Abstract
São Miguel, the largest island of the Azores, corresponds to one of the most seismically active areas within the archipelago. During historical times it was affected at least six times by destructive earthquakes, which were responsible for thousands of deaths. Several seismic crises are related both to volcanic eruptions and unrest episodes, being the cause of significant damage over more restricted areas. The central region of the island has been the most active in recent years and is characterized by the occurrence of seismic swarms, generally with low-magnitude events (MD≤2), that are clustered both spatially and temporally. A significant relevant seismic swarm occurred between May and September 2005 and was associated with an episode of volcanic unrest. Another part of the seismicity recorded on São Miguel is associated with submarine volcano-tectonic structures that are located near the island, some of which are the source of the strongest earthquakes. The spatial distribution of b-values shows that high values (i.e. b>1) are mainly associated with the Fogo–Congro region, whereas low b-values (i.e. <1) are more common offshore. The spatial distribution of b-values in the Fogo–Congro region is suggestive of the existence of material and/or stress heterogeneities, and this is supported by a mixture of fault plane solutions.
The Azores archipelago, because of its plate boundary geodynamic setting, is active seismically and volcanically (Gaspar et al. 2015; Madeira et al. 2015). São Miguel, located in the southeastern part of the archipelago, is the scene of significant seismic activity that is characterized by the occurrence of frequent seismic swarms. These are mainly concentrated in its central region, which encompasses Fogo Volcano and Congro Volcanic Fissural System. This type of seismicity normally occurs in volcanic and/or geothermal regions and presents high b-values in the magnitude–frequency distribution, because the swarms are dominated by events of low magnitude (e.g. Fischer 2005). The occurrence of earthquake swarms in active volcanic regions has been attributed to subsurface magma movement at shallow depths (e.g. Spicák & Horálek 2001; Fischer & Horálek 2005; Hensch et al. 2008) and/or are associated with hydrothermal fluid circulation through new or pre-existing fractures (e.g. Husen et al. 2004; Farrel et al. 2009). Therefore, the swarms are commonly limited to regions with exceptionally heterogeneous distributions of material properties and stress concentrations (Hill 1977). This is the case in the central region of São Miguel. In the last few decades several seismic crises have been recorded in this region. The first accurately located crisis occurred in 1989 (e.g. Gaspar & Queiroz 1989; Tryggvason et al. 1989) and the 2005 swarm was the longest that has been instrumentally recorded. The latest crisis occurred in 2011–12. Seismicity is not restricted to either onland locations or to low-magnitude events, and several strong earthquakes have struck São Miguel, most of which are associated with the submarine volcano-tectonic structures that surround the island.
Destructive earthquakes
Almost all knowledge of past seismicity relies on historical records of earthquakes and their impacts. Since the Azores archipelago was settled, in the fifteenth century, about 30 destructive earthquakes and countless seismic crises have resulted in thousands of deaths and severe damage (Gaspar et al. 2015). Among the major tectonic historical earthquakes in the Azores, six were located on São Miguel or in its submarine surroundings (Fig. 17.1). These events were relocated and reassessed by Silveira et al. (2003) using EMS-98 (Grunthal 1998) and are briefly described below. Other destructive seismic activity is related to the historical volcanic eruptions in 1563, 1652, 1630 and 1811.
Map of São Miguel Island. Circles mark the epicentres of the most destructive earthquakes that have occurred in this region (adapted from Silveira et al. 2003). Volcanic systems: A, Sete Cidades Volcano; B, Picos Volcanic Fissural System; C, Fogo Volcano; D, Congro Volcanic Fissural System; E, Furnas Volcano; F, Povoação Volcano; G, Nordeste Volcanic System. Localities: 1, Água Retorta; 2, Lomba do Loução; 3, Lomba do Cavaleiro; 4, Ribeira Quente; 5, Ponta Graça.
22 October 1522
This was the first and strongest earthquake to strike São Miguel in historic times and caused the destruction of Vila Franca do Campo, which was at the time the capital of the Azores. The epicentral location was in the central part of the island, about 2 km to the north of the village, and the maximum intensity was estimated at X (EMS-98). Historical accounts by Frutuoso ([1522–91] 1981) describe how the main shock triggered numerous landslides, mainly in the central and eastern parts of the island. Two major landslides completely buried Vila Franca do Campo, increasing the destruction and producing a death toll of up to 5000 people. The source areas of these landslides were located to the NW and NE of the village, close to the epicentre, and were channelled along two pre-existing stream channels (Marques et al. 2009).
26 July 1591
The recently rebuilt village of Vila Franca do Campo was again destroyed by a strong earthquake that struck approximately the same area as the 1522 event. The epicentre was located off-shore, near the south coast, and the earthquake had a maximum intensity of VIII–IX (EMS-98). This earthquake destroyed several buildings and caused many fatalities. Several landslides were triggered and a new spring flowed for 4 days. This earthquake marked the beginning of an episode of severe seismic activity that lasted until 12 August.
16 April 1852
This earthquake had its epicentre to the north of São Miguel. It reached a maximum intensity of VIII (EMS-98), mainly on the north flank of Sete Cidades Volcano, and triggered several landslides along the coast. The city of Ponta Delgada was damaged. The 1852 earthquake was the disaster that produced the most severe damage on the island during the nineteenth century, causing nine deaths and leaving several people seriously injured.
5 August 1932
The eastern part of São Miguel was severely affected by this earthquake. The epicentre was located offshore, SE of Água Retorta, and the maximum intensity was VIII (EMS-98). Santa Maria Island was also affected with an intensity of III (EMS-98). In the Concelho da Povoação (e.g. in Água Retorta and Lomba do Cavaleiro), buildings underwent total or partial collapse, causing no casualties but more than 3000 people were made homeless. Landslides were triggered mainly along the south coast from Ponta Garça to Lomba do Loução.
27 April 1935
São Miguel was strongly affected by a severe earthquake. The epicentre was located a few kilometres to the south of Ribeira Quente and the maximum intensity was IX (EMS-98). Destruction extended across almost the entire central and eastern zones of the island, but only caused one death. Several landslides were triggered along the south coast, from Ribeira Quente to Lomba do Loução.
26 June 1952
This earthquake was located south of Ribeira Quente, in almost the same epicentral area as that which gave rise to the 1935 event. The maximum intensity reached VIII (EMS-98) at Ribeira Quente, with Povoação and Vila Franca do Campo villages also being affected. In some places, the total or partial collapse of buildings occurred and more than 600 people were made homeless. Several landslides were triggered along the southern coast near to Ponta Garça, Ribeira Quente and Lomba do Cavaleiro.
The majority of these destructive earthquakes had their epicentres off-shore, mainly to the south and SE of the island, and were usually associated with submarine volcano-tectonic structures. The 1522 earthquake appears to have been the only event that had an epicentre inland, in the central region of São Miguel. The 1852 earthquake is unique in that it had an epicentre located to the north of the island.
Temporal and spatial distribution of seismicity
In the Azores archipelago, the first two seismic stations were installed in 1902. After the 1st January 1980 Terceira earthquake, a sufficient number of seismic stations were available and they could begin to act as a seismic surveillance network. Today, the CIVISA seismic network comprises 37 short-period stations deployed at seven of the nine islands of the Azores archipelago (Gaspar et al. 2015). On São Miguel, the seismic network is composed of 16 stations deployed around the principal volcanoes (Fig. 17.2): Sete Cidades, Fogo and Furnas.
CIVISA seismic network of São Miguel Island.
Instrumental data clearly demonstrate that most of the seismic activity in the Azores takes place along plate boundaries, both the Mid-Atlantic Ridge and the Terceira Rift (TR) (Gaspar et al. 2015). Some of the seismicity that occurs on São Miguel is related to TR activity.
Until 1997, seismic data were sparse and insufficient for detailed study. In this chapter the seismicity recorded from 1997 to 2010 is characterized. During this period several seismic crises have occurred on São Miguel, where hundreds to thousands of events (Fig. 17.3a) were recorded over time-scales ranging from hours to a few months. The majority of the earthquakes had low magnitude, with MD values ≤2, and were located in a depth range of between 0 and 10 km (Fig. 17.3b, c).
(a) Monthly total earthquakes with determined MD. The solid line represents the cumulative number of locatable events as a function of time. (b) Magnitude (MD) distribution as function of time. (c) Depth distribution as a function of time.
Over 17 000 locatable earthquakes have been recorded with epicentres on the island or in nearby submarine areas. More than 13 000 are associated with the main volcanogenic zones of São Miguel: Sete Cidades Volcano (Queiroz et al. 2015); Picos and Congro volcanic fissure systems (Ferreira et al. 2015); Fogo Volcano (Wallenstein et al. 2015); and Furnas Volcano (Guest et al. 1999). However, far more seismic events were recorded but, because of their low magnitude and/or reduced number of phases, they could not be located. In fact, in 2005 more than 40 000 seismic events were recorded between May and September, but only around one-sixth of them could be located. The number of seismic events associated with each volcanogenic zone is presented in Figure 17.4, where it may be seen that the seismicity associated with the Fogo–Congro region, which comprises Fogo Volcano and the Congro Volcanic Fissural System, is much higher than any other area. In contrast, the Picos Volcanic Fissural System represents the volcanogenic zone with lowest seismic activity.
Temporal distribution of seismicity for each volcanogenic zone. VZ01, Furnas Volcano; VZ02, Fogo–Congro region; VZ03, Picos Volcanic Fissural System; VZ04, Sete Cidades Volcano.
The evolution of the spatial distribution of seismicity on São Miguel for the period 1997–2010 is presented in Figure 17.5. From 1997 to 2001, seismicity was sparse and with a low number of events each year. A more detailed examination highlights, however, a small increase in seismicity at Sete Cidades Volcano and surrounding areas in 1998, which became more intense in 1999 (Fig. 17.5b, c). In 2002, the seismicity at Sete Cidades Volcano decreased, but a slight increase started to occur in the Fogo–Congro area (Fig. 17.5f). This activity continued to increase during 2003 and 2004 and reaching a peak in 2005 (Fig. 17.5g–i), with an intense swarm that lasted from May to September. From 2006 to 2008 (Fig. 17.5j–l), the seismicity in the central region became sparse, and it was possible to distinguish two areas: a broad zone near to Fogo Volcano and another associated with Furnas Volcano, located on the eastern rim of the caldera. In 2009, Sete Cidades Volcano showed only a short-lived seismic swarm of a few hours’ duration that was highly clustered, not only temporally but also spatially (Fig. 17.5m). During 2010, seismicity was again dispersed, but with a slight concentration of epicentres close to Furnas Volcano (Fig. 17.5n).
Seismicity evolution from 1997 to 2010 (data from CIVISA). Red dots correspond to the yearly epicentres. The yellow dots represent the overall seismicity for the considered time period.
The general distribution of seismicity across São Miguel (yellow dots, Fig. 17.5a–n) presents a general NW–SE trend, extending from the submarine structure, known as the Hirondelle Basin, to the Seamount Grande Norte and Povoação Basin. This pattern suggests tectonic control exerted mainly by the TR. The seismicity recorded in the central region of São Miguel presents some particularities, with the existence of ‘volcanic-type’ events that cannot be explained by regional tectonics and is suggestive, rather, of the presence of a local stress field that is controlled by the volcanic/hydrothermal system associated with the Fogo–Congro region.
Seismicity on São Miguel is characterized mainly by low MD values. Nevertheless, since 1997 more than 200 earthquakes have been felt (with MD>2) with intensities ranging up to VI (MMI), 185 of these being located on land and the remaining in the surrounding submarine areas. The majority of the felt seismicity was located in the Fogo–Congro region (176 felt earthquakes; Fig. 17.6), mostly during the 2005 seismic swarm (Fig. 17.7), with 126 felt events occurring between May to September.
Yearly number of felt earthquakes for each volcanogenic zone (data from CIVISA). VZ01, Furnas Volcano; VZ02, Fogo–Congro region; VZ03, Picos Volcanic Fissural System; VZ04, Sete Cidades Volcano.
Temporal evolution of the felt seismicity in São Miguel Island (data from CIVISA).
Completeness
The magnitude of completeness (Mc) may be defined as the magnitude at which cumulative frequency–magnitude distributions diverge from a linear trend, if the dataset strictly follows the Gutenberg–Ritcher law (Wiemer & Wyss 2002). The Mc and the slope of the frequency–magnitude distribution (b-value) are directly related. If the Mc is underestimated, then the b-values will be too low; conversely, if Mc is overestimated, then the amount of data available will be smaller and uncertainties associated with the estimate will be higher (Wiemer & Wyss 2002). In a given region and for a given period of time, the relationship between the size of an earthquake and the frequency of occurrence can be represented by the Gutenberg–Ritcher law (Gutenberg & Ritcher 1944). However, it is commonly found that the frequency–magnitude distribution deviates from the linear power law for smaller magnitude events. This deviation may be due to some of the smaller events being missed by the network and may be attributed to several causes, including: (1) seismic events being too small to be recorded at a sufficient number of stations; (2) operators of the network deciding that events below a certain threshold were of no interest; and (3) events being too small to be detected within the coda of larger events, for instance in aftershock sequences.
Since the configuration of São Miguel's seismic network has changed over the time period 1997–2010, it is important to determine the detection threshold. In the CIVISA catalogue there are two types of magnitudes available: duration magnitudes (MD) and local magnitudes (ML). However, the most representative for this study period is MD, and this is the reason why it is used in the present chapter. MD values were calculated using the empirical relationship of Lee et al. (1972). The histogram of the number of events v. magnitudes (Fig. 17.8a) can be seen as a first indication of the Mc value expected for the whole catalogue. Figure 17.8a shows that magnitudes near zero are sometimes reported, but that the highest number of events reported is near to magnitude 1, giving the first indication of the expected value of Mc, which can be equal to or higher than 1. There are different techniques that can be used to estimate Mc (e.g. Wiemer & Wyss 2000; Woessner & Wiemer 2005). A simple technique often employed is based on estimating Mc from the frequency–magnitude distribution; however, in this study the Mc was computed using the ZMAP program (Wiemer 2001), which automatically estimated Mc=1.4±0.09 for the study area catalogue (Fig. 17.8b).
Average magnitude of completeness. (a) Histogram of the number of events plotted against magnitudes for all the earthquakes in the study area. The magnitude bins are equal to 0.1. (b) The frequency–magnitude distribution for earthquakes located in the study area. The line represents the maximum likelihood (Aki 1965) fit to the data.
Spatial changes in b-values
The b-value is one of the most studied parameters of seismicity and it is strongly related to the completeness of the seismic catalogue. Recent studies have demonstrated that b-values vary spatially at local to regional scales, providing information about the tectonics, material properties, pore pressure conditions, material heterogeneities and/or differences in stress of a given region (e.g. Wyss et al. 2001; Sánchez et al. 2005; Bridges & Gao 2006; Zhao & Wu 2008). Areas with high b-values are usually associated with volcanic regions, linked to magmatic and/or hydrothermal activity (e.g. Wyss et al. 2001; Wiemer & Wyss 2002; Sánchez et al. 2004, 2005; Bridges & Gao 2006; Farrel et al. 2009; Lin et al. 2007).
A detailed analysis of the spatial distribution of b-values in two dimensions was made for São Miguel island and its surrounding submarine areas, in an attempt to find regions with anomalously high b-values, which according to Bridges & Gao (2006) may be associated with magma chambers, not directly, but linked to fractured areas that are affected by the presence of the magma reservoirs.
In order to map the b-values, the ZMAP program (Wiemer 2001) was employed, which requires the area of study to be divided into a grid. At each node the Mc and the b-value were calculated from the Ni closest events to the node, with Ni being a fixed number. Instead of considering a unique Mc value, the Mc is automatically calculated according to the frequency–magnitude distribution in each grid sample, which allows a more correct estimate of b-values to be made. It also requires setting a minimum number of earthquakes (Nmin) with MD≥Mc, in order to estimate the b-values for each node, based on the equation:
where N is the cumulative number of earthquakes with magnitude greater than M, and a and b are constants that represent, respectively, the productivity of a given volume and the relative size distribution of seismic events. The b-value corresponds to the slope of the best-fitting line between the number of earthquakes and their magnitudes.
Map resolution depends on the grid nodes and earthquake density (Nuannin 2006) and, therefore, a map with a high density of earthquakes and small grid spacing will present a high resolution. Thus, in order to spatially map the b-values, the study area was divided into a grid of 0.01×0.01° (∼1 km), and Ni established to 50 earthquakes, since Wiemer & Wyss (2002) suggested that this number is the minimum needed in order to obtain a statistically significant result. The Nmin was set to 5 to avoid too many gaps, corresponding to regions with low seismicity. For samples that contained fewer than 50 events, the b-values were not calculated.
Significant spatial variations in frequency–magnitude distribution are well defined in the central region of São Miguel, and are probably related to variations in tectonic and/or volcanic processes. The map (Fig. 17.9) shows that regions of b-values ≤1 are more common offshore, but, in contrast, regions with b-values >1 are located inland (yellow to red colours) and are mainly associated with the Fogo–Congro region. At Furnas and Sete Cidades volcanoes, it is also possible to visualize some areas with b-values >1.
Spatial b-value distribution for São Miguel and surrounding submarine areas.
Fault plane solutions
Fault plane solutions can be misleading, particularly if the seismic network configuration does not allow a good coverage of the area of interest. For this reason and to obtain well-constrained fault plane solutions for São Miguel, only data from the Fogo–Congro region (2003–08), which has better network coverage, were used by Silva (2011). Eight representative multiplets (M1–M8) (Fig. 17.10) were identified based on waveform cross-correlation methods. The fault plane solution for each of their master events was determined.
Spatial distribution of the seismic families and their representative focal mechanisms. The solid and dashed lines represent observable faults and inferred lineaments, respectively (from Silva 2011).
The focal mechanisms for M7 and M8 multiplets present oblique-normal faulting and strike-slip faulting styles, respectively, suggesting transtensional to extensional regimes in the area, which are in good agreement with the known regional stress field, with its dominant NE–SW trend for the T-axis (Carmo et al. 2015).
The multiplets located at the northern border of Fogo Lake do not show any preferential lineament. In this area compressive and tensional styles coexist, reflecting a complex system. This mixture of fault plane solutions seems to be the result of heterogeneous stress sources that are common in volcanic and/or hydrothermal areas, where structural features generally result from the local superimposition of gravity, tectonic and/or magmatic/hydrothermal-induced stress, promoting the rotation of the major axis of the stress tensors. The apparent oblique-reverse mechanisms of M4 and M5 seem to be rotated relative to those shown by the oblique-normal solutions of the nearby seismic families, M1 and M2, respectively. This apparent rotation may represent local changes in the stress field, or slip on different oriented faults under a uniform stress field.
Discussion
During historical times São Miguel was affected by destructive earthquakes, which have been responsible for thousands of deaths, much homelessness and extensive damage. The epicentral locations of the strongest historically known earthquakes are an essential contribution to the process of evaluating the seismic hazard of a given area. On São Miguel, the greatest numbers of strong earthquakes have occurred offshore to the south and SE and are mostly associated with the Povoação Basin submarine volcano-tectonic structures. The exception is the 1852 earthquake, which was located to the north. Another zone of high seismic activity is the central region of the island, where the strongest historically known earthquake occurred in 1522.
Recently, improvements in instrumental seismology have allowed the detection of smaller magnitude events, which results in a more complete knowledge of the seismic activity of the region. The seismicity recorded on São Miguel, and in its submarine surroundings structures in the period of 1997–2010, seems to be associated both with deformation that takes place along the TR regional tectonic structures and with episodes of magma intrusion. In fact, the activity related to the TR deformation does not explain all of the recorded seismicity and a local stress regime must be present, such as in the central region of São Miguel, in order to explain the occurrence of some volcanic-type signals.
The seismicity of the island shows several different patterns, reflecting high stress complexity, where more than one stress field is present. The seismicity located in the Fogo–Congro region is clearly influenced by the volcanic/hydrothermal system present in this area (Silva et al. 2012). In contrast, the seismicity of Sete Cidades Volcano, and its surrounding areas, seems to be influenced by regional tectonics, namely TR activity, which shows a NW–SE trend.
Higher levels of seismic activity in the Fogo–Congro region from 2003 to 2008 were accompanied by ground deformation, and this indicates that an unrest episode had occurred that reached a higher intensity from May to September 2005. The velocity field obtained from GPS data spanning the period January 2000 to July 2007 (Trota et al. 2009) defines a clear outward radial pattern whereas the vertical component shows stepwise inflation/deflation episodes. The deformation source was attributed to a shallow magma intrusion emplaced at a depth of 2.1 km. This unrest episode was also characterized by the occurrence of low frequency seismic signals, which previously had not been frequent.
The existence of high b-values in the Fogo–Congro area (the central part of the island) and the fault plane solutions determined for this region support the hypothesis of a highly heterogeneously fractured medium. Zones of high b-value are interpreted as being influenced by high crustal heterogeneities in a local stress regime, which causes numerous small cracks/fissures with a diversity of orientations, but fewer large cracks/fissures (Wyss et al. 2001; Farrel et al. 2009).
Some authors have suggested a fluid-driven origin, which induces a pore pressure increment affecting an extremely heterogeneous medium, as the cause of the recorded swarm activity between 2002 and 2004 (e.g. Bonagura et al. 2004; Saccorotti et al. 2004; Silva et al. 2008). In fact, Silva et al. (2012) demonstrated – using stress inversion based on fault plane solutions – that in 2005 the central region of São Miguel presented a localized heterogeneous stress field that was responsible for numerous small cracks that were oriented in all directions. This stress field was superimposed on to the regional stress field and is compatible with the 3D deformation pattern associated with a magma intrusion episode (Trota et al. 2009).
In marked contrast, the seismicity recorded offshore has average magnitude values that are higher than the ones found for São Miguel. This fact is corroborated by the existence of low b-values in the surrounding submarine areas (mean value <1). These b-values can be attributed to high stress accumulation outside the dominant influence of any magmatic and/or hydrothermal systems associated with the central volcanoes located inland. The seismicity offshore should be analysed carefully, because the configuration of the seismic network does not allow the detection of small-magnitude events and may lead to erroneous low b-values.
Conclusions
The seismic activity recorded on São Miguel and its surrounding areas presents a general NW–SE trend, reflecting the influence of the TR. Nevertheless with respect to the central volcanoes, the seismicity is highly affected by a local stress regime that sometimes overlaps the regional stress field. In the case of the central region of São Miguel, the local stress regime is responsible for the microseismicity recorded and is associated with the presence of magma reservoirs and/or magma intrusions, which favours the formation of numerous microfractures with different directions.
- © 2015 The Geological Society of London
References
Chapter 17 Seismic activity on São Miguel Island volcano-tectonic structures (Azores archipelago)
