Leeds
AS THE weary inhabitants of Beirut struggle to rebuild their war-ravaged city, evidence is emerging that it faces another threat. In AD 551 ancient Beirut was levelled, not by civil war but by a natural disaster. Some historians think the devastation was wrought by tsunamis triggered by earthquakes elsewhere in the eastern Mediterranean. But new archaeological excavations in downtown Beirut point strongly towards a local event-an earthquake so severe that scarcely a pillar, column or arch of the city built by the Romans was left standing. What鈥檚 more, though the standard wisdom is that the city should escape the worst of the region鈥檚 quakes, research published this month reveals an active earthquake fault running right through its heart.
My interest in Beirut鈥檚 fate arose almost by accident. I am a structural geologist, so my business is to try and understand how the Earth鈥檚 crust is reshaped by faults and folds as the tectonic plates interact. Lebanon should be a good place for investigating such matters, because it lies astride part of the world鈥檚 network of plate boundaries. But because of the long civil war, the country has seen virtually no detailed geological study since the invention of plate tectonics. For the past two years, I have been working with Sara Spencer of the American University of Beirut to establish how the uplift of the country鈥檚 mountainous backbone, the 3000-metre-high Mount Lebanon chain, relates to the region鈥檚 major seismic faults. Coincidentally, our research has revealed vital clues to the likelihood of earthquakes happening elsewhere in the country.
In June this year I paid my latest visit to downtown Beirut-until recently part of the 鈥淕reen Line鈥�, a devastated corridor of pancaked reinforced concrete and gutted buildings. Now regeneration is under way, and there are ambitious plans to make Beirut one of the world鈥檚 great financial centres. High-rise, high-tech buildings will cover an area extending into what is now the Mediterranean, on land reclaimed using the bulldozed rubble of the war zone. Putting large buildings on such unconsolidated foundations requires considerable confidence that there will be no big earthquakes.
Advertisement
As bulldozers prepare the area, they are uncovering the old levels of the city, which ironically hold evidence of past quakes. Frantic teams of archaeologists are racing the construction crews to record what they can before the new foundation piles go in. I visited one excavation site with Ziad Beydoun, emeritus professor of geology at AUB, and his archaeologist wife Muntaha Beydoun. She showed us the Roman high street, its granite columns and facades reworked into the walls of the buildings erected immediately after the disaster of 551. Tessellated floors had collapsed onto the 6th-century inhabitants in their workshops-scenes of devastation akin to those at Pompeii. This archaeological evidence leaves little doubt that the Roman city was destroyed by strong ground motion, the hallmark of a large-magnitude earthquake, not a tsunami from a distant part of the Mediterranean.
Safe as houses
Modern Lebanon seems blithely unaware of this long-standing hazard to its capital. When Ziad Beydoun helped to draft the country鈥檚 building code, he and his colleagues had only scant knowledge of Lebanon鈥檚 geological faults to call on. For their assessment of the seismic risk they had to trust this existing wisdom and assume that Beirut should be relatively safe from quakes. Yet even the lenient code they devised is being flouted as builders struggle to keep costs down. On the slopes of Mount Lebanon, the city鈥檚 expensive suburbs seem to have been built in an exercise of competitive cantilevering, as apartment blocks arch outwards for better views. Driving into Beirut we passed one of the latest office blocks, a 10-storey edifice of gently overhanging smoked glass. 鈥淧retty good for earthquakes,鈥� says Beydoun, with sad irony.
Everyone realises that Lebanon as a whole should be prone to earthquakes. After all, the country lies astride the Dead Sea Transform-the western boundary to the Arabian plate. The transform links the new, opening plate boundary in the Red Sea with a destructive plate boundary, in southeastern Turkey, where the Arabian subcontinent collides with Eurasia (see map). Outside Lebanon, the transform forms a prominent physiographic feature: the long valley running north from the Gulf of Aqaba, through the Dead Sea itself and along the valley of the River Jordan. In the 1940s and 1950s Albert Quennell, a geologist with the Land and Surveys Department of what was then Trans-Jordan, recognised the importance of this fault system and estimated that Arabia, on the eastern side, had moved about 105 kilometres northwards compared to Sinai on the west. From geological studies in Israel we know that this slip happened in the past 18 million years. This implies an average slip rate of about 6 millimetres per year-quite enough to generate regular, sizeable earthquakes.FIG-mg21005601.GIF
According to conventional geological wisdom, the active part of the fault system passes safely east of Beirut. However, it is hard to know for sure. The southern part of the Dead Sea Transform is restricted to a narrow zone of faulting only around 10 kilometres wide. But as the transform heads north from the Jordan valley and into Lebanon, its structure becomes far more complex. The main fault zone splits into a horsetail of five major strands together with blocks of rock, raised high by the seismic slipping.
Faulty reasoning
So the question is on which part of this horsetail the bulk of the slipping takes place. Until now, the main candidate has been the Yammouneh Fault. Its southern portion runs along the eastern side of the Mount Lebanon range, forming the western edge of the Bekaa Valley. But towards the north of the country the fault splits the high ground of Mount Lebanon. This pattern is particularly clear on satellite images-the only way there has been of surveying Lebanon during the civil war-so most geologists have interpreted the Yammouneh Fault to be the main active arm of the Dead Sea Transform. If they are right, then this fault is likely to be the channel for most of the energy released by earthquakes, leaving the rest of Lebanon with a much lower seismic risk.
It is this model that underpins the building code that Ziad Beydoun helped to draw up. Rather more than $64 000 rests on the question of whether it is correct.
The first port of call for any earth scientist asking this sort of question is the catalogues of the International Seismology Centre in Newbury, England, which records all the earthquakes ever detected anywhere in the world. The data come not only from the modern seismometers but also from historical accounts stretching back over the centuries. Look at the ISC catalogue for Lebanon and you will find a swarm of earthquakes in the southwestern corner of the country, but very few in the north. Much of the Yammouneh Fault is disturbingly quiet.
There are two possible reasons for this. The first is that the fault slips only rarely, causing a big earthquake when it does. If this is so, the seismic stresses in northern Lebanon should be growing in advance of a big, releasing earthquake, possibly in the near future. If there were seismometers in place in the region, they would be able to detect the tiny tremors that presage such a quake. Unfortunately, there are none. The alternative explanation is that the fault might be dead. This would transfer the strain of slippage to other parts of Lebanon, away from the Yammouneh, and put them at much higher risk of a serious quake.
So how can we tell if the Yammouneh fault is dead, or just resting? In search of an answer, Spencer and I headed off to a valley in the northern part of the country, on the edge of the limestone highlands of Mount Lebanon near the border with Syria. These rocks had been uplifted and shifted by earth movements associated with the Yammouneh Fault, but in this region they plunge beneath a covering of lava flows called the Homs Basalt. Where the southern edge of the basalt crosses the trace of the Yammouneh Fault it is deflected by about 10 kilometres-suggesting that the fault has moved by this amount since the lava first formed. The basalt is believed to be less than 5 million years old, implying an average slip rate of at least 2 millimetres a year. This should be fast enough to produce devastating earthquakes every few hundred years.
Shattered rock
In 1995, Spencer and I decided to take a closer look at the relationship between the Homs Basalt and the Yammouneh Fault. We started out in Wadi Chadra, a quiet valley of small citrus orchards. After parking at the last farm, we picked our way through the prickly vegetation and scrub, following goat tracks southwards along the Yammouneh Fault. Shattered outcrops of bright white limestone on our left bore witness to the destructive power of ancient earthquakes. After about a kilometre we came to a side valley heading westwards into gullies that cut back into the basalt. Here we could look at the basalts and how they had been affected by past movements and earthquakes on the Yammouneh Fault.
Working up the side valley we crossed limestones that became increasingly shattered. The last outcrop was a weakly cemented, intensely ground-up material called fault gouge. We had expected this-the product of grinding along a fault. However, if the grinding had continued after the lava flowed, there should also have been ground-up basalt in the fault. But wherever we looked, the only fragments we could see were limestone. Right next to the limestone fault gouge were outcrops of basalt, barely fractured. There was only one conclusion: the fault must have stopped moving before the eruption of even the oldest basalts in Wadi Chadra.
But if the Yammouneh Fault is dead, why does it seem to jut 10 kilometres into the Homs Basalt? Looking out from Wadi Chadra the answer was obvious. The limestones to the east of the fault form a high ridge called Jabel Akroum, 300 metres above the basalts to the west. We concluded that the lavas flowed around Jabel Akroum like the sea round a promontory. Samples of basalt that we collected later were between 6.5 and 5.2 million years old. The Yammouneh Fault, at least in northern Lebanon, had been dead for one-third of the transform鈥檚 history.
So what is taking the strain of the transform as it passes through Lebanon? We decided to check the southwest of the country, where modern seismic activity seems to be concentrated. We hit on the Roum Fault, named after a village in the Chouf hills 35 kilometres south of Beirut, which runs along the eastern edge of the Tyre-Nabatiyet plateau. A few million years ago, this plateau lay below sea level, but it has since been raised more than 400 metres-another legacy of tectonic activity.
This is not easy country for geological fieldwork. The uplifted ridges on the western flank of the fault are prominent enough to have been of strategic importance for centuries. Crusaders built their stronghold of Beaufort Castle on the southernmost of these ridges, overlooking a spectacular bend on the Litani river. Although Beaufort was destroyed by shelling when Israel invaded Lebanon in 1982, it retains its strategic importance and is occupied still by Israeli troops and armour maintaining the 鈥渟ecurity zone鈥� of south Lebanon. In the land below, minefields are strewn over much of the now desolate landscape. On the few roads, vehicles cannot stop without fear of being taken for terrorists and fired upon by the Israelis.
The right stuff
These problems, together with the civil war, had made it impossible for geologists to look at the Roum Fault. But during a lull in hostilities in December 1995, Spencer and I made a foray into the south. First we had to establish whether the Roum was the right kind of fault to be part of the transform system. Sure enough, striations on the fault surface showed sideways motion, which is diagnostic of transform faults.
Next we needed to establish how much movement had occurred on the Roum Fault. For this we had a ready-made set of markers: the rivers of southwest Lebanon are deflected across the Roum Fault, showing that the eastern side has moved northwards over time. To work out how far the fault had moved, we needed to measure the deflection and establish the ages of the river valleys concerned.
The age is usually the tricky part of this equation. But in this case we knew that many of the gorges dated back to the Messinian stage, between 5 and 6 million years ago. At this time, the Mediterranean was isolated from the rest of the world鈥檚 oceans and the eastern Mediterranean fell to more than 2 kilometres below its present level. Rivers do not cut down farther than the prevailing sea level. Yet modern rivers running across the Tyre-Nabatiyet plateau lie in deep gorges extending far beyond today鈥檚 shoreline-showing they have existed since the Messinian.
Knowing the age of the gorges, it should be a simple matter to establish the distance the fault has slipped by measuring their offset across the Roum Fault on a map. However, it is easy to be fooled. Rivers can alter their courses for reasons quite unrelated to earthquakes. For instance one river can 鈥渃apture鈥� the drainage basin of another. To avoid this problem, we decided not to look at just one or two rivers, but to analyse the features of the region as a whole.
So we turned to a map showing river systems and their headwaters in southern Lebanon and northern Israel, and taking a pair of scissors we cut along the Roum Fault, from the Jordan valley to Beirut (see p 47). If we moved the eastern side of the map southwards we should be able to match the various headwaters that exist to this day with their ancestral outflows, whether or not they still contain a river.
Great divide
The most prominent feature on the western side of the fault is the Litani valley, the greatest gorge in this part of the Levant. On the other side of the divide, working north from what are now the headwaters of the Litani, we come the rather small drainage basin of the Zahrani river. It seems unlikely that this small catchment area would ever have provided the water to carve the Litani gorge, so we looked further. Sure enough, north of the Zahrani is the much bigger drainage basin of what is now the upper Aouali river. When we shifted the two sides of the map so the Aouali headwaters fed the lower Litani, all the catchment areas lined up behind gorges. We had reconstructed the Messinian geography of the rivers.
The difference between this ancient arrangement and the modern one requires 30 kilometres of slippage on the Roum Fault. If this has occurred in the 6 million years since Messinian times, then the long-term average slip rate on the Roum Fault has been 5 millimetres per year-close to the total slip rate of 6 millimetres per year for the whole transform. In other words, the Roum looks to be the main active segment of the transform in Lebanon. And here鈥檚 the worrying part: it is heading straight for downtown Beirut.
Spencer and I published our findings this month, and we hope they will give the developers rebuilding Beirut some food for thought. But much remains unknown. Though we have established that the Roum Fault is active, we don鈥檛 know what the earthquake activity is like. A quake of magnitude 5 occurred on the fault at Easter this year. Should we expect many more small quakes like this one, which caused only minor damage? Or does the Roum Fault typically slip with large earthquakes, like the one in 551? The lack of an adequate network of seismic monitors makes it impossible to decide.
Lebanon鈥檚 National Research Council has proposed an array of six instruments to be scattered throughout Lebanon. They have yet to be installed, and in any case their sites have been planned on the assumption that the Yammouneh Fault is the principal source of earthquakes. Virtually nothing is known of the recurrence interval of the devastating high-magnitude earthquakes in Lebanon. In California, geologists investigate faults by digging trenches and looking for disrupted soil horizons which can be dated using radiocarbon or other methods. Nothing like this has yet been done in Lebanon.
But the destructive effects of earthquakes themselves can yield valuable data, too. It is not only buildings that fall down during big quake. The delicate carbonate edifices that slowly build up inside caves also collapse, and these toppled and tilted stalagmites can also be dated. This approach was pioneered by Paolo Forti of the University of Bologna, who has used data gathered in a cave in Italy鈥檚 Apennine mountains to build up a record of the country鈥檚 major earthquakes, which he has linked with historical accounts. The Lebanese Speleo-club is currently starting such a study in caves close to Beirut.
All these investigations would greatly enhance the evaluation of earthquake hazards. But in the meantime Beirut鈥檚 developers have a problem. Plans to rebuild the city are well advanced. Capital has been raised on the world鈥檚 stock markets, and construction work has already started. But if a devastating earthquake strikes again, will Beirut be ready?
