2008.10.002
Acocella, V., Neri, M., & Sulpizio, R. (2008). Dike propagation within active central
volcanic edifices: constraints from Somma-Vesuvius, Etna and analogue models. Bulletin of
Volcanology, 71(2), 219–223. Doi: 10.1007/s00445-008-0258-2
82
Angelier, J., Bergerat, F., Dauteuil, O., & Villemin, T. (1997). Effective tension-shear
relationships in extensional fissure swarms, axial rift zone of northeastern Iceland. Journal
of Structural Geology, 19(5), 673–685. Doi: 10.1016/S0191-8141(96)00106-X
Ayele, A., Jacques, E., Kassim, M., Kidane, T., Omar, A., Tait, S., King, G. (2007). The
volcano–seismic crisis in Afar, Ethiopia, starting September 2005. Earth and Planetary
Science Letters, 255(1-2), 177–187. Doi:10.1016/j.epsl.2006.12.014
Barisin, I., Leprince, S., Parsons, B., & Wright, T. (2009). Surface displacements in the
September 2005 Afar rifting event from satellite image matching : asymmetric uplift and
faulting. Geophysical Research Letters, 36, 1–6. Doi: 10.1029/2008GL036431
Bjornsson, A., Saemundsson, K., Einarsson, P., Tryggvason, E., & Gronvold, K. (1977).
Current rifting episode in north Iceland. Nature, 266, 318–322.
Bonaccorso, a. (2003). Dynamics of the December 2002 flank failure and tsunami at
Stromboli volcano inferred by volcanological and geophysical observations. Geophysical
Research Letters, 30(18), 1941. Doi: 10.1029/2003GL017702
Bonafede, M., & Danesi, S. (1997). Near-field modifications of stress induced by dyke
injection at shallow depth. Geophysical Journal International, 130(2), 435–448.
Doi:10.1111/j.1365-246X.1997.tb05659.x
Brandsdóttir, B., Menke, W. H., Einarsson, P., White, R. S., & Staples, R. K. (1997).
Faroe-Iceland Ridge Experiment. 2. Crustal structure of the Krafla central volcano. Journal
of Geophysical Research, 102(B4), 7867–7886.
Biggs, J., Amelung, F., Gourmelen, N., Dixon, T. H., & Kim, S.-W. (2009). InSAR
observations of 2007 Tanzania rifting episode reveal mixed fault and dyke extension in an
immature continental rift. Geophysical Journal International, 179(1), 549–558.
doi:10.1111/j.1365-246X.2009.04262.x
Buck, W. R., Einarsson, P., & Brandsdóttir, B. (2006). Tectonic stress and magma
chamber size as controls on dike propagation: Constraints from the 1975–1984 Krafla rifting
episode. Journal of Geophysical Research, 111(B12), B12404. Doi: 10.1029/2005JB003879
Calais, E., d’Oreye, N., Albaric, J., Deschamps, A., Delvaux, D., Déverchère, J., Wauthier,
C. (2008). Strain accommodation by slow slip and dyking in a youthful continental rift, East
Africa. Nature, 456(7223), 783–7. Doi: 10.1038/nature07478
Cervelli, P., Segall, P., Johnson, K., Lisowski, M., & Miklius, A. (2002). Sudden aseismic
fault slip on the south flank of Kilauea volcano. Nature, 415(6875), 1014–8. Doi:
10.1038/4151014a
Dieterich, J. H., & Decker, R. W. (1975). Finite element modelling of surface deformation
associated with volcanism. Journal of Geophysical Research, 80(29), 4094–4102. Doi:
10.1029/JB080i029p04094
Dvorak, J., A, and Dzurisin, D. (1997). Volcano geodesy: the research for magma
reservoirs and the formation of eruptive vents. Reviews of Geophysics, 35(3), 343-384.
Dzurisin, D. (2006). Volcano deformation: geodetic monitoring techniques. (P. Blondel,
Ed.) (Springer-P., p. 441).
Ebinger, C., Ayele, A., Keir, D., Rowland, J. V, Yirgu, G., Wright, T., Hamling, I. (2010).
Length and Timescales of Rift Faulting and Magma Intrusion : The Afar Rifting Cycle from
2005 to Present. Annual Rev. Planet. Sci., 38, 439–466. Doi: 10.1146/annurev-earth040809-152333
Ebinger, C. J., & Casey, M. (2001). Continental breakup in magmatic provinces : An
Ethiopian example. Geology, 29, 527–530. Doi: 10.1130/0091-7613(2001)029<0527
Ebinger, C. J., Wijk, J. Van, & Keir, D. (2013). The time scales of continental rifting :
Implications for global processes. Geological Society of America, Special Paper, 500(11),
1–26. Doi:10.1130/2013.2500(11).
Forslund, T., & Agust Gudmundsson. (1991). Crustal spreading due to dikes and faults
in SW island. Journal of Structural Geology, 13(4), 443–457.
83
Gee, M. A. M., Taylor, R. N., Thirlwall, M. F., & Murton, B. J. (2000). Axial magma
reservoirs located by variation in lava chemistry along Iceland’s mid-ocean ridge. Geology,
28, 699–702. Doi: 10.1130/0091-7613(2000)28<699
Grandin, R., Socquet, A., Binet, R., Klinger, Y., Jacques, E., de Chabalier, J.-B., Pinzuti,
P. (2009). September 2005 Manda Hararo-Dabbahu rifting event, Afar (Ethiopia):
Constraints provided by geodetic data. Journal of Geophysical Research, 114, 1–20. Doi:
10.1029/2008JB005843
Grant, J. V, & Kattenhorn, S. A. (2004). Evolution of vertical faults at an extensional plate
boundary, southwest Iceland. Journal of Structural Geology, 26, 537–557.
Doi:10.1016/j.jsg.2003.07.003
Gudmundsson, A. (1987). Tectonics of the Thingvellir fissure swarm, SW Iceland. Journal
of Structural Geology, 9(1), 61–69.
Gudmundsson, A. (1995). Infrastructure and mechanics of volcanic systems in Iceland.
Journal of Volcanology and Geothermal Research, 64, 1–22.
Gudmundsson, A., & Backstrom, K. (1991). Structure and development of the Sveinagja
graben, Northeast Iceland. Tectonophysics, 200, 111–125.
Gudmundsson, A., & Loetveit, I. F. (2005). Dyke emplacement in a layered and faulted
rift zone. Journal of Volcanology and Geothermal Research, 144, 311–327.
Helgason, J., & Zentilli, M. (1985). Field characteristic of laterally emplaced dikes:
anatomy of an exhumed Miocene dike swarm in Reydarfjordur, Eastern Iceland.
Tectonophysics, 115, 247–274.
Hoek, E. (1983). Strength of jointed rock masses. Géotechnique, 23(3), 187–223.
Doi:Doi:10.1680/geot.1983.33.3.187.
Hubbert, M. K. (1937). Theory of scale models as applied to the study of geologic
structures. Geological Society of America Bulletin, 48, 1459–1520.
Jónsson, S., Einarsson, P., & Sigmundsson, F. (1997). Extension across a divergent plate
boundary, the Eastern Volcanic Rift Zone, south Iceland, 1967–1994, observed with GPS
and electronic distance measurements. Journal of Geophysical Research, 102(B6), 11,913–
11,929. Doi: 10.1029/96JB03893
Jónsson, S., Zebker, H., Cervelli, P., Segall, P., Garbeil, H., Mouginis-mark, P., &
Rowland, S. (1999). A shallow-dipping dike fed the 1995 flank eruption at Fernandina
Volcano, Galàpagos, observed by satellite radar interferometry. Geophysical Research
Letters, 26(8), 1077–1080.
Keir, D., Hamling, I. J., Ayele, a., Calais, E., Ebinger, C., Wright, T. J., Bennati, L. (2009).
Evidence for focused magmatic accretion at segment centers from lateral dike injections
captured beneath the Red Sea rift in Afar. Geology, 37(1), 59–62. Doi:10.1130/G25147A.1
King, G. C. P., Stein, R. S., & Lin, J. (1994). Static stress changes and the triggering of
earthquakes. Bulletin of the Seismological Society of America, 84(3), 935–953.
Mastin, L. G., & Pollard, D. D. (1988). Surface deformation and shallow dike intrusion
processes at Inyo Craters, Long Valley, California. Journal of Geophysical Research,
93(B11), 13,221–13,235.
Mathieu, L., van Wyk de Vries, B., Holohan, E. P., & Troll, V. R. (2008). Dykes, cups,
saucers and sills: Analogue experiments on magma intrusion into brittle rocks. Earth and
Planetary Science Letters, 271(1-4), 1–13. Doi:10.1016/j.epsl.2008.02.020
Merle, O., & Vendeville. (1995). Experimental modelling of thin-skinned shortening
around magmatic intrusions. Bulletin of Volcanology, 57, 33–43.
Merle, O., Vidal, N., & Vries, B. V. W. De. (2001). Experiments on vertical basement fault
reactivation below volcanoes. Journal of Geophysical Research, 106(B2), 2153–2162.
Nobile, A., Pagli, C., Keir, D., Wright, T. J., Ayele, A., Ruch, J., & Acocella, V. (2012).
Dike-fault interaction during the 2004 Dallol intrusion at the northern edge of the Erta Ale
84
Ridge
(Afar,
Ethiopia).
Geophysical
Research
Letters,
39,
1–6.
Doi:
10.1029/2012GL053152
Okada, Y., and E. Yamamoto (1991). Dyke intrusion model for the 1989 seismovolcanic
activity off Ito, central Japan. Journal of Geophysical Research, 96, 10,361–10,376.
Okada, Y. (1992). Internal deformation due to shear and tensile faults in a half-space.
Bull. Seismol. Soc. Am., 82, 2, 1018–1040.
Opheim, J. A., & Gudmundsson, A. (1989). Formation and geometry of fractures, and
related volcanism, of the Krafla fissure swarm, northeast Iceland. Geological Society of
Aamerica Bulletin, 101(12), 1608–1622. Doi: 10.1130/0016-7606(1989)101<1608
Pallister, J.S., McCausland, W.A., Jonsson, S., Lu, Z., Zahran, H.M., El Hadidy, S.,
Abrukbah, A., Stewart, I.C.F., Lundgren, P.R., White, R.A., & Moufti, M.R.H., (2010). Broad
accommodation of rift-related extension recorder by dyke intrusion in Saudi Arabia. Nat.
Geoscience 3(10), 708–712. Doi: 10.1038/ngeo966
Panien, M., Schreurs, G., & Pfiffner, A. (2006). Mechanical behaviour of granular
materials used in analogue modelling: insights from grain characterisation, ring-shear tests
and analogue experiments. Journal of Structural Geology, 28(9), 1710–1724.
Doi:10.1016/j.jsg.2006.05.004
Paquet, F., Dauteuil, O., Hallot, E., & Moreau, F. (2007). Tectonics and magma dynamics
coupling in a dyke swarm of Iceland. Journal of Structural Geology, 29(9), 1477–1493.
Doi:10.1016/j.jsg.2007.06.001
Patton, T. L., & Fletcher, R. C. (1995). Mathematical block-motion model for deformation
of a layer above a buried fault of arbitrary dip and sense of slip. Journal of Structural
Geology, 17(10), 1455–1472. Doi: 10.1016/0191-8141(95)00034-B
Perlt, J., & Heinert, M. (2006). Kinematic model of the South Icelandic tectonic system.
Geophysical
Journal
International,
164(1),
168–175.
Doi:10.1111/j.1365246X.2005.02795.x
Perlt, J., Heinert, M., & Niemeier, W. (2008). The continental margin in Iceland — A
snapshot derived from combined GPS networks. Tectonophysics, 447(1-4), 155–166.
Doi:10.1016/j.tecto.2006.09.020
Pollard, D. D., Delaney, P. T., Duffield, W. A., Endo, E. T., & Okamura, A. T. (1983).
Surface deformation in volcanic rift zones. Tectonophysics, 94, 541–584.
Purdy, G. M., Kong, L. S. L., Christeson, G. L., & Solomon, S. C. (1992). Relationship
between spreading rate and seismic structure of mid-ocean ridges. Letters to Nature, 355,
815–817.
Ramberg, H. (1981). Gravity, deformation, and the earth’s crust: In theory, experiments,
and geological application. (p. 452). London and New York: Academic Press.
Roche, O., Druitt, T. H., & Merle, O. (2000). Experimental study of caldera formation.
Journal of Geophysical Research, 105(B1), 395–416.
Rowland, J. V., Baker, E., Ebinger, C. J., Keir, D., Kidane, T., Biggs, J., Wright, T. J.
(2007). Fault growth at a nascent slow-spreading ridge: 2005 Dabbahu rifting episode, Afar.
Geophysical
Journal
International,
171(3),
1226–1246.
Doi:10.1111/j.1365246X.2007.03584.x
Rubin, A. M. (1992). Dike-induced faulting and graben subsidence in volcanic rift zones.
Journal of Geophysical Research, 97(B2), 1839–1858.
Rubin, A. M., & Pollard, D. D. (1988). Dike-induced faulting in rift zones of Iceland and
Afar. Geology, 16, 413–417.
Ruch, J., Acocella, V., Geshi, N., Nobile, a., & Corbi, F. (2012). Kinematic analysis of
vertical collapse on volcanoes using experimental models time series. Journal of
Geophysical Research, 117(B7), B07301. Doi: 10.1029/2012JB009229
85
Ruegg, J. C., Lépine, J. C., & Tarantola, A. (1979). Geodetic measurements of rifting
associated with a seismo-volcanic crisis in Afar. Geophysical Research Letters, 6(11), 817–
820.
Sigurdsson, O. (1980). Surface deformation of the Krafia fissure swarm in two rifting
events, Journal of Geophysics, 47, 154-159.
Swanson, B. D. A., Duffield, W. A., & Fiske, R. (1976). Displacement of the south flank of
Kilauea Volcano : the result of forceful intrusion of magma into the rift zones. Geological
Survey Professional Paper, (963), 1–39.
Tentler, T. (2005). Propagation of brittle failure triggered by magma in Iceland.
Tectonophysics, 406(1-2), 17–38. Doi:10.1016/j.tecto.2005.05.016
Tentler, T., & Mazzoli, S. (2005). Architecture of normal faults in the rift zone of central
north Iceland. Journal of Structural Geology, 27, 1721–1739. Doi:10.1016/j.jsg.2005.05.018
Thordarson, T., & Self, S. (1993). The Laki (Skaftlir Fires) and Grimsvotn eruptions in
1783-1785. Bulletin of Volcanology, 55, 233–263.
Tryggvason, E. (1980), Subsidence events in the Krafla area, North Iceland, 1975 – 1979,
Journal of Geophysics, 47, 141 – 153
Trippanera, D., Acocella, V., & Ruch, J. (2014). Dike-induced contraction along oceanic
and continental divergent plate boundaries. Geophysical Research Letters, 40, 1–7. Doi:
10.1002/2014GL061570
Walker, G. P. L. (1958). Geology of the Reydarfjordur Area, Eastern Iceland. Quarterly
Journal of the Geological Society, 114(1-4), 367–391. Doi:10.1144/gsjgs.114.1.0367
Walker, G. P. L. (1960). Zeolite zones and dike distribution in relation to the structure of
the basalts of Eastern Iceland. Journal of the Geological Society (London), 68, 515–527.
Walker, G. P. L. (1963). The Breiddalur central volcano, eastern Iceland. Quarterly
Journal of the Geological Society, 119(1-4), 29–63. Doi:10.1144/gsjgs.119.1.0029
Wright, T. J., Ebinger, C., Biggs, J., Ayele, A., Yirgu, G., Keir, D., & Stork, A. (2006).
Magma-maintained rift segmentation at continental rupture in the 2005 Afar dyking episode.
Nature, 1–5. Doi: 10.1038/nature04978
Wright, T. J., Sigmundsson, F., Pagli, C., Belachew, M., Hamling, I. J., Brandsdóttir, B.,
Calais, E. (2012). Geophysical constraints on the dynamics of spreading centres from rifting
episodes on land. Nature Geoscience, 5(4), 242–250. Doi: 10.1038/ngeo1428
86
Figures and Tables
Depth
Experiment
Setup
Type
Tip
(cm)
Thickness
or dilation on
top (cm)
A1
A
rising
0.05
A2
A
rising
0.20
B1
B
opening
rectangle
1
1.00
B2
B
opening
rectangle
2
1.00
B3
B
opening
rectangle
3
1.00
B4
B
opening
rectangle
4
1.00
B5
B
opening
rectangle
4
1.00
B6
B
opening
rectangle
8
1.00
C1
C
opening
triangle
2
0.20
C2
C
opening
triangle
4
0.20
C3
C
opening
triangle
8
0.20
asymmetric
sand cone
Table 1: List of the experiments and relative parameters. In bold the models described in
this work.
87
Fig.1: Scheme of the apparatus used for (a) the rising (setup A) and (b) the thickening
intrusion models (setup B and C). The yellow arrows indicate the rise of the plates in (a) and
the thickening of the intrusions in (b). Schematic section view of setup A (c), B (d) and C (e).
88
Fig. 2: Summary of the results of experiment A2 (upward propagating dike; Table 1). Figs.
a-c: map views of the central part of the experiment (the dashed rectangle in the inset of a)
89
shows the investigated area of the sandbox): (a) before the experiment starts, (b) after 1.5
cm and (c) after 2.5 cm of rise of the metal plates. The plates position is indicated at the
lower and upper edges of the maps. Figs. d-e: (d) Cross sections A-A’ and (e) B-B’; the main
faults are marked by solid lines; the arrows indicate the sense of slip. Figs. f-g-h-i: (f) Vertical
surface displacement map obtained with the laser-scanner at the end of the experiment and
(g) vertical displacement profiles at different time frames of the experiment. (h) Horizontal
surface displacement map at the end of the experiment and (i) horizontal displacement
profiles at different time frames of the experiment. We only consider the horizontal
displacement perpendicular to the dike strike, neglecting any component parallel to it. In
both maps, the vertical dashed black lines show the position of the dike and the horizontal
dashed grey lines the location of the time series profiles. Figs. l-m-n: vertical (l), horizontal
(m) and total (n) displacement field resulting from the PIV analysis of the side view images.
90
Fig. 3: Summary of the results from experiment B1 (shallow thickening rectangular
intrusive complex; Table 1). Figs. (a-d): Map views of the central part of the model for
91
different steps of the experiment. The amount of intruded plates is reported in each image.
(e) Cross section A-A’ in which the vertical throws of the faults are calculated at different
depths, plotted in (f). Each square on the graph shows the amount of vertical offset of the
marker layer along the fault. Squares on the graph correspond to those of the section. (g)
Vertical displacement map and (h) vertical displacement time series profiles; (i) horizontal
displacement map and (l) horizontal displacement time series profiles. In the time series
profiles the color of the profiles indicates the number of intruded plates. Figs. (m-n-o): results
of the PIV analysis from side view.
92
Fig. 4: Summary of the results for experiment B4 (medium depth opening rectangular
intrusions; Table 1). Figs. (a-d): Map views of the central part of the model for different
93
evolutionary stages. (e) Vertical displacement map and (f) vertical displacement time series
profiles; (g) horizontal displacement map and (h) horizontal displacement time series
profiles; (i) cross sections A-A’ and (l) B-B’; (m) variation in the vertical displacement of the
faults in section B-B’; symbols on the graph correspond to those of Fig. 4l.
94
Fig. 5: Results for experiment B6 (deep thickening rectangular intrusion; Table 1). Figs.
(a-d): Map views of the central part of the model for different stages. (e) Vertical
displacement map and (f) vertical displacement time series profiles. (g) Cross section A-A’
and (h) B-B’; (i) vertical throws pattern of the faults in section B-B’.
95
Fig. 6: Results of experiment C1 (shallow triangular intrusive complex with a cone above);
Table 1). Figs. (a-d): Map views of the central part of the model at different stages. (e)
Vertical displacement time series profiles across the graben in the central portion of the
experiment. (f) Cross section A-A’; (g) vertical throws pattern of the faults in section A-A’.
96
Fig. 7: Results of experiment C2 (medium depth triangular intrusive complex; Table 1).
Figs. (a-d): Map views of the central part of the model at different stages. (e) Vertical
displacement map and (f) vertical displacement time series profiles. (g) Cross section A-A’
and (h) B-B’; (i) vertical throws pattern of the faults in section B-B’.
97
Fig. 8: Surface vertical displacement, expressed as a function of the amount of uplift h
with regard to the maximum uplift
Hmax versus the lateral distance from the intrusion axis
X with regard to the top of the intrusion Y, for a numerical model, our A2 and B2 experiments
and four rifting events in Afar and Iceland (see text for details).
Fig. 9: Model of dike propagation in two stages, which repeat indefinitely in a continuum
process: a) the dike focuses the stresses at its tip, inducing a mode I fracture; b) the magma
progressively penetrates the fracture, enlarging it and pushing the two walls aside.
98
Fig. 10: Scheme of the fault patterns at depth (below) and of the vertical displacement
profiles at the surface (above) obtained in (a) setup B and (b) setup C. The scale of the
profiles is different from that of the fault patterns. Labels below each model refer to Table 1.
99
Fig. 11: Summary of the data obtained in the 9 intrusion models of setup B and C. (a)
Mean distance between the conjugate normal faults bordering the graben measured at the
surface (the mean distance is extracted from measurements in five cross section along the
graben); (b) absolute vertical displacement at the surface (positive and negative values
indicate uplift and subsidence, respectively); (c) variation of the depression depth, that is the
sum of the mean subsidence above the intrusion and the mean uplift to the sides.
100
Fig. 12: Evolution of the analogue experiments and of the natural prototypes with only
normal faults (a-c) and with arcuate and normal faults (d-f). The development of the
downward propagating normal faults is explained by figures g and h. (g) Coulomb stress
change on optimally oriented planes induced by an opening fracture simulating the upper
portion of a thickening dike (dark gray rectangle) in section view at few km from the surface;
(h) total Coulomb stress, where the total normal stress is equal to the lithostatic pressure
plus the normal stress change induced by the same fracture of Fig. h. The stress
concentrates at the dike tip, as expected, but also at the surface, where it is large enough
101
to cause faulting. (i) The development of the arcuate reverse faults depends on the
distribution of the maximum compressive stress trajectories induced by an opening dike
[after Patton and Fletcher, 1995].
Fig. 13: Comparison between experiment C1 and the 1783-1784 Lakagigar eruptive
fissure. (a) Panoramic view of the central Lakagigar fissure and its border faults, as seen
from Laki hill southward; (b) cross section of experiment C1; (c) map view of the central
Lakagigar fissure extracted from Google Earth; the general crater alignment (red dashed
lines) and the two border faults (solid lines) are outlined.
102
Fig. 14: Comparison between experiments and mature rift zones. (a) Map of the Krafla
fissure swarm, after Opheim and Gudmundsson (1989). (b) Two dimensional,
103
compressional velocity model derived from travel time data along the Krafla magmatic
system [Brandsdóttir et al., 1997]; velocities are in km/s. Panoramic view and schematic
section of: (c) fault and tension fractures within the Krafla caldera; (d) the Grjotagja fault to
the south of Krafla, showing a tilted hanging wall and basal contraction; location in (a). (e, f)
Contractional structure parallel to the Grjotagja normal fault. (g) View and schematic section
of an open normal fault with horizontal hanging-wall at Reykjanes, southern Iceland. (h)
Cross sections of experiments B1 and (i) B5.
Fig. 15: (a) Plot of the dike depth vs. thickness when faults appear at surface in the
experiments with setup B (black) and C (grey); the relative fitting lines are also shown. (b)
Plot of the dike depth vs. thickness during diking episodes in which new faults formed at the
surface. References: (1, 8) Jonsson et al., 1997; (2) Wright et al., 2006; (3) Sigurdsson,
1980, in Rubin, 1992; (4) Rubin, 1992; (5) Stein et al., 1991; (6) Pallister et al., 2010; (7)
Tryggvason, 1980; (9, 10, 11, 12) Chadwick and Embley, 1998; (13, 14, 15, 17, 18, 19, 20
23) Dvorak and Dzurisin, 1997, and references therein; (16) Acocella and Neri, 2003; (21,
22), Pollard, 1983.
104
Chapter 4
(Published in Journal of Geophysical Research Letters)
105
106
107
108
109
110
111
5. General conclusions
The general conclusions of the PhD thesis can be summarized as follows:
1) Shallower and thicker dikes, inducing a high stress concentration to their tips, usually
form at the surface grabens bounded by normal faults that propagate downward. These
dikes, depending on the interplay between their thickness and depth, may feed eruptive
fissures and cause up to 10 m of slip on the faults, as at Lakagigar, higher than previously
described.
2) Results from analogue models of dike emplacement confirm the deformation pattern
found on the field, as given by a graben structure above the dike. Similarly to what observed
to the sides of eruptive fissures on the field, the normal faults propagate downward from the
surface. In the case of deeper dikes, arcuate faults with a reverse component at the surface
may form; these latter features are also commonly observed in Iceland and Ethiopia, at the
base of the tilted hanging wall of normal faults, suggesting an origin of the structures through
dike emplacement.
3) The deformation pattern along the studied rift segments consists of an overall repetition
of the structural features found along the eruptive fissures forming wider graben-like
structures (as at Thingvellir in Iceland) or juxtaposed and nested grabens (as at Fantale in
Ethiopa, and Krafla in Iceland) hosting several fissures or scattered vents. Moreover,
different specific features found along the studied rift segments are magma-induced: these
include the presence of diffuse contraction at the base of the tilted hanging wall, that is
induced by upward propagating arcuate faults (Krafla, Vogar, Fantale), the downward
propagation of the fault in magmatic systems undergoing higher extension rate (> 10mm/yr;
as central Krafla), the upward propagation of reactivated blind faults along magmatic
systems with lower extension rate (<10 mm/yr; as Vogar, Thingvellir, Fantale). All these
features may be simply explained through dike emplacement, as also observed in the
experiments, and therefore suggest that the structure of the rift segments is the product or
repeated diking episodes
4) These results suggests that shallow dike emplacement may explain all the observed
deformation along eruptive fissures as well as along rift segments, so that the overall shape,
112
structure and development of divergent plate boundaries may be in principle entirely
magma-induced. Regional tectonics, in the form of ridge push, certainly plays a fundamental
preparatory role in shaping the plate boundaries, influencing the direction of the dikes and
separating the plates apart on the longer-term and over longer distances. Regional tectonics
may also directly activate normal faults, seismically or not, during inter-diking episodes..
However, the collected data suggest that in a magmatically active rift, where plate separation
is constantly maintained by diking, the regional tectonic stress may rarely reach values high
enough to be released and activate the normal faults. Therefore, any direct role of regional
tectonics in separating and shaping the plate boundaries appears negligible compared to
that of diking.
113