Four years ago I wrote this article on the dangers of Lake Kivu’s gas resource. Its eruptive potential is huge, enabled by a stored gas volume of about 80% of its 500 cubic km water volume. Its renewable gas inventory is growing at perhaps 0.5% a year, arguably, so it can erupt with a very large, sudden energy injection. But will that eruption happen now with the 22 May eruption and the follow-on seismic activity? How serious was this event and what does it say about the risk level?
Like any of the experts who worked on a World Bank-sponsored program to figure out how to safely extract gas from Lake Kivu, we asked the questions and gathered the expertise to get a consensus in answers. Most of the six experts who contributed to the document “Management Prescriptions for Lake Kivu Development” spent many thousands of hours of their own time working that out. Not all agreed on everything.
In a Q&A format below, I try to clarify in a summary below what we know and what that means for people’s safety.
The longer-term cause of limnic eruption is for the renewable gas (CH4, CO2) production to build up enough gas content, within any of the deepest 5 layers of the lake, to reach 100% gas saturation. At 100% saturation, gas will bubble out of solution and start to rise to the lake’s surface. At this stage, it can become something akin to a chain reaction. The rising column continues drawing in more gas-saturated water from the deep until it becomes a rising column with explosively degassing water and entrained gas.
But there are at least two ways of pre-empting reaching 100% saturation that can cause an earlier eruption. These all require a massive heat energy injection. Think of a megaton thermo-nuclear device going off deep in the lake. Three possible causes are (1) a massive inflow of lava from one of the two volcanos on the north shore; (2) A large new volcano opening up in the floor of the lake; (3) A massively long and wide seismic rift opening on the lake floor, exposing hot magma.
There is evidence of at least two lake-floor volcanos forming, as shown in the bathymetric map below by Data Environnement. The two live volcanos today, Mt Nyiragongo and Mt Nyamulagira, are approximately 10,000 years old. These mountains are much larger and probably pre-date the two small cones formed and visible on the lake floor. But the deep water probably froze their growth potential.
Studies have looked at the core, drilled into the lake bottom, for evidence of previous lake eruptions. The two small peaks on the lake floor may have come close to causing an eruption but were not big enough to trigger the “big one”. But Lake Kivu sits right on one of the planet’s tectonic hot spots.
Christoph Hormann’s brilliant 3D rendition of the region (below-original text) shows how a string of volcanos were all sited at one time over the rifting hot spot. They have discharged trillions of tons of lava, both fast-flowing lava and flying ash and rocks up to 50 km from more explosive events. The lava plains around the older volcanic peaks, some hundreds of metres deep, show up as smooth slopes in contrast to the high ridges from rifting action. These ridges show clearly in the 3D rendition, hundreds of kilometres on either side of Lake Kivu, created by millions of years of rifting movement. Africa is slowly and occasionally violently splitting into two. This is one of the most active zones down the thousands of kilometres of the north-south rift, pushing East Africa out eastwards to the Indian Ocean.
From the surface of the lake, if you were unlucky enough to be caught right there close to the epicentre, the lake waters would look like they were boiling. The ferocity of this boiling would rapidly escalate until it looked like a watery volcano if it was visible through a fog. A strong odour of H2S, like rotten eggs, would come on strongly until it kills your sense of smell within seconds. That does not signify the H2S gas has gone, but it has become dangerously strong.
The rising waters would rush upwards and outwards rapidly, causing tsunami-like waves. The core of the eruption could grow larger, eventually wider than 1km in diameter. The rapidly forming, dense gas cloud would appear like a foggy hurricane, blasting out in all directions. Nobody could survive for even minutes at this location. We estimate this eruptive storm blasts out for as much as one day.
From the lake shore over 25km away from the site of an eruption’s epicentre, the eruption could look like a rapidly forming cumulo-nimbus storm cloud. But instead of growing upwards, it grows out laterally and fast. Under it, the tsunami-like waves would radiate outwards, rapidly to all shores. These could be metres high, perhaps 5-10m high, crashing onto the shore.
The fog-like windrush would follow, enveloping the shore and getting deeper and penetrating much further ashore on low-lying ground with time, as the eruption continued. This cloud could get as deep as 100-200m above the lake level, with depth depending on whether it was a still or windy day and which way it blows.
There are more frequent occurrences of this lesser phenomenon. There have been reports over the years of swimmers and boats, hundreds of metres offshore, disappearing into such clouds of gas bubbles. These events are most likely caused by the sand ouflow fans from the river flowing into the lake east of the Serena hotel in Gisenyi.
The sand fan builds up over time, hundreds of metres offshore, also forming the beach along the shoreline. But there is a steep drop-off on the fringes of this fan, into deeper water perhaps 200-300m depth. With seismic activity, as has been frequent since the May 22 flank eruption of Mt Nyiragongo, sand slips can be triggered on these steep slopes as they are unstable. With a large volume of sand slipping downslope like an avalanche, deep water is displaced upwards to the surface.
If this water was between 200-260m deep, it would de-pressurise and bubble-out on its way towards the lake surface. The mini-eruption can last several minutes. The water bubbles as it exceeds 100% gas saturation in shallower depths, but it remains cool. The density of bubbling water is lower than water, causing any swimmers and overloaded boats in the area to sink rapidly.
For the big eruption, do not pause to witness this event unfolding. Get up to high ground quickly, climb hundreds of metres above the lake and stay above the visible cloud. Carry water and food with you as the event can last for days and rescuers will be slow to help. As with escaping a tsunami, getting to high ground (above any fog or cloud from the eruption) should save your life. Stay there or go even higher if possible. Do not turn back into the cloud to try help those left behind. That would be fatal.
In calm weather this toxic gas cloud could take days to weeks to disperse, even after the foggy appearance disappears. The toxic cloud remains dangerous for days at least and could disperse to higher ground. So re-entry to low ground needs to be avoided by everyone, at least until gas tests show that the toxicty level has dispersed. Even professional rescue teams should avoid re-entry until H2S levels have diminished drastically to less than 2 ppm. Gas masks or breathing apparatus would be essential, but cannot last long enough to effect any rescue. If anybody survived the toxic gas, they would most likely to be found only near the outer edges of the gas cloud. The only possible way to survive within the cloud would be to shelter in perfectly sealed rooms with an oxygen supply that can last for weeks.
Rescue teams should concentrate on the people who escaped to high ground and who will need water, food and some shelter. Only days or weeks later will rescuers be able to return to the lake shore. A vast recovery operation will be required to clear huge numbers of bodies, whether human, animal or even fish. One cannot expect any survivors among those who did not escape to high ground. Over the days and weeks the danger will become the health hazard from the potentially countless dead.
As for the smaller eruption described above, avoid swimming or boating in the area between Gisenyi and Cap Rubone, especially in the area within hundreds of metres of the Gisenyi beach.
We are describing what a very large potential limnic eruption may look like. The description above is not based on empirical evidence as no such major eruption has occurred in the past few millennia. Perhaps nobody who ever witnessed it would have lived to tell the tale. The last one occurred as the result of a prolonged (centuries-long) drought, about 3500 years ago. The drop in the lake level was enough to reduce hydraulic pressure that serves to keep the gas in solution at depth.
The next predictable lake eruption can occur either from: (a) a further half century or so of gas build-up from now to reach 100% saturation, as discussed above, or (b) pre-emptively triggered by any of the above-described volcanic/seismic events.
The duration of any eruption depends on two factors; the intensity of the eruption and the inventory of gas in the lake. Both factors are strongly dependent on how close the lake’s gas content has approached to 100% saturation. The intensity rises exponentially with % saturation. Roughly speaking, duration depends on the continued availability of gas-rich water drawn into the centre of the eruption, but likely will have depleted inside 24 hours. Gas may continue to bubble up for days, but at a fraction of the intensity.
The inventory of gas in the lake is presently 60-65 bcm (billion cubic metres) of methane, or 20% of the total gas content, and about 300 bcm of carbon dioxide. Other gases include about 2000 ppm of hydrogen sulphide and some nitrogen and argon. compared to the 500 bcm of water in the lake. Any eruption would contain similar proportions of these gases.
The most dangerous gases in any eruption are:
Unlike what was suggested in the current BBC reporting, all these gases are mixed. With 80% CO2 in the mix, it is denser than air and not potentially explosive at all. Such fire or explosion fears are unfounded. That would be the case with >60% methane in a gas emission. But if the gas was mostly methane it would be lighter than air and would escape rapidly to the upper atmosphere.
Yes, but there is no instant panacea. We have the means of progressively reducing the risk of an eruption by producing gas from the lake. Extraction needs to be faster than new gas formation but must be done strictly according to the rules to reduce danger. It takes time and long-term, consistent application of the safety rules.
A key concern among experts is that several developers on Lake Kivu have skirted the rules, resorting to legal arguments to avoid being subject to the mandatory and legally recognised Management Prescriptions. The Lake Kivu Monitoring Program (LKMP) has been observing the impact of discharge plumes from all production barges. The rules on the discharge of degassed water and wash water are strict. They are required to protect against disruption of the stability structure of the lake – the best of all defences against an eruption. There is evidence of a concerning level of weakening of the gradient layers as the result of non-compliance, where such disallowed re-injection plumes are breaking through gradients. The damage will soon be irreversible.
A compliant extraction technology, working strictly to the rules, has the ability to reduce the level of risk of a lake eruption by as much as 90% within ten years and by as much as 99% within 20 years. While this de-risking is significant, it takes time to be fully effective and that takes investment to build out the required capacity.
A full-scale de-risking project will take billions of dollars. The upside of that is that the optimal energy derived from degassing the lake can earn 15-20 times as much as the capital spent. So degassing is commercially viable. Among many benefits, it can make the two countries that share the lake net zero in terms of carbon emissions while saving lives too. There is a potentially very good economic result that can come out of de-risking the danger of Lake Kivu, as long as it is done the right way.
Almost certainly not. This is a bold statement but made with sufficient certainty with the facts available. Considering all the above factors the recent eruption of Mt Nyiragongo, through a flank eruption. It released much of the built-up energy in the system by draining the caldera. It will take time, perhaps as much as a decade, to rebuild the lava lake in the main crater after this release. The lava lake is a better safety valve for the magmatic system than a lava lake that has solidified. The last flank spill, a bigger event than this one with lava penetrating Lake Kivu to 80m depth, occurred in January 2002.
Another form of danger is for a large seismic rift to open up, but that occurs through a separate mechanism and is less predictable. Darchambeau, a seismologist who has spent years studying Lake Kivu, monitors the seismic events that show up as a wedge of around a hundred events, running north-south 30km into the lake from close to the Rwanda/DRC land border.
We must acknowledge the work of the Observatoire Volcanologique de Goma in collecting and reporting data. Celestin Kasarkea Mahinda and his small team work diligently, with virtually no regular budget, to maintain monitoring of the volcanos and to share reports on the danger levels based on their observations. Here is a recent report that commented on concerns emanating from journals of rising lava levels in the crater lake. COMMUNIQUE (version française)
Their monitoring equipment is only accessible through climbing the mountain and it is too frequently vandalised. However, their information gathering is much improved on that available in 2002. The observatory’s view before this volcanic eruption was that vigilance was required, not panic. Even now there is a a tendency to panic and a lack of coordinated information sharing and warning systems.
We also recognise the many years of support from volcanologists to the Goma team, including Prof. Dario Tedesco from the University of Campania, Napoli. He has provided the lake experts with valuable input and support.
Our original page follows below:
Lake Kivu contains an important methane gas resource. It is one of the Great Lakes of Africa in the Western (Albertine) Rift Valley. It lies on the border between the Democratic Republic of the Congo and Rwanda. The lake has a length along the dividing border of 98km and at its widest it’s 48km. The lake is some 10 000 years old. This is because it was dammed by the lava fields of two active volcanoes in the north. With its ancient outlet to the Nile River blocked by the lava fields and volcanoes, it now spills to the south. It is one of many reasons it began to create a Gas Resource.
3D Compilation of Lake Kivu in the Africa Great Lakes Region (c Christof Hormann)[/caption]
The Ruzizi River flows south from Lake Kivu through a deep canyon. It barely shows on the Hormann 3D picture above, but like the Grand Canyon in the USA it is evidence of one or more major outflows from Lake Kivu into Lake Tanganyika. It is up to 2km wide and hundreds of metres deep.The river flows south in a cascade which drops over 700m in 20km to Lake Tanganyika. However, the river has a series of hydro-electric run-of-river power sites which are exploited to only 5% of their 560 MW potential. In contrast, Lake Tanganyika, in contrast, has no methane resources as it has no trap mechnism capable of storing any that is generated in its great depth.
Lake Kivu’s methane gas resource can now be measured very quickly, cheaply and accurately. But it must be measured every few years. This is because gas is continuously forming, making it a biogenic and renewable resource. The methodology for updating resource measurement is simple. It is described below.
However, the greatest uncertainty associated with Lake Kivu’s resource is the recoverability of gas in situ. This varies greatly by the gas recovery method used. This in turn is hugely variable. Indeed it is mainly dependent on the technology and plant design deployed.
The Kivu gas resource certainty is in stark contrast to typical gas in-situ measurement for conventional oil and gas. In fact, such data and their interpretation are slow to acquire. They are also expensive and yet most often lack great certainty.
Along with the inaccuracy of predicting the contents of the resource, there is the uncertainty multiple of % recovery. One calculates reserve estimates by multiplying the resources in place by their recovery. This percentage is highly variable, depending on the characteristics of the oil and gas resources present. It is also impacted by the “tightness” of the reservoir rock. Producers can choose or vary their production techniques to increase recovery potential.
For conventional oil and gas, the field discovery and delineation drilling are just the beginning of the process. Typically, discovery requires several methods used in sequence, in a process that may take years. For instance, these will include geological mapping for initial targeting. For deeper interpretation, geophysicists will select the respective methods and tools for onshore and offshore survey.
The survey data produce 2-D and 3-D seismic mapping. These maps are used by experts to delineate a potential reservoir. Experienced technicians identify their drilling targets for reserve delineation and quantification to gather the field data. Essentially, they use drill data logged in the database such as depth, metres of intersect, pressure, flow-testing and assays. More recently, newer geophysical methods use sophisticated down-hole probes to establish more extensive data. New tools such as ML and AI are thus becoming the tools of choice for quicker interpretation of vast “data lakes”. The risk of “dry” wells and uneconomic reservoirs remains high, despite expensive seismic modelling.
Indeed, such oil or gas reservoirs can have estimates of P50 and P90 certainty reserves that are wide apart. The certainty and cost both hinge on the number of holes drilled, their depths, the grid spacing etc. The costs of completing a conventional field’s reserve estimate can add up to hundreds of millions of dollars. The process can easily take ten years to complete.
For Lake Kivu, the reservoir capacity is known quite precisely by depth in a 3-D mapped lake (see maps above & below). The second factor in the calculation, the concentration profile of gases by depth, is measured roughly every five years. The concentration of gases by depth is highly consistent across the lake. Therefore, for any given year, taking a few profile measurements allows for a quick reserve recalculation.
EAWAG has estimated dissolved gas resources in Lake Kivu several times. Reserve estimates were made recently in 2004 and 2010. The prior estimate was back in 1975. In these cases, they calculate the lake’s volume from the lake bathymetry maps (shown below). They assessed detailed gas concentrations from sampling across the lake, at all depths. Multiplying gas concentrations, for each metre of depth by the gas concentration, they can quantify the resource.
These re-estimates can be very accurate, much more so than typical for conventional gas reservoirs. This accuracy is good because gas concentrations at any depth are very consistent across the whole lake. Therefore a few measurements across the lake can confirm the lake-wide dissolved gas concentration profile by depth. 2010 assessments of lake methane resource were 65 km3 (>2 tcf) and carbon dioxide 260 km3 (10 tcf).
Lake Kivu is arguably the world’s largest natural freshwater bio-digester. The lake contains vast amounts of algae, providing feedstock for producing recoverable biogas. The lake may also be one of the world’s largest carbon sinks, after the oceans, due to it unique capability to store vast volumes of these greenhouse gases. Gas resources in-place can provide inexpensive power for the region’s population for generations. But a program to draw-down gas inventories is also essential to avert a massive limnic eruption.
Distinct layers within the lake form five unique, stratified gas containments. In fact, the lake’s salinity-based density gradients (chemoclines) clearly separate and define these strata. Therefore, left undisturbed, these containments retain the continuously generated, biogenic CH4 and CO₂. The gases are trapped in solution, as strata of varying concentrations. The richest biogenic gas inventory is stored in the lower two strata of the lake. Significant volumes of CO₂ of magmatic origin also accumulate within this total gas inventory.
Gases are therefore accumulating ever closer to saturation levels. As such, they threaten to cause a future limnic eruption. For example, an eruption occurred at Lake Nyos in Cameroon in 1986 with 1746 reported casualties. In this case, an eruption is inevitable if not actively prevented. However Lake Kivu has the size, potential, location and gas inventory for even more dire consequences. An eruption may be of the order of 1000 times larger.
In a detailed study, Philip Morkel has co-authored a paper with Dr Finn Hirslund of COWI. The paper outlines a novel hypothesis to interpret an improved understanding of the lake’s history. We used a complex, multidisciplinary approach in this work to analyse the lake’s unique situation. Our hypothesis identifies the impact of diffusion mechanisms on vertical transport of gas and nutrients in the lake. Through this body of work, one can foresee the probable outcomes and recognize the effects of system constraints.
The paper outlines how one should initiate gas extraction projects on a basis that enables three vital outcomes. These are; a) for developers to pursue economic projects, while b) society enforces developers to follow extraction methods that ensure safety, through lifetime prevention of disasters, while c) gas extraction maximizes useful energy output.
The key to safety is proper management of the chemoclines. This must be achieved while drawing down the gas inventory. The instrument to achieve both, safely and productively, is derived from defining and making the right processing choices. The choices correctly specify all critical requirements for the plant design of gas production facilities. Indeed, these are derived from a thorough analysis of what may happen and what must be done technically.
We had to first deal with the primary concern for continued public safety. Thereafter we could examine, through further analysis, ways of minimizing environmental impact on the mixolimnion. Slow, natural upwelling of meteoric water out of surrounding lava strata causes impactful changes to the stratification. These meteoric waters flow into the mixolimnion of one of the world’s most nutrient-rich water bodies.
In addition, we need to consider the impact of washing the extracted raw gas with some of this water. This scrubbing of unwanted gas is essential to upgrade gas quality for use in power-generating engines. The wash water source is from the shallow stratum of the mixolimnion.
Finally, we considered how to continue holding the chemoclines at their current levels. To achieve this need, we discussed how to evacuate a fraction of the degassed water from deeper resource strata. Counter-intuitively, the best solution may be to re-inject degassed water into the mixolimnion, in the top 60m of the lake. Displacing degassed water can maintain chemoclines in place.
However, it introduces an environmental challenge. The challenge lies in defining how to minimize any safety-driven or environmental impacts on the mixolimnion. Potential toxic effects come from H₂S, from CO₂-induced acidity, and from oxygen depletion caused by CH4 and H₂S.
Defining the role and methods of sustainable commercial gas extraction as the tool that for the foreseeable future, serves to avert a disastrous limnic eruption caused by nature’s continued accumulation of gases in Lake Kivu.
EAWAG, at Kastanienbaum on Lake Lucerne in Switzerland, is a faculty specialising in studies of lakes and their behaviours. EAWAG Studies of Lake Kivu demonstrate how it is one of the most-studied lakes on the planet because of its unique characteristics.
The Lake Kivu methane is partly magmatic in origin. It is noteworthy that microbial reduction of the volcanic CO2 produces some of the methane. The role of archaea methanogens is being studied. But the gas inventory is mostly biogenic. This gas forms through the anaerobic bio-digestion of dead algae and fish on the lake floor. The gas resource is renewable, with a rate of renewal at present about 0.5-1.0% per annum. The rate of renewal has increased over recent decades, possibly due to changes in nutrients present in the Biozone.
Kivu is one of three known turn-over lakes. But Kivu shares this distinction with Cameroonian volcanic lakes, Lake Nyos and Lake Monoun. With turnovers recurring less often than 1000-year intervals, history fails to provide a detailed record any one incident.
Analysis of Lake Kivu’s geo-morphology history indicates multiple major disturbances from the examination of drill-cores. The cores show evidence of one or more massive biological extinctions on millennial timescales. These extinction events relate to gaseous eruptions that occur every few millennia. These eruption events can release hundreds of billions of cubic metres of gas in short catastrophic events.
But the gaseous chemical composition of releases from exploding lakes can be unique to each lake. Lake Kivu has a signature gas content of 20% methane and 80% carbon dioxide. It has some nitrogen and sulphur dioxide (2000 ppm) as lesser constituents.
Potential eruptions release emissions of gases that pose a serious risk to all oxygen-dependent life, anywhere in the vicinity of the lake. But the inventory of methane resource gas in situ also provides an accessible, exploitable, renewable energy resource. If we properly exploit these, the value of production from the methane resource can provide tens of billions of dollars for the Kivu region.
The lake covers a total surface area of some 2,400 km2 (1,040 sq mi). Its surface elevation lies at 1,463 metres (4,790 ft) above sea level. But the lake also lies in a rift valley that is slowly being pulled apart. This rifting results in volcanic activity in the area. Rifting also makes the lake particularly deep. Its maximum depth of 486 m (1,575 ft) ranks fifteenth in the world.
EAWAG studies assess Lake Kivu to contain approximately 65 billion cubic metres dissolved methane gas. A density gradient at 260 m (850 ft) traps most of the gas below it. This density gradient blocks any upward diffusion of gases and nutrients into shallower water.
Until 2004, extraction of the gas was done on a small scale by a 1965 era extraction plant. The piped extracted gas was being used to run boilers at the Bralirwa (Heineken-owned) brewery in Gisenyi, Rwanda, until 2003.
Lake Kivu has an extraordinary feature in the stratification and resultant stability of the deep waters of the lake. The vertical density profile is a key feature defining the lake’s behaviour. However, this phenomenon is not evident in any other lakes to this degree. No other is known to have five strata.
Oxygenated water persists only the uppermost layer, the Biozone. This Biozone supports aquatic life in the form of small, sardine-like fish called “Sambaza”. Indeed, seasonal turnovers oxygenate the Biozone. Hence this occurs more strongly in the dry, windy seasons. Oxygenation penetrates to a depth of up to 60m. However, oxygen-rich water only penetrates to as shallow as 40m during the calmer, wet seasons. These wet seasons run from January to April and from October to November due to the lake’s equatorial latitude.
The deeper layers are progressively more dense with depth. Increasing concentrations of salts dictate the density. These consist mainly of magnesium carbonates and sulphates. For that reason, density layers separate the lake into definable strata. Figure 5 show these density gradients, illustrating the defined zones. Gradient layers act as virtual traps, preventing significant vertical migration of gases out of the main Gas Resource.
Gas formation happens on the lake bottom. The bulk of the gas resource accumulates in the Resource Zones, as shown above. Gas reserves remain in the same lake strata as as they are generated. This is because the strength of the gradient layers prevents gas diffusion through a sharp change in density.
This is a sectional view of Lake Kivu, looking to the north from the deeper northern basin. We show this view schematically. The vertical transitions characterise the lake as it is at present. The lake bottom composition consists of many layers of sediments. We know that some of the sediment layers date to the older and shallower lake’s existence, prior to 10,000 years ago. The belief is that the anoxic bio-degradation of sediments occurs mainly on top of the sediments.