At one time, Portugal ruled the world. Its explorers discovered the Cape of Good Hope, created colonies around the Indian Ocean, reached China and Japan, founded the city of Nagasaki, and claimed much of South America. The glory days of the Portuguese empire lasted from 1415 to 1750, and left a rich heritage in Portugal. But geology was the nation’s Achilles heel.
Portugal’s capital, Lisbon, midway along the Portuguese coast, is not close to a plate boundary and is not a place to expect a major earthquake. History decided differently. The Lisbon earthquake of 1755 is among the worst disasters to hit Europe in modern times. It devastated the Portuguese empire. Still, many articles, newspapers and even some scientific literature overstate the magnitude, number of casualties, and impact on society; scientific papers comment on the fact that some statements appear exaggerated. Let’s look at the facts.
It came from the Atlantic Ocean, during the celebrations of All Saints day. In Lisbon, Sunday, 1 Nov 1755 was a sunny day, warm for the time of the year. Around 9:35am, gentle shaking started, as from heavy carts, and a rumbling sound came from the ground, lasting about 1.5 minute. A minute later, there was a sudden violent shock: strong shaking lasted 2.5 minute, causing panic. Another minute later, there was another shock, perhaps stronger than the previous one; this time the shaking lasted 3.5 minutes, slowly becoming less violent. These nine minutes left Lisbon in ruins, its population destitute from the immense destruction. All major churches in Lisbon collapsed, killing many of the congregations. Over 80% of the houses were damaged. Deep fractures were left in the city. The first aftershock came at 11am. The aftershocks continued for months and some were themselves violent.
Lisbon was not the only place that was badly affected. There was major damage in the Algarve; Faro was almost completely destroyed. Damage in southern Spain and Morocco was extensive. North of Lisbon the damage was considerably less. The shaking was felt across Spain, France and Italy, to about 2500 km away.
There is a detailed and precise report on the earthquake from Cadiz, in southern Spain, 350km southeast of Lisbon. The same three phases were observed with a similar duration (perhaps the last phase lasted a bit longer), but the third phase was notably less violent than the second, whereas in Lisbon they were more comparable.
The population of Lisbon fled to safety, away from buildings. Many went for the harbour. Around 10am, some 30 minutes after the quake, a mountainous wave came up the Tagus river. Within 5 minutes the water rose to a height of 5 to 6 metre; the wave penetrated 250 metres inland impacting especially the western areas of Lisbon. Many boats overturned. A large pier with many people sheltering on it sank and completely disappeared, probably due to liquefaction.
The first wave was followed by 2 or 3 more, over 1-2 hours. The first major aftershock occurred in between these waves. In the midst of the destruction, there was one lucky break: the tsunami hit an hour before low tide, so that the base water level was low and the tsunami was slowed down by the outgoing tide and flow of the river.
On the coast, the harbour of Setubal was swamped by the tsunami with water reaching the first floor of buildings, 6 metre high. Further south, the harbour of Lago was ruined and had to be moved afterwards due to the sediment. Here the waves came over the city walls, 11 meter high: the water traveled 2.5 km in-land. Blocks up to 4500 kg were deposited on the coast. Portimao was badly hit. Faro escaped lightly, protected by a sandbank in front of the city (but this was scant consolation as the city had been destroyed by the earthquake instead). Elsewhere in the Algarve, the highest waves reached 20 metre, perhaps higher. In Agadir, Morocco, the water also came over the city walls. There was tsunami damage on the island of Terceira on the Azores. 3.5 meter high waves reached Antigua 10 hours after the earthquake. There was damage in Recife, Brazil. In Cornwall, UK, the waves were 3 meter high.
Some reported tsunami heights seem overstated. For instance, for Cadiz, in Spain, a wave height of 17-19 metre is often stated. But the town survived unscathed: as it is 11 metre above sea level, the tsunami was less than that, probably 5-10 metre. Some reports mention very high tsunamis in Morocco, but this is disputed: the wave heights may have been around 5 metres instead.
In the majority of places, the reports state that the water initially came up: there are far fewer descriptions of the water initially receding. That may be related to the fact that it was already low tide, and an initial outflow may not have been noticed.
Elsewhere in Europe, to the north, the earthquake triggered resonance waves (seiches) on lakes, rivers and harbours. They reached 60 cm amplitude in Loch Ness, and were observed from Switzerland to Finland including the Netherlands and Stockholm, and possibly even along the coast of Greenland.
In the UK, large waves were seen along the south coast but little actual damage was reported. Computer models indicate that the region between Penzance and Lizard Point, near the tip of Cornwall, would have been worst affected; elsewhere wave heights were less than 2 meters. The extensive continental plateau around the British Isles disperses the waves and provides a degree of protection against tsunamis. But Spain and Portugal lack this protection.
After the earthquake, fires broke out, some from cooking fires and candles, and some set by looters. These quickly raged out of control and large areas of downtown Lisbon but also surrounding districts burned down. Some buildings which had escaped major earthquake damage, including the Royal Palace and the Patriarchal Cathedral were instead gutted by the fires. Lisbon burned for 5 days. In medieval times, fires were a major hazard for cities as the narrow streets and inflammable building materials made them impossible to control. Once a fire started, chances were that the entire town would burn down, with many casualties. Better control of building materials later improved the situation but city-wide fires remained common.
Lisbon was destroyed by earth, water, and fire, but most of the physical damage was done by fire.
The death toll in Lisbon is generally given as between 30,000 and 60,000. Recent evaluations of the historical records have suggested that some 12,000 died in Lisbon, close to 10% of the population. Theses studies put the total death toll across Portugal and Spain at 15,000 – 20,000, in agreement with the 14,000 recorded fatalities in Portugal and Spain. The death toll from the tsunami is estimated at 900, of whom 600 were on the sinking pier.
The recorded population in Lisbon fell by just over 31000 people, from before to two years after the quake (these records exclude young children below the age of 7). The decline includes people who moved away from the destruction and it can be considered as an upper limit to the death toll. Portugal was well governed and the records are believed to be accurate. Taking an average of the minimum and maximum numbers gives perhaps 20,000 casualties in Lisbon. The death toll away from Lisbon was likely a few thousand, noting that the badly-hit Algarve was thinly populated. This gives a total estimate of 25,000, a bit lower than commonly quoted.
The number of casualties in Morocco is not well known. The coastal towns were hit but how badly is disputed. There was another, unrelated earthquake in the Southern Atlas mountains on Nov 18 or 19, which although weaker caused destruction and casualties in in-land towns such Fez, and its effects are often confused with the Nov 1 earthquake. Casualties attributed to the Lisbon earthquake may have been very limited, or have been as high as 10,000.
The total death toll of the Lisbon quake is thus in the region 25,000 – 35,000. This was not the deathliest European earthquake in recent history (60,000-80,000 died in the Sicily earthquake of 1908, including 2,000 in the tsunami). But the Lisbon earthquake hit the heart of an empire.
The economic cost was extreme. In Portugal, it amounted to 40% of the national GDP. This is among the worst on record. For comparison, the 2004 Indonesian earthquake and tsunami led to damages of order 7% of GDP in all affected countries apart from the Maldives. (For those of a curious nature: the direct cost in the US of the 2008 financial crisis came to 15% of GDP.) The 1755 Lisbon event was a human disaster but an economic calamity.
Taxes were increased by 4% to pay for reconstruction and for safeguarding the devastated regions from pirates. Even so, the cost exceeded the capacity of the Portuguese economy. Portugal ceased to be the economic powerhouse it once was. In many ways the events of 1755 signalled the beginning of the decline of its empire.
Characteristics of the quake
The Lisbon earthquake has some unusual aspects. These include the type of damage, and the extreme distance over which the seiches were seen.
Multiple reports note how large buildings, such as castles and churches were worse affected than smaller houses. Many of these buildings, supported by arches, did not survive the shaking. Large buildings have low resonance frequencies. Earthquakes bring shaking with a range of frequencies, both high and low, but for this earthquake, the low frequencies (slow shaking) seem to have dominated.
Also notable were the three different phases. The first phase showed minor shaking accompanied by a strong rumble. These were the so-called P-waves (P stands for primary), a sound wave through rock, often heard but rarely felt. P-waves travel at typical velocities of 4-6 km/s and are the first waves to arrive. The second phase would have included the S-wave (S for secondary), which travels about 40% slower. S-waves are much stronger than the P-waves.
Reports from ships in the harbour mention how the city seemed to be rolling back and forth. This is typical of a third type of waves, Rayleigh waves. While P and S-waves travel through the rock, Rayleigh waves travel over the surface of the Earth. They give a rocking motion, back and forth, and up and down. Rayleigh waves have a typical period of around 20 sec, longer than S and P-waves which have periods of around 7 sec. A longer period means a lower frequency: strong Rayleigh waves could therefore explain the pattern of destruction, worse for larger buildings.
Rayleigh waves fade more slowly with distance than other kind of waves: this and their low frequency can explain the seiches seen on lakes thousands of kilometres away, without requiring an excessively large earthquake.
Surface waves are produced in shallow earthquakes. They come in two types, Rayleigh waves and Love waves (the latter rock sideways, a motion not reported in Lisbon). Love waves tend to be absent from oceanic earthquakes. They strengthen from reflections of the Moho separating the crust and mantle, but the Moho is much less pronounced underneath oceans. This indicates that we are looking at a shallow earthquake, somewhere in the ocean, away from the continental shelf.
Houses in Lisbon streets with a south or south-west orientation appeared to have had less damage. These houses supported each other. Thus, the Rayleigh waves came in from that direction. The earthquake occurred somewhere in the deep Atlantic Ocean, south-west of Lisbon.
Finally, the second and third phase of the quake, felt both in Lisbon and Cadiz, were too far apart in time to be different waves from the same event. It is likely there were in fact two earthquakes in quick succession. As the strength ratio differed in Lisbon and Cadiz, the two quakes probably did not come from the same location. The second one may have been a bit closer to Lisbon.
Distance and location
The earthquake happened roughly south-west of Lisbon, and there is no doubt it was out at sea. If nothing else, the tsunami provides a clue! The indications for Rayleigh waves also fit this. The absence of Love waves suggest a location off the continental shelf, in the deep ocean. However, there is a lot of deep ocean in that direction. Can we get an indication for how far away it was?
The image shows the shaking map, which shows that the worst effects were along the coastal strip, from Lisbon down around the tip of Portugal. Between Portugal and Morocco is the Sea of Cadiz. The map indicates that the epicentre would have been to west of this, closer to Portugal than to Morocco.
A first indication of the distance from Lisbon comes from the time delay between the P-wave and the S-wave. Typical P-wave velocities through crustal rock are 4-6 km/s, whilst S-waves travel at 2.5-3.5 km/s. There was a 2.5 minute delay between the two: this gives a distance to the epicentre of roughly 300 kilometre.
There was also a time delay between the earthquake and the tsunami, around 25 minutes for Lisbon. The tsunami travels much slower: a 25-minutes delay is roughly 300 kilometre.
The image shows three proposed locations for the earthquake (it is taken from a study of tsunami risks to the UK). The distance of 300 km from Lisbon is consistent with locations A or B. Location C may be a bit too close to Lisbon although it is not excluded. If there were indeed two earthquakes, they may not have the same location and only one of the two may have caused the tsunami.
The Lisbon earthquake is often stated to have been an M9. This extreme size should be taken with caution. Earthquakes of this size only occur in subduction zones, and there is no such zone south-west of Lisbon (sometimes it is argued that there may be one developing, but this is based on the presumed size of the earthquake and the argument becomes circular). Calculations of the magnitude based on the damage and extent give lower values. The most recent estimate of the magnitude is 8.5+-0.3 (Martinez-Solares & Lopez Arroyo 2004). Earlier authors estimated M8.7+-0.4.
Now consider that weaker earthquakes are much more common than strong ones. If you have a range of possible magnitudes (8.2-8.8, in the first case), it is more probable to be in the lower range as these occur more often. There is a trick of statistics for taking this into account. The black curve in the figure shows the published likelihood distribution for the magnitude, 8.5+-0.3, (assuming a normal distribution). The dashed line shows the probability of an earthquake of a certain magnitude. The actual probability is obtained by multiplying the two: that is the red curve. This reduces the probable magnitude to about M8.35+-0.25, say M8.4.
The 1755 event was thus likely low-to-mid 8. This is just about within the range of what the off-shore area is capable off. M9 would be difficult to explain.
There is little doubt that the Lisbon event is connected to the Gibraltar-Azores transform fault, separating the African and European plates. Near the Azores, in the middle of the Atlantic ocean, this fault is clear (see this excellent article for a description), but follow it towards Europe and the fault becomes diffuse: rather than a single fault line: it becomes a series of parallel, short features over a width of 200-300 kilometre. Close to the Azores it is a transform fault, with sideways motion at a snail-pace of 4mm per year. Near the continent the two plates move towards each other. There is no subduction: subduction happens if one of the plates is colder (denser) than the other, and therefore sinks. In this region, both plates have the same age, temperature and density. Like sparring politicians, neither is willing to give way. In the absence of subduction, the region becomes a crumple zone, and tries to form an oceanic mountain range.
Some earthquake data suggests a new Africa-Europe plate boundary may be forming, which in the future will replace the current diffuse boundary. In such a complex region, the plate boundary can move, with bits of plate jumping back and forth between Europe and Africa, a kind of pass-the-parcel.
The figure above shows the location of recent earthquakes around the Azores-Gibraltar fault, and is taken from Bezzeghoud et al 2013. An earthquake-free zone is visible between 24 and 18 degrees longitude: this gap is the seismically inactive Gloria Fault (GF). East of this, from 18 to 11 degree longitude, the Gloria Fault enters a complex region which has relatively frequent major earthquakes, include an M8.4 in 1941, M7.8 in 1969 and M7.0 in 1975. Around the Gorringe Bank (GB), the fault becomes indistinct and deformation is spread out over a width of 300 kilometre: this is the diffuse region. While elsewhere along the fault the earthquakes are shallow, in this region they extend to 150 km depth.
The Gorringe Bank (sometimes called the Gorringe Ridge; it is named after the discovery in 1875 by Captain Henry Gorringe) may be a crucial part of the story. Whereas elsewhere the two plates slide along each other, with varying degrees of violence, here the earthquakes indicate a head-on collision. In the absence of subduction, there has been as much as 50 kilometres of crustal shortening and folding. An underwater mountain chain has formed; two of the seamounts, Gettysburg and Ormonde, reach to less than 100 metres below the sea surface. (During the ice age parts of it would have been near the sea surface, and perhaps wave erosion have affected the peaks.) To the north is the Tagus abyssal plain, 5100 meter deep, and to the south is the Horsehoe abyssal plain, only slightly shallower. The latter is where the Gloria Fault would run had it been well defined, and is the location of most of the large earthquakes. Remove the sea, and the Gorringe Bank would be the second highest mountain chain in Europe, higher than the Alps! At the northern edge of the Gorringe Bank runs the Gorringe Fault.
Early reconstructions assumed that the Lisbon earthquake occurred in the Horsehoe plain, where the 1979 earthquake was located. But the timings of the tsunami do not quite fit, and they favour a location a bit north. That would place it in or near the Gorringe Bank. But this is also problematic as the Gorringe Bank appears inactive. A third possibility is the MPF (Marques de Pombalt Fault), but this seems too small for such a major quake. It has been suggested that the MPF extends further north, perhaps connected to the Lower Tagus Valley fault (underlying Lisbon), and that the region in between snapped. Finally, a horizontal fault plane has been suggested underlying the region. The strongest argument for this is based on a lack of evidence against it! It is perhaps a bit worrying that 260 years after one of the strongest and most destructive earthquakes in modern Europe, the fault which caused it has gone in hiding.
Part of the solution may come from the study of the tsunami wave heights in the Americas. These are quite dependent on the location and orientation of the fault. Detailed modeling has shown that it is not easy to get strong tsunamis in the Caribbean from Spain. The only model that predicted this is one with a NE-SW fault within the Horsehoe plain. None of the quake models mentioned above have this! An ancient transform fault may have existed here dating from the opening of the Atlantic, but there is no reason to assume that such a 100-million-year old fault remains active. Ancient faults, like old-age pensioners, should enjoy the modern world, not try to reshape it!
The longest fault in the region is the Horsehoe fault, at 120 km length. Other faults are no more than 100 km long. The largest earthquake that can be expected from the Horsehoe fault is M8-8.3. This fits our estimated magnitude of the Lisbon earthquake of 8.4+-0.3. Larger quakes are difficult to get within this fault system, unless several faults rupture simultaneously.
The fact that Lisbon experienced two separate jolts is often ignored. This may not have been a single quake, but two in quick succession. There may be more than one smoking gun on the sea floor.
A fault ruptures at about 2 km/s, about the same as the speed of the S wave. If the fault was pointing towards Lisbon, the total duration of the earthquake is the time of rupture, plus the difference in travel time between one side and the other. That gives a rupture length of 60-120 kilometres, in good agreement with the known lengths of faults in the region. Do remember that you would need two.
The Iberian peninsula, encompassing Spain and Portugal, has been a semi-independent plate. 40 million years ago it was attached to Africa, south of a spreading ridge, now extinct, in the Bay of Biscay. Later Iberia separated from Africa, rotated 35 degrees counter-clockwise, and attached itself to Europe. This plate-jumping was part of the closing of the Tethys Ocean and formed both the Mediterranean and the Pyrenees. The fault line separating Africa from Europe jumped south to accommodate the changing allegiance of Iberia. This jump, and Iberia’s rotation, may have left fractures behind.
Risks for the Future
Recurrence times for intraplate quakes are long. For the 1755 quake, the recurrence time has been estimated as between 1000 and 10,000 years. Older coastal tsunami deposits have been found along the Algarve, similar to those left by the 1755 tsunami, approximately dated to 2400BC and 6000BC. This gives a recurrence time of perhaps 4000-5000 years. Smaller tsunamis are more common in the region, however. Under-water studies have found turbidite (deep sea mud-flow) deposits, as could have been generated by large earthquakes. Both the 1969 and the 1755 event have left well-dated turbidites. Earlier ones have been dated to 218 BC, 3010–3960 BC and 7765–8065 BC. Turbidites indicate a recurrence time of 1800 yr, but not all of these will have generated major tsunamis.
There is evidence for a large landslide on the northern edge of the Gorringe Bank. It left a scar 2 kilometer deep, leaving a hole of 30km3, and has been carbon-dated to 30,000 years ago. It is not known whether this caused a tsunami but it could have caused destruction across the Atlantic.
Further faults of Lisbon
This was not the first devastating earthquake to hit Lisbon. The Tagus river follows a fault line, and large earthquakes can and do occur along it. In 1531, the city was hit hard, when an earthquake along this fault struck the centre of Portugal, east-northeast of Lisbon, with an estimated magnitude of M7. The main shock occurred on 26 Jan, early in the morning; there had been two precursor shocks earlier that month. A third of the city was destroyed, and the shaking caused flooding in the Lisbon estuary, possibly due to a landslide in the river. Some reports suggest that the housing stock may have been worse affected than in the 1755 event. If correct, apart from a lesser quality of the houses, it may also mean more intense high frequency vibrations. There was also a severe earthquake in 1321, again with widespread destruction, and other significant quakes occurred in 1147 (also leaving the city, just captured from the Moors, in ruins), 1334 (destroying the cathedral roof) and 1356. In 1909, a quake hit north-east of Lisbon, with a magnitude of 6.5. Since 1909, the region has been seismically calm.
There have been many discussions of the chances of a re-occurrence of the 1755 event. But such an event may only occur once every 5000 years or so, leaving the next one a long way off. A more immediate danger is a potential repeat of those medieval quakes. Lisbon is a beautiful city built in a dangerous place.
-Review of the 1755 Lisbon Earthquake Based on Recent Analyses of Historical Observations. C.S. Oliveira, 2008, in: Historical Seismology. Modern Approaches in Solid Earth Sciences, vol. 2, p 261-300
– The effects of the 1755 Lisbon earthquake and tsunami on the Algarve Region, Southern Portugal. David K. Chester, University of Liverpool
– The Historical Review of the 1755 Lisbon Tsunami. Angela Santos, Shunichi Koshimura. Journal of Geodesy and Geomatics Engineering 1 (2015) 38-52
– Tsunamigenic earthquakes in the Gulf of Cadiz. L. Matias et al. Nat. Hazards Earth Syst. Sci., 13, 1–13, 2013
Large, deepwater slope failures: Implications for landslide-generated tsunamis. Claudio Lo Iacono et al. Geology (2012), 40, 931-934
– Seismicity along the Azores-Gibraltar region and global plate kinematics. M. Bezzeghoud et al., Journal of Seismology, (2014) 18:205-220
Far field tsunami simulations of the 1755 Lisbon earthquake: Implications for tsunami hazard to the U.S. East Coast and the Caribbean. Roy Barkan, Uri S. ten Brink Jian Lin, Marine Geology 264 (2009) 109–122
– Holocene earthquake record offshore Portugal (SW Iberia): testing turbidite paleoseismology in a slow-convergence margin. E. Garcia et al, Quaternary Science Reviews 29 (2010) 1156–1172
The great historical 1755 earthquake. Effects and damage in Spain. J. Martinez Solares, A. Lopez Arroyo, Journal of Seismology 8: 275–294, 2004.
– The 1755 Lisbon tsunami; evaluation of the tsunami parameters. M. Baptista et al., J. Geodynamics Vol. 25, pp. 143-157, 1998
– Earthquakes and tsunami in November 1755 in Morocco: a different reading of contemporaneous documentary sources P. Blanc, Nat. Hazards Earth Syst. Sci., 9, 725–738, 2009