Late on 8 September 2023, a major earthquake hit the region of the High Atlas mountains in Morocco. As of writing, the full extent of the damage is not yet known. The ancient city of Marrakesh, just north of the mountains, was hit badly. Settlements in the mountains themselves have not yet been reached. The precise details of the earthquake itself are also not known yet: the best numbers at the moment are magnitude (moment) M6.9 and fairly shallow (10-30 km) but this may be revised. The earthquake occured close to Morocco’s highest mountain, Djebel Toubkal, measured (pre-quake) at 4167 meters. This mountain has been said to be volcanic, the only one in the High Atlas. It is indeed built on a basaltic plateau which is Cambrian or even Edicaran in age (550 to 650 million years). Clearly the mountain itself is younger: the mountain is dated to 3 to 5 million years ago. But little is known about any volcanic eruption, and perhaps this only ever referred to the ancient volcanic plateau.
About 4500 years ago, the Toubkal region suffered the largest collapse known in North Africa, the Arroumd rock slide, which was likely triggered by an earthquake, perhaps similar to that of last week. Old, high mountains or volcanoes can be a risk for large collapses.
The news stated that the recent earthquake was by some margin the largest recorded in Morocco. That is true for the 20th century, when the largest event was an M5.9 in Agadir in 1960. But a similar size earthquake is known from November 1755, shortly after the the Lisbon earthquake and likely triggered by it. That earthquake, though, was in the north of Morocco and not related to the high Atlas. Other large historical earthquakes were also in the north of Morocco, but smaller events are more common across the country. The current event occured midway between the 1960 Agadir quake and the 1967 Mgouna earthquake (M5.1), both on the southern border of the High Atlas. Further to the east, the 1986 Tinghir earthquake (M5.1) and the 2019 Midelt (M5.0) earthquake are about equally spaced along this track. The current event may appear to fill in a gap between these previous quakes, but the fact that it is so much larger suggests it is more than that. The second earthquake last week, an M4.9 a little south of the main event, would fit this sequence better. And it is a rare event. Marrakesh has a 1000-year history, but none of the records mention damaging earthquakes. The time between large earthquakes may be anywhere between 1000 and 5000 years.
Morocco and the Atlas Mountains
The High Atlas run across southern Morocco, from Agadir on the coast to Algeria. To the south is the Sahara. To the north and west is the Meseta plateau. There are more mountain ranges in Morocco, almost all carrying the name ‘Atlas’, but the High Atlas are the highest and longest. In fact, the High Atlas run for 2000 km, ending only in Tunisia.
The existence of fault zones along the High Atlas has long been known. The South Atlas Fault was found in 1934, running along the southern edge. It was seen as a natural division between Africa and Europe, and it was proposed to be a transform fault with movement of 200 km. The nature of this fault was later reconsidered. The evidence for transform motion is not strong: it is not really visible in the landscape. The fault also seemed to be rather fragmented. Nowadays we see it as a collection of shorter separate faults, which together create a fault zone along the mountains.
A more recent view is shown in the figure above. It lists many shorter faults. The South Atlas Fault is now renamed as South Atlas Fault System, a vaguer term whilst keeping the SAF abbreviation. There is a fault further south, running along the Anti-Atlas Mountains, the AAF. To the north is the North Atlas Fault System, while the TAF (Trans Alboran Fault) runs off at an angle to the northeast. Both of these have had earthquakes before: the NAF had a number of high M4 events during a few years around 2010, and the TAF had a few such events in 1934 and 2006.
The High Atlas Mountains themselves are riddled with short faults. Why so many faults? And why does Morocco have so many mountains anyway?
In the north of Morocco are the Riff mountains, a series of ridges running southeast and east. They are a direct relation to the mountains in Gibraltar and Spain. They show that the current collision zone between Africa and Europe runs along this this line. In Spain it bends north and even northeast, delineating the western edge of the Mediterranean Sea. The collision of Africa and Europe is not, and has never been, straightforward. It involved many parts making early crossings, changing affiliation and now acting for the defence. Iberia is one of these fragments. It originally was related to Cornwall and Brittany, and was perhaps even part of the proposed microcontinent of Armorica. Over 150 million years Iberia rotated by 180 degrees, moved sideways, and slotted into place some 30 million years ago. The Pyrenees are the trace of the final reconnection. The curved fault line from the Riff through Spain shows the effect of Iberia’s rotation. A little (very little) of that story is in our Christmas Ghosts story.
South of the Riff mountains are the Middle Atlas mountains. This chain is about 300 km long and consists of four long ridges. This mountain grew from a Jurassic valley. It is a remnant of a basin which formed before and during the opening of the Atlantic Ocean. The Jurassic basin was mostly part of the Tethys Ocean but at the western end it was Atlantic. The basin formation thinned the lithosphere. Later, there was contraction, caused by a mixture of Spain rotating and sliding into pace and the continuing slow motion of Africa northward. This contraction together with the thin lithosphere pushed up the region. The basin inverted and became an elevated region. There is still a basin further to the east, a deep desert extending across Algeria and bordered by the Sahara Atlas mountains in the south and the Middle Atlas in the west.
South of the High Atlas is the Anti Atlas. This is an elevated area, similar to the Meseta. It has been called an ‘upwarp’. The thin lithosphere runs along the line of the Middle Atlas and along the edge of the Anti Atlas. The thin lithosphere allows for buoyancy: a warmer mantle here has pushed up the region. The uplift amounts to around 1 km, and it occured in the last 10-15 million years. This mechanism is responsible for much of the Anti Atlas uplift and about half of that of the Middle Atlas. Both ranges contain a volcanic centre which both were active some 10 million years ago. A magma reservoir pushed up each of these places. The better known is the Siroua mountain, a remnant stratovolcano.
The High Atlas is a bit different. It runs at an angle to the Middle Atlas- Anti-Atlas (the ’hot line’). The High Atlas was raised up by contraction – pushed up by horizontal forces while the other mountains were raised from below, i.e mantle upwelling. Africa is still moving into Europe at 5 mm per year. (It is also sliding westward at 24 mm per year relative to Europe.) Diverting actions, such as using Spain and Italy as cover, have not stopped the motion. Some of this movement is accommodated in the Atlas mountains. This has caused earthquakes which are mainly in northern Algeria but there have been seismic events in the Morocco High Atlas too, including the M6 in Agadir in 1960. The area is not particularly active but does show up in maps of seismic events.
The High Atlas consists of three branches: the western, the middle and (no surprise here) the eastern. The division between the west and the middle lies at the Marrakech High Atlas, the massif that peaks at Djebel Toubkal. The main seismicity has been in the west (Agadir) and east (Midelt). The current event was at the Toubkal massif.
GPS measurement show that the main motion of north Morocco relative to the African plate is WNW-ESE (which is under a slight angle to the High Atlas mountain range here), by 4-5 mm/yr. This move is in the same direction as Spain, albeit smaller, and it appears that Morocco north of the High Atlas is acting as an initial buffer zone for Europe. The full motion of Europe with respect to Africa is 24 mm/yr, so Morocco accounts for about 20% of that. It is a useful stress relief for Europe, but the motion leads to some compression in Morocco.
The High Atlas accommodates around 1-2 mm/yr of this motion. GPS data has shown that the central High Atlas moves by only 0.14 mm/yr, rising to 0.72 mm/yr for the eastern High Atlas (Ciunha et al 2012, Geophys J Intern 188(3): 850–872; Medina & Cherkaoui 2021, Arabian Journal of Geosciences (2021) 14: 1717). In hindsight, this might have been a warning sign. The manyfold of High Atlas faults can act as a dispersal for the movement, limiting the slip on each individual fault. But the movement on one fault can also increase the stress on the next one. The fact that the observed movement was so much less in the central section could be seen as accumulating stress – a locked region.
The Tizi ‘n Test fault
Which particular fault failed? That is hard to tell from the available data. There may have been two failures, as the main event and the second smaller earthquake appear to be on different sides of the mountains and not along a single fault. There is however a main suspect (or a usual suspect, in Casablanca terms) which would be the so-called Tizi ‘n Test fault, TTF in the map above. Does it fit?
The 2023 earthquake is currently classified as ‘oblique-reverse fault’. This situation means a steep fault where the movement is up on the overhanging side (blue on the diagram). (The opposite of ‘reverse’ is ‘normal’ where the yellow part would move up with respect to blue.) ‘Oblique’ means that the fault is steep, between 45 and 90 degrees. If it is shallower than this, it is called ‘thrust’. The movement that we saw made the High Atlas even higher. Was this related to the fact that it was next to the highest mountain of the HighAtlas, Djebel Toubkal? The area that failed is estimated at 20 to 30 km by USGS.
The upward motion was on the southeast side. This earthquake was caused by the movement of the Meseta to the east-southeast causing compression.This compressional stress was relieved in this event.
Let’s come back to the main suspect. The Tizi ‘n Test fault is best known from a scary mountain road. It was originally seen as ancient, millions to tens of millions of years. But it reactivated in the Pleistocene. The fault runs in the right place and has the right shape for the 2023 earthquake. The suspect can be charged.
There is comprehensive evidence for past activity of this fault. Detailed study of river valleys around the faults showed evidence for slope changes, reversal of river flows, changes of river directions across the fault and catastrophic rockfalls, showing that the fault at one time re-activated. This was the uplift of the Central High Atlas which had happened along this fault. The activity was not dated and perhaps the impression was given that the fault had gone dormant again. But the uplift activity agrees closely with what is seen in the 2023 earthquake. Whether the rockfalls have repeated as well is something that will have to see from satellite images.
How often could this happen? The amount of movement is estimated at 1.7 meter at depth (not much surface rupture is expected by USGS). For 1 mm/yr, that is is 1700 years. But the movement is mostly parallel to the fault. Assuming this angle is 25 degrees to the fault, the compression component is 2.5 times less. That would give a time scale of 4000 years. The correct number may be in between.
Why here? On the image, the Tizi ‘n test fault runs more northeast here, before turning again east. This puts the movement of the block to the north more perpendicular to the fault. So this region becomes a focal point for the compression. This may explain why Djebel Toubkal became so high: it is pushed up more than any other area because of the direction of the fault line along it. No need to invoke a volcano!
This suggests that there may be much less risk of a similar event on either side. The USGS estimate of 30 km of failure corresponds to approximately half the length where the fault runs northeast. Much of what could break has broken in the earthquake. In this situation, the risk of powerful aftershocks may not be as high as usual.
The flank failure of Djebel Toubkal occurred 4500 years ago. It is possible that this was the last time the region suffered a failure like last week! Or there may have been one such event in between.
The damage from this earthquake is major. Could more have been done? It is dangerous to judge after the fact. This is not Turkey, where building codes existed for very good reasons, but which were not enforced. The buildings in the region are often old and even historic. They have survived for a very long time. The 1960 earthquakes in Agadir did tremendous damage and showed that earthquake risk was present. In fact, Agadir had also been destroyed in 1731, so there was history, although this was not recognized until later.
But Marrakesh did not have such a history. Its buildings had not been tested against the faulting. The rebuild will need to take more care. But it is hard to see how this could have been avoided. Remember that the southeast of the UK is also susceptible to earthquakes up to M6. How resilient will the buildings there be? I hope we will never find out. Few places on Earth are prepared for even a one-in-a-millennium event. This earthquake was rarer than that. And it is likely that only the particular area around Djebel Toubkal is affected. Chances of similar events elsewhere along the fault are low.
There were papers indicating potential danger. Some were even in particular about the Tizi ‘n Test fault. But these papers are technical and would be difficult to read for a non-specialist. Some are behind extreme paywalls. (The Geological Society of America is especially good for people wanting to avoid any risk of anyone actually being able to read their work.) Scientists prefer to present data and science, but would rather not go for public warnings. We are also often told to stay out of politics – by the people who don’t want to hear.
Faults are not just tectonics – people carry enough faults of their own.
Albert, September 2023