Some research results:


Laboratory simulations of volcanic and lake eruptions

CO2-driven lake eruptions

1986 Lake Nyos and 1984 Lake Monoun eruptions

Dome collapse, magma fragmentation and pyroclastic flow

Age and Accretion of the Earth






Support by the National Science Foundation

The research on this web page is based on work partially supported by the National Science Foundation under Grants 9118155, 9304161, 9315918, 9458368, 9526980, 9706107, 9725566, 9815351, 9911352, 9972937, 0106718, 0125506, and 0228752.  Any opinions, findings, and conclusions are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.




  CO 2-driven lake eruptions: Disasters of Lakes Nyos and Monoun


     Gas-driven eruptions are powerful, destructive, and not uncommon.  The more familiar type is the violent volcanic eruptions powered by the exsolution of H 2O gas initially dissolved in magma.   Several famous eruptions include the 1991 eruption of Pinatubo in the Philippines that killed 300 people and devastated the Clark Air Force Base, the 1980 eruption of Mt. St. Helens in the USA that killed 57 people and reduced the height of Mount St. Helens by 400 meters, the 79 AD eruption of Vesuvius that buried the Roman City of Pompeii, and the ~1600 BC eruption of Santorini that destroyed the Minoan Civilization.  

     A new type of eruption has been recognized in the 1980’s.  On the evening of August 21, 1986, a massive CO2 -rich cloud, released from Lake Nyos in Cameroon, swept down the valleys and brought devastation to a densely populated area (Kling et al., 1987).   The death toll was estimated to be more than 1700.   Cattle and other animals (including birds), some of them at hills 120 meters above the lake surface, were also killed but vegetation was largely unaffected.  A similar though smaller event occurred at Lake Monoun (also in Cameroon) on August 15, 1984, killing about 40 people.  Witness accounts included the following: some heard loud rumbling or bubbling sound from the lake, some saw white mist/cloud rising from the surging lake , some smelled peculiar odor, many lost conscience and some fortunately awoke later.  

     In the aftermath of the Lake Nyos disaster, many countries not only offered humanitarian assistance, but also sent teams of scientists to investigate the cause of the disaster.   The Cameroon Ministry of Higher Education and Scientific Research organized a conference in Yaoundé (the capital of Cameroon) in March 1987 on the cause of the Lake Nyos disaster.   Two hypotheses emerged from the conference about the origin of the massive CO2 gas releases.   One can be referred to as the volcanic hypothesis.   The other can be referred to as the limnic hypothesis.   As evidence, scientists advancing the volcanic hypothesis cited the violence and localized nature of the process plus some other anecdotal evidence regarding smell of the gas, color of the lake, etc.   However, survey of the lake and surroundings failed to discover fresh volcanic rock, volcanic vent or disturbed sediments.   In the limnic hypothesis, the process was often referred to as a lake overturn.  In retrospect, the choice of the term “lake overturn” probably hindered the acceptance of the limnic eruption hypothesis because many people have the notion that a lake overturn is nonviolent.  In the Yaoundé Conference, a group of French scientist led by J.C. Sabroux coined the term “limnic eruption” to describe the internal lake process that released massive amount of CO 2 gas.  The limnic hypothesis was based on both the high concentrations of CO 2 in deep lake water and the absence of clear evidence for a volcanic eruption.  The limnic hypothesis gradually evolved to provide a clear picture for the process.   In the context of the hypothesis as it now stands, the gas release was due to the degassing of lake water that was nearly saturated with CO2 prior to the eruption although there is still debate about the triggering mechanism and the depth at which the eruption initiated. 

     Considerable progress has been made after the Yaoundé Conference in both monitoring lake-water composition (especially the gas concentrations) and theoretical understanding of CO 2-driven lake eruptions.  It is now generally understood that the CO 2 gas release represents CO 2 -driven water eruptions due to the rapid exsolution of CO 2 gas bubbles from a supersaturated CO 2 aqueous solution.  Plans to pump out and degas lake bottom water are being carried out.  

Solubility of CO2 in water

     The solubility of CO 2 in water is well known and listed in standard chemistry handbooks.   It increases with increasing pressure and decreases with increasing temperature.   At 25°C, the solubility of CO2 in water is 0.15 wt% at one atmosphere pressure, 1.0 wt% at 7 atmospheres and 3.0 wt% at 21 atmospheres.   The greatest depth of Lake Nyos is 208 meters, corresponding to a pressure of about 21 atmospheres.   At saturation (that is, when the partial pressure of CO2 is the same as the hydrostatic pressure), Lake Nyos bottom water can dissolve about 3 wt% CO2.   If the CO2 content dissolved in water is less than the solubility at the given pressure (depth), no bubbles form and more CO2 can be dissolved into water.  If CO 2 content dissolved in water is more than the solubility at the given pressure, CO2 exsolves from the water to form a gas phase.   That is, CO2 bubbles nucleate and grow from the initially homogeneous solution.  

Mechanisms of lake eruptions

     Measurements show that the bottom water of both lakes contains high concentration of CO 2.  When CO 2 becomes oversaturated (possible triggers are discussed below), gas bubbles nucleate and grow.  The volume of the bubbly system increases rapidly.   The bubbly water is less dense than the surrounding water and rises.  As it rises, the surrounding pressure becomes less, and more bubbles grow.   The density decreases further and the bubbly water rises with increasing velocity.  Therefore, there is strong positive feedback.  As the vesicularity (volume fraction of bubbles) of the bubbly water increases to a threshold, water film on bubble walls ruptures and the flow fragments into a gas flow carrying water droplets.   As the flow exits the surface, larger droplets rain down near the eruption vent and finer droplets are carried further away, similar to a mist.   The exact location of the eruption conduit (or conduits) depends on the nature of the trigger.   Once an eruption conduit forms, it can sustain itself by drawing more saturated water into it through suction.  

Dissolved CO2 in lake bottom water

     A research team led by G.W. Kling monitored the change of the water composition of Lakes Nyos and Monoun since the 1986 eruption (Kling et al., 1994).  Both lakes contain high concentrations of CO 2.  The dissolved CO2 concentration in water increases with depth and is increasing with time.   The increase is inferred to be due to leakage of CO2 into the lake from below, probably a magmatic source along the Cameroon Volcanic Line.   As of March 1992, CO2 content in the bottom water corresponds to a CO 2 partial pressure of 12 bars at Lake Nyos (hydrostatic pressure is 21 bars) and a pressure of 8 bars at Lake Monoun (hydrostatic pressure is 11 bars).  By January 2001, Kling’s measurements show that the partial pressure of CO 2 increased to almost 16 bars (Fig. 1 ).  At the present rate of CO2 concentration increase, saturation of CO2 at lake bottom could be reached in about 10 years for Lake Nyos.   Water with dissolved CO2 is denser than pure water.   Therefore, before saturation, the lake is hydrostatically stable and is stratified.   However, as the dissolved CO 2 content in water increases, the partial pressure of CO 2 in the water increases and the hydrostatically stable water at the lake bottom becomes increasingly unstable with respect to perturbation.   A small perturbation due to a landslide, sinking of cold rain water, an internal wave, a small volcanic injection, or a heavy flood of water into the lake might move the water up sufficiently to reach oversaturation (partial pressure of CO2 is greater than the hydrostatic pressure).   Even though the exact triggering mechanism for the eruption of Lake Nyos is still unknown, the eruption was inevitable as long as the bottom water (or any layer of water) was near the saturation.  


Fig 1


Laboratory simulations of CO2 -driven water eruptions

     Starting from 1991, my coworkers and I have carried out experimental simulations of CO 2-driven water eruptions in the laboratory in the so-called Champagne-type experiments (Mader et al., 1994; Zhang et al., 1997; Zhang, 1998).   The experimental apparatus consists of a test cell that exhausts upward into a tank that can be evacuated.   The tank and the test cell are separated by an aluminum foil diaphragm.  During an experiment, the test cell is partially filled with CO 2-saturated H2 O (~1.0 wt% CO 2 in water).   Rapidly accelerating flow is initiated by cutting the diaphragm with a pneumatically driven knife blade inside the tank.   The process is recorded by high speed motion picture photography at up to 4000 frames per second.  The films are then examined frame by frame using an analysis movie projector.   The basic questions to be addressed are whether limnic eruptions can be violent, and the conditions for and the dynamics of a violent gas-driven eruption (including volcanic and limnic eruptions).   The experiments were recorded by high speed camera at 4000 frames per second.  Some pictures are transferred to video and can be seen at .  Experimental data clearly show that CO2-driven water eruptions can be violent under the appropriate conditions, providing direct evidence for the limnic hypothesis.

Comparison of limnic and explosive volcanic eruptions

     The CO2 -driven water eruptions are similar to H2O-driven violent volcanic eruptions in that both are driven by the exsolution of a gas component from an initial solution that is roughly saturated with the gas component.  Bubble growth plays a critical role in both cases.  The role of buoyancy, however, is different.   For a CO2-driven water eruption through a lake, the flow rises due to buoyancy because the surrounding water can deform and flow into the lower pressure region.   A CO2-driven water eruption ends when sufficiently saturated bottom water is used up.   Water sucked into the conduit but without sufficient degree of saturation in CO 2 will not rise much because there is no expansion and buoyancy induced by bubble growth.   For a H 2O-driven volcanic eruption, if the wall rocks surrounding the conduit is cold and rigid, the rocks cannot deform on the time scale of an eruption, and the eruption is not due to buoyancy but due to volume expansion.   When the conduit or magma chamber pressure becomes sufficiently low compared to lithostatic pressures, wall rocks can cave in and therefore terminate the eruption.  

Dynamics of lake eruptions

     With understanding of the physical process gained from experimental studies and from comparison with gas-driven volcanic eruptions, the dynamics of CO2-driven lake eruptions are semi-quantitatively modeled using the Bernoulli equation and an equation of state for CO 2-water assuming equilibrium between CO2 bubbles and water (Zhang, 1996).  It was found that the kinetic energy per unit mass of the erupting flow is roughly proportional to the solubility coefficient of the gas and the initial saturation pressure.   For Lake Nyos with a depth 208 meters, if bottom water was erupted, the exit velocity can reach 90 meters per second (200 miles per hour).   At this exit velocity, the erupting cloud can rise to a height of 400 meters if the entrainment of air is ignored, consistent with observed dead cattle at 120 meters above lake surface and dead birds.   A schematic drawing of the eruption conduit and column is shown in Figure 2.


Fig 2


     If the initial saturation depth is 96 meters (for Lake Monoun, or for an eruption initiated at mid-depth in Lake Nyos), the maximum exit velocity would be 50 meters per second (110 miles per hour).   The erupting column would rise to a height of 130 meters.   Calculated erupting flow velocity at a given depth of the flow is shown in Figure 3a.   Calculated exit velocity as the bubbly flow exits lake surface is shown in Figure 3b.  

Fig 3a

Fig 3b

     Because CO 2-rich gas is denser than air, the gas cloud eventually collapses down and becomes a ground-hugging density flow.   Such a flow is dynamically similar to a pyroclastic flow.   However, since the flow is not hot (pyro) and does not contain clastic particles, Zhang (1996) coined the term ambioructic flow (“ambioructic” is formed by combining “ambient” and “eruct”) to describe the CO2 flow carrying water droplets at ambient temperatures.   The density flow is the killing agent for people and animals on its path through asphyxiation.

Degassing Lakes Nyos and Monoun

     Understanding the process and realizing the danger that Lake Nyos and Lake Monoun still pose, scientists are removing dissolved CO2 from Lake Nyos.   Please see the following French web site: .

Is there danger for the lake near your home to erupt?

     You may wonder whether the lake next to your home may erupt tomorrow.  Five conditions must be met for a violent lake eruption to occur.  One is that the lake must be deep.   A ten-meter deep lake would not be able to erupt violently.   The deeper the lake, the more potential there is for a violent eruption.  The second condition is a continuous source of gases supplied to the bottom of the lake so that bottom water gradually becomes supersaturated.   The gas supply often comes from escaped volcanic gas from a deep-seated magma chamber.  The third is that the gas supplied into the bottom water must have a high solubility.   The erupting power is proportional to the solubility coefficient (solubility at a specific pressure).   The greater the solubility coefficient, the more violent the eruption can be.   For example, the solubility coefficient of CO2 in water is 45 times that of air and 26 times that of CH 4 (methane).  Therefore, energy obtained from the exsolution of air from water or methane from water will be less than that obtained from the exsolution of CO 2 from water by the appropriate factor.   The fourth is that the gas must accumulate at the lake bottom.   That is, the surface and bottom water of the lake must not mix every year.  Due to seasonal temperature variations, most lakes mix every year due to the sinking of surface water as the surface temperature becomes 4 °C (where pure water reaches its highest density).  Therefore, small annual temperature variation is a necessary condition for gas-driven lake eruptions.  The fifth is that the dissolution of the gas must increase the density of water.   Otherwise, the dissolution of the gas would lead to buoyant rise of the water and hence the loss of the gas from the lake.   These conditions mean that gas-driven lake eruptions are most likely to occur in deep equatorial lakes near volcanic zones where CO 2 leaks from deep magma chambers into rocks and lakes.   Bottom water from Crater Lake and Lake Tahoe have been collected some years ago and no high concentrations of CO 2 found. 


Youxue Zhang

Department of Geological Sciences

The University of Michigan

Ann Arbor, MI 48109-1063



Further reading:

Kling G. W., Clark M. A., Compton H. R., Devine J. D., Evans W. C., Humphrey A. M., Koenigsberg E. J., Lockwood J. P., Tuttle M. L., and Wagner G. N. (1987) The 1986 Lake Nyos gas disaster in Cameroon, West Africa. Science 236, 169-175.

Kling G. W., Evans W. C., Tuttle M. L., and Tanyileke G. (1994) Degassing of Lake Nyos. Nature 368 , 405-406.

Mader H.M., Zhang Y., Phillips J.C., Sparks R.S.J., Sturtevant B., and Stolper E.M. (1994) Experimental simulations of explosive degassing of magma, Nature 372 , 85-88.

Zhang Y. (1996) Dynamics of CO2-driven lake eruptions. Nature 379 , 57-59.

Zhang Y., Sturtevant B., and Stolper E. M. (1997) Dynamics of gas-driven eruptions: Experimental simulations using CO 2-H 2O-polymer system. J. Geophys. Res. 102 , 3077-3096.

Zhang Y. (1998) Experimental simulations of gas-driven eruptions: kinetics of bubble growth and effect of geometry. Bull. Volcanol. 59, 281-290.