Few people would have heard of Mount Nyiragongo before its current eruption. There are some cities that we know live in the shadow of a volcano. Naples and Vesuvius, Catania and Etna, Seattle and Rainier (although the city itself is unlikely be affected by an eruption), Fuji and Tokyo (with the same proviso). But Nyiragongo is worse than any of these. It erupts often (the current eruption is number 35 since 1882). The effusive lava is very fluid and moves fast. Worse, the mountain is extremely steep for this type of eruptions, reaching 50 degrees near the summit. The lava flow can reach speeds approaching 100 km/h. And the city of Goma, with half a million people, lies directly in one of the paths the flows can take. Finally, this is in the DRC, perhaps the country worst equipped to deal with volcanic risks. Both the 1977 and 2002 flank eruptions had a significant number of fatalities in Goma. In 2002, 13% of the city was covered by 2-meter thick lava as it flowed towards Lake Kivu. And fatalities continued long after the eruption had ended.
Since 1970, there have been 25 volcanic eruptions worldwide with fatalities. By far the worst was Nevado del Ruiz, where 23,000 people died. Second is El Chichon, followed by Pinatubo, Anak Krakatau and Merapi. (In my opinion, Kazbek should be added to the list.) Nyiragonga was already the only volcano to appear in this list twice, with both appearances among the 10 worst eruptions of the past 50 years. And now it is in this list 3 times.
(The number of eruptions worldwide with fatalities has increased dramatically in recent years. Since 2010, there has been on average one such event per year. Before that it was much less. In two cases tourism was involved, but the others were people living in the wrong place. With volcanoes, precaution is the only effective strategy.)
We often discuss the dangers of explosive eruptions, and point at Krakatau, Tambora or El Chicohn. Those kinds of eruptions are huge, devastating, and unpredictable. Often the culprit was not recognized as dangerous prior to the explosion. Effusive eruptions are more sedate, tourist-friendly, and predictable but also more frequent. We acknowledge the damage they can still do, and the desperation of people seeing their home engulfed by lava. There is no such thing as a friendly volcano. But Goma finds itself in the worst of both worlds. Its mountain suffers frequent effusive eruptions which come fast and without warning, and are as deadly as explosive eruptions would be. And it is not only the lava. The gas emissions cause acid rain which can damage plant and tree growth on the mountain. Near the crater rim, underneath the volcanic plume, the rain can have pH as low as 2. Low pH and high fluoride has been measured as far as the village of Rusayo, 10 km from the summit.
Geologically, Nyiragongo is part of a cluster of volcanoes, two of which remain active. Together they form the Virunga bulge, home of wonders including the impenetrable forest and its mountain gorillas. The bulge sits at the intersection of the Albertine rift (the western branch of the African rift, running north-south – the volcano, as well as Goma and Lake Kiva are located in the rift), and the Kamatemba rift (running northwest to southeast into the Albertine rift). Exactly what feeds the volcanoes is being studied, but a crossing of two rift indicates this is a weak spot. Two magma chambers are suspected, one shallow and one 14 km deep.
The unusual lava of Nyiragongo is attributed to a melting of a carbonated mantle. It differs from all other lavas in Virunga. This seems related to the depth: its magma formed from a much deeper melt than elsewhere in the Virungas. The lava is often claimed to have the lowest viscosity of any volcano known but this is disputed. It is fluid, but the extreme speed of the lava is caused mainly by the steepness of the cone.
The steep cone is not what one would expect from low viscosity lava. It should have build a very broad shield, not a cone. The mountain cannot have been build by these lava flows. There are several possibilities. There may have been a very recent change in the magma. But none of the lavas seen in this area can build such a steep cone. The second possibility is that the mountain was not build from the outside but from the inside. The mountain is in effect the sides of the conduit. Magma rises, forms a lava lake, and the lava lake solidifies against the sides and adds bulk to the mountain. The magma builds the mountain around it. Because this builds the mountain with solidified rather than liquid rock, it can be much steeper than flowing lava would allow. The lava lake has several benches: these benches are how the mountain grows.
The 1.2-km wide, deep crater of Nyiragongo hosts the largest lava lake in the world, some 200 meters wide. The mountain is 3470 meters tall; the lava lake is typically at a height of 2700-3000 meters, while Lake Kivu is at 1470 meters. You can imagine the stress that the sides of the mountain are under! It carries a 1-km wide bucket of lava at the narrow tip of a steep cone. The sides of Nyiragongo are indeed riddled with rifts, many of which are unmapped because of the dense vegetation. These rifts can feed the eruptions, as they did in 1977 and 2002, and apparently also in 2021.
The volcano is about 12,000 years old; the current deep crater is though to have formed a few hundred years ago through caldera collapse. There have been previous collapses: there are two other but extinct craters of similar size, one located two kilometers north (Baruta: visible on the image) and one 2 kilometer south of the current crater (Shaheru). The three cratered cones are aligned: there seems to be a short north-south rift underneath Nyiragongo. This rift follows the direction of the Albertine rift and therefore may have benefited from spreading in the Albertine. In contrast, the Virunga volcanoes overall follow an east-west line. Although the parasitic cones are themselves extinct, they are not immune. During the 1977 eruption, lava erupted inside Shaheru crater.
The 1977 eruption formed a rift just above the Shaheru crater; lava flowed into this crater. The 2002 eruption showed lava fountaining above Shaheru from the same fracture. But the 2002 eruption extended this rift much further. The southern flank of Shaheru was ruptured in two places, 300 meters apart. The rift extended downward towards Goma, causing small grabens with lava fountains in several place along the way. These fissures reached the city itself.
The 2021 eruption appeared to have started just southeast of the Shaheru crater, close to previous flows. The lava appear to have come through the same fissure as in 1977. The flow split in two, with the longest one heading south, ending within the Goma suburbs close to the airport runway. In the day after the eruption, Goma was hit by several earthquakes which caused cracks in roads. This seems to have extended the activity of 2002, when minor rifting and gas emission was seen 500 meters beyond the lava flows, as far as the airport. In 2021 it reached just off-shore into Lake Kivu. A long crack in the N2 road parallel to the coast also suggests some ground movement within the lake. The rift has thus progressed from 1997 when it formed as 2-kilometer long fracture, to 2002 when it extended by 15 kilometer, to 2021 when it added a further kilometer. Note that this does not mean it is exactly the same crack. Rifts have a certain width and the new crack may be next to and parallel to the previous one.
The eruptions are fed not from a magma chamber but from the existing lava lake. This means there is little or no tremor preceding the eruption as there is no rising magma. Instead, they start when the growing pressure of the lava lake breaks through the steep sides of the cone (a fate the Geldingadalir walls are trying to avoid), involving tectonic earthquakes. In 2002, the eruption was preceded by an M4 earthquake two weeks earlier. This earthquake may be when the break first began to form. The 2021 eruption did not have a clear precursor earthquake. (An earlier claim in this post for a precursor earthquake was based on an erroneous location.) The aftermath is also the same. In 2002 there was earthquake activity on a line between Goma and Nyiragongo for 5 days after the eruption, with the strongest even M5, and in 2021 there were several earthquakes in the day after the eruption also peaking at M5. This pattern adds some predictability to the Nyiragongo eruptions.
But this expanding rifting activity points to another major danger, beyond that of an eruption in the heart of the city. This mountain seems unstable. The rifts in the side are integral to the mountain (that can be seen because many of them directly feed volcanic gasses into the water system). There could be a failure where two of the side rifts combine and one side of the mountain slides down-hill. Such an event would be catastrophic. Most volcanoes in the Virungas show no evidence for a flank failure, but the extinct Mount Sabyinyo does have a big gap in its side and on satellite mages a 1-km wide crater, 2 km southeast of Mount Karimbi also seems to have a gap on the side facing away from Karimbi.
But let’s look at Goma itself. Why build a major city in a place where it has been hit by lava streams three times in the past 50 years?
Goma is at the border with Rwanda. Gisyeni, the part of Goma on the Rwandan side is well connected to the Rwandan capital, Kigali. Travel in the DRC can be more difficult. The area is very densely populated, fertile, with mild tropical climate. The Rwandan genocide in 1994 caused many refugees to flee to Goma, with rebel movements following. The area has remained rebellious and it is not always under control of the central government. People are attracted by the mineral wealth of the DRC. This should be a fabulously wealthy country. Instead, the wealth goes to however manages to profit most. Some of the rebels have a cause. Others just want the wealth for themselves. And earlier this year, the Italian ambassador (who also helped to run the world food program here) was killed by rebels, in a kidnap that went wrong.
And in spite of the insecurity, Goma has grown rapidly. The current population is estimated at 670,000, growing at 5% per year. It is now three times the size it was during the devastating 2002 eruption, and ten times that during the 1977 eruption! New roads have been build (many by Chinese contractors), on top of the 2002 lava.
In view of such unstoppable growth, where do people live? They can’t go east, towards Rwanda, as Gisenyi is too small and itself has no room to expand. West gives some space but quickly takes them out of the rift valley. South lies the lake. So the city grows west along the lake shore, and north on the plains leading to the Virunga mountains. Much of the land is already taken, and so people build houses where they find unoccupied space. People have to get to work, using the taxis, any ride they can get, or by walking. So they live near the roads. Conveniently, there is space here which is not farmed or otherwise occupied. That is, they build their houses on top of the recent lava flows.
A time line shows the growth of Goma. The image below is from 2005, when the 2002 lava flows are still visible.
The most recent image, from 2019, shows dramatic changes.
Putting the Google earth images together (thanks Gijs, Lughduniense!) shows the rapid expansion in a time line of Goma’s growth towards Nyiaragongo..
As you can see, the expansion of the city follows the 2002 lava flows very well, along the roads build on those flows.
But the lava flowed along those paths for a reason. Nyiragongo can erupt in all directions, but in the three major eruptions of the past 50 years the lava has gone south, following the underlying north-south weakness in the mountain. This rift formed in 1977 eruption, was re-activated in 2002 and possibly again in 2021, extending through the city and now into the lake. The rift provides a pathway for the lava which inevitably ends up in the same area as before, following very similar flow paths towards the lowest points. The people are living on a ticking time bomb.
They do so not by choice, but because they too have to live. These are not refugee camps, put wherever others can’t see their plight. They need houses to live and places to work. They are people just like us.
And now Nyiragongo has erupted again. The precursors were too weak to be seen on the few available instruments and by the few scientists. Do not expect magma signals: the lava is already in situ and all it needs is to break through the crater wall. The precursor was probably a faint tectonic signal. People knew that a new eruption was imminent. The lava lake had grown too big. But they did not know when. Suddenly the sky went red and the lava came. Eruptions of Nyiragongo only last a few hours. By the time the alarm was raised and the evacuations had started, the lava was already among the people. We do not know yet how many casualties there were. So far, 500 houses are reported buried, 23 people died from the eruption and 9 from a big traffic accident in the evacuation. The real toll will be higher and may never be known.
Proposals have been made to safeguard Goma. Dams could be build to divert the lava around the most endangered area in east Goma. The problem is that this would divert the lava towards other areas that are just as densely populated. Warning systems have been proposed, already 25 years ago. But hardware on the ground does not last long, often due to theft. Before the 2002 eruption, only 2 of 5 installed seismographs were still working. The rangers who lead the expeditions to the lava lake do report anything unusual – but it needs people to collate and analyse the reports. The 2002 eruption has been predicted in advance by the local scientists, based on reports from the rangers and the little instruments they had. There was failure of action on these warning by the authorities – who in any case do not carry much authority here. The 2021 eruption was expected but was not predicted.
What can be done? Monitoring from space is difficult because of the near-perpetual cloud cover over the peaks. Seismographs could be placed in Rwanda, not many kilometers away. But that is a different country and they cannot issue warnings for Goma. With such a volcano, an eruption warning needs to be instantaneous (in 2002 lava reached Goma within 20 minutes, and because the rifts extends into the city the eruption could in theory begin within the city limits), and should be unmissable. It can be done – even here, most people have mobile phones. Pre-eruption warnings need to be accurate, otherwise they will quickly be ignored. That requires significant scientific effort. And finally, we need predictions where lava is most likely to flow, and discourage people from living there. East Goma in particular should not be there. That may be the hardest task of all.
Goma was lucky this time. But there will be a next time. And one after that. And one day this could become the Pompeii of the tropics.
Albert, May 2021
Addendum 27 May 2021
A new INSAR image has appeared show ing spectacular changes in the rift valley. The image shows the changes between 19 and 25 May (i.e. including the eruption but not the subsequent tectonic activity). It is reproduced here (at reduced resolution), with the original taken from https://pbs.twimg.com/media/E2U9GXEWEAU55XF. The INSAR data is also available from https://sarviews-hazards.alaska.edu/Event/221?pinned=d989972a-8f8e-4337-9744-f9dbb668ef33,406295fb-403e-4604-acc9-4cd65ddeae09
This looks very much like a dike intrusion but there are two aspects worth noticing. First, the coloured contours are best visible along the side of the rift valley. Second, the colour sequence on the left and the right of the valley are reversed (look for the order yellow-green). It is always hard to separate up/down from left/right in these plots (the satellite measures the distance from the satellite to the ground, and if it is not directly overhead it will see both). The reversal suggests that the sides of the rift valley are moving away (or towards) from each other, i.e. rifting. If they were moving up (or down) you would see the same pattern on both sides.
So there is rifting. But there is not much indication for a large dike. That would show up as inflation in the centre and that is not really seen. (Unless it is localized in the feature near the lake coast which coincides with a ridge in the landscape).
We will wait for more data. I think there is rifting between Shaheru and Goma. Whether there is lava inside the rift remains to be seen.