Hypothesis: the earthquake swarms report relaxation of accumulated, compressive tectonic stresses
The crust under the Canaries, part of the African tectonic plate, is moving north eastwards at 24mm annually in response to drag from the upper mantle.
The crust/upper mantle boundary under Hierro plunges to at least 30km deep below the island . Gorbatikov et.al.: www.gradient-geo.com/library/storing.php?doing=T91oVy5lrThe MOHO on the ‘bulk plate’ is at ca. 10km depth between the islands. Watts: http://www.earth.ox.ac.uk/~tony/watts/Oceanislandsandseamounts.htmSo the island presents a 30-40km tall obstruction that distorts mantle flow and so reduces the propulsive force on the the island compared with the bulk plate. This results in a compressive force, directed NE-wards, being exerted on the SW arc of the island by the bulk plate. So what might be the response of Hierro’s crust to these forces?
Compressive crustal strain
Compressive stress will accumulate in the country rock throughout an arc around the south west of the island. The initial eruptions that formed Hierro occurred in the north east of El Hierro (Tinor, SSW of Valverde). The site is revealed in Gorbatikov’s microseismc survey as a massive, 5km diameter, 30km deep towering structure. So a NE-wards directed loading will lead to accumulation of compressive stress between this Valverde stack and the bulk plate to the SW of the island. This compressive stress accumulation must be almost silent to seismometers: very few earthquakes have been recorded at Hierro over 25 years prior to 2011. The relentless force from plate motion has the desired property to sustain compressive loading. Note that for simplicity we ignore mantle flow forces directed downwards and around the island. We can now offer explanations for the patterns of the earthquake swarms.
Swarm location where stress relaxation is maximal
It is the sudden relaxation of accumulated compressive stress that causes the earthquakes. The earthquake swarms are proposed to result from relaxation of compressive stress. So the stress gradient is important. We can see this gradient in simplified form by looking at the flow of an incompressible fluid around a cylinder representing Hierro:
While compression will be maximal ( red) on the central axis ( SW-NE) it is only when the compression can relax into less compressed rock that we detect earthquakes in elongated swarms.
The stress gradient ( red going to blue) will be maximal around the western arc (and southern arc) of the island. The two 2011 swarms follow the stress gradient. The 2012 swarm is on the central axis and did not appear to follow stress gradient (see below). Note that the precise pressure distribution around Hierro will be much more complex than depicted and will be maximal at the 10km depth of the MOHO, so the outline of the the island is merely a very rough guide to locations.
Patterns in swarms
Several patterns support a stress relaxation hypothesis.
- The first two swarms showed distinct north-south linearity, which is at 45 degrees to the SW-NE central axis and plate motion. The orientation is what would be expected if earthquakes occur where the stress gradient is greatest ( red-blue on diagrams above). The swarms’ southern limits are where the gradient reduces, despite the absolute level of compression being greatest on the central axis (dark red)
- The 2012 swarm was orthogonal to the first two swarms and may not have followed the circumferential pressure gradient (see 6 below).
- The distinct conical shape in vertical section to the first swarm suggest a stress-relaxation failure process,with a greater gradient at shallower depths permitting relaxation upwards. The conical pattern is clearly seen in Geolurking’s plot of earthquakes onto the microseismic image.
- The maximum compression from the bulk plate impinging the island would occur at MOHO depthof ca 10km, coinciding with the maximum width of the first swarm. This region, of all, would be expected to be exposed to the greatest compression and this, combined with it the shallowest swarm, may explain why it was the first zone to fail. The trend to events spreading southwards during the first and second swarm likewise could reflect an ‘unzipping’ of stress, with prior relaxation influencing subsequent events.
- The gently curving upper surface of the first swarm, deepening to the south, putatively follows the lower bound of the sedimentary layer. The sedimentary layer in this model does not accumulate stress, but rather accommodates compressive strain by sliding in shear. So few earthquakes arise in the sedimentary layer, which also insulates the overlying erupted edifice from strain and hence explains its lack of stress accumulation and earthquakes.
- The curved lower bound to the base of the second swarm may reflect the transition from colder crustal rock that has accumulated compressive stress, compared with underlying more plastic upper mantle that deforms and accommodates loading without accumulating stress. The second swarm was deeper, ca 20-25km, and again elongated N-S. It was separated from the base of the first swarm by an unusual aseismic layer at ca 15-20km depth. It can be seen as the clear band here http://www.02.ign.es/ign/resources/volcanologia/jpg/Eventos_HIERRO_2012.jpg
The south end of this aseismic layer became active in the third swarm suggesting that stress had accumulated there but was unable to be relaxed during the second swarm. Interestingly the azimuths in this region differed in the two swarms- see 7 later)
- The third swarm (2012) produced rapid GPS uplift and strong NE-wards horizontal motion. Daniel’s plot gives a clue to the geometry of stress relaxation horizontally: source: http://earthquake-report.com/2011/09/25/el-hierro-canary-islands-spain-volcanic-risk-alertincreased-
The geometry of the GPS motions suggests an on-axis origin for the driving force, corresponding to the high pressure zone (Fig2, dark red) and directed NE-wards across the SW of the structure The GPS motion suggests the entire island edifice relaxed to the NE, as shown by Valverde H100 – the massive stack – shifting in precisely the direction of African plate motion. In 2011 H100 was immobile. Other motions suggest a a radial relaxation towards low pressure zones from the geometric centre. The alarmingly rapid GPS uplifts ( not shown) may result from a component of the stress relaxation being directed upward.
- Data on earthquake azimuth in the two 2011 swarms are compatible with azimuth reflecting the direction of the relaxations.
(source: Chryphia). As seen at 1-34min in the video the azimuths in the first and second swarms were either northwards or southwards ( red or dark blue) with few intermediate angles (pale bluegreen). This alignment, predominantly north or south, is in agreement with the compression gradient. The third swarm, 2012 started with eastward azimuths (Chryphia’s plot: yellow, green) followed by southward (blue) as activity spread westwards The temporal relationship between the eastward azimuths in the 2012 swarm and the eastward GPS motion deserves investigation.
- Conventionally the swarms are believed to be generated by magma intrusion. Interestingly IGN’s deployable seismometer array recorded no magmatic signatures such as harmonic tremor or LP events during the 10,000 earthquakes of the first swarm prior to the submarine eruption. http://meetingorganizer.copernicus.org/EGU2012/EGU2012-11986.pdf
- Tenerife has shown a low intensity linear eq swarm with N-S orientation to the west of Teide, the island’s central volcano, as seen in Schteve’s plot:
The mechanism may resemble Hierro’s except that the gradient of compression is less due to the larger size of the island, hence the low intensity of earthquaking. The swarm under Emedio shows no linearity, which would be predicted as its erupted mass is considerably less than Hierro’s and compression does not accumulate, and magmatic processes dominate.
Hypothesis and implications
The earthquake swarms are envisaged as reflecting the sudden relaxation of tectonic compressive stress that had accumulated over several hundred years. The 2011-2012 swarms may be a repeat of the swarm in 1793. However, the model predicts that swarms will also occur as a mirror image, extending at 45 degrees south of the central axis. If the 1793 activity was focussed to the south then the period of stress accumulation could be longer for the recent swarms. If the repose period is say 200 years, a NE-wards GPS motion of 15cm would represent about six years’ lost motion. So Hierro would lag the bulk plate motion by just 3%.
If the hypothesis is valid, it means that considerably less magma (if any) has arrived under El Hierro than indicated by the conventional interpretation of seismic and GPS data. Relaxation of compressed rock will involve some increase in volume as reflected in the GPS motion many km above. We have no knowledge on the motions at depth, but the relaxation of a column of compressed rock will create space in neighbouring rock that allows decompression degassing and melting of emplaced magma.
Without detailed analogue or in silica modelling of the compressive stresses, or knowledge of deep structure on which to base those models, we cannot predict with any confidence if the swarms are complete. Swarms in a mirror image configuration south of the axis are a possibility. Fluid flows past obstructions often oscillate side-to-side,so perhaps in another 200 years the Restinga swarms will recur?
The best initial test of the hypothesis might lie in the 2004 microseismic study by Gorbatikov at al. The location of the microseismic sources, reflecting compresional stressing, would be predicted to be greatest in an arc of compression extending across the south western arc of the island.
The 2011- 2012 swarms were north of the central, SW-NE compressive axis. The model predicts similar swarms to its south, running W-E ( El Julan-Las Playas) under La Restinga. So the hypothesis presented here can potentially explain the ‘Mercedes star’ shape, together with its orientation on the plate. Those zones of the island that have been repeatedly fractured by cycles of compression and relaxation-fracturing will provide the network of small conduits and sills needed to explain rapid magma ascent. So the Mercedes star and its orientation on the plate are proposed to be the result of tectonic forces. The sedimentary layer, possibly unique to the Canary Islands, may play a pivotal role in allowing tectonic earthquake swarming by removing restrictions from the erupted edifice on deeper compression-relaxation motions. Alternatively, the mechanism proposed here may be rare by being dependent upon the size of the initial eruption being small enough to create a compression gradient that permits stress/relaxation cycling. The gradient around larger initial eruptive edifices (eg putatively Gomera) may not permit relaxation, so intense swarming and fracturing and the star shape do not occur.
This hypothesis should in no way used to negate the safety measures promoted by the authorities. It is speculation by an amateur.
PETER COBBOLD, Professor in cell biology, not the professor of geology of similar name.