Content uploaded by Minoru Kusakabe
Author content
All content in this area was uploaded by Minoru Kusakabe on Aug 09, 2019
Content may be subject to copyright.
GEochemistry Monograph Series, Vol. 1, No. 1, pp. 1–50 (2017) www.terrapub.co.jp/onlinemonographs/gems/
Received on December 5, 2015
Accepted on May 11, 2016
Online published on
April 7, 2017
Keywords
•Cameroon
•Lakes Nyos and Monoun
•gas disaster
•crater lakes
•magmatic CO2
•limnic eruption
•disaster mitigation
•degassing
•Cameroon Volcanic Line
•SATREPS
© 2017 TERRAPUB, Tokyo. All rights reserved.
doi:10.5047/gems.2017.00101.0001 ISSN: 2432-8804
degassing of CO2 is the quiet discharge of gas often
derived from a magmatic source, with varying degrees
of contamination by crustal or biological CO2. Crater
lakes usually sit on top of volcanic conduits and act as
condensers or traps for magmatic volatiles. The Lake
Nyos gas disaster in 1986, and a similar event in 1984
at Lake Monoun, both in Cameroon, Central Africa,
resulted from an excessive accumulation of magmatic
CO2 in the bottom layers of the lakes. These volcanic
Abstract
This is a review paper on the Lakes Nyos and Monoun gas disasters that took place in the
mid-1980s in Cameroon, and on their related geochemistry. The paper describes: (i) the
gas disasters (the event and testimonies); (ii) the unusual geochemical characters of the
lakes, i.e., strong stratification with high concentrations of dissolved CO2; (iii) the evolu-
tion of the CO2 content in the lakes during pre- and syn-degassing; (iv) the noble gas
signatures and their implications; (v) a review of models of a limnic eruption; (vi) a
revision of a spontaneous eruption hypothesis that explains the cyclic nature of a limnic
eruption (Kusakabe 2015); (vii) a brief review of the origin of the Cameroon Volcanic
Line (CVL) and the geochemistry of CVL magmas; (viii) a brief review of other CO2-
rich lakes in the world; and (ix) concluding remarks.
Degassing of the two lakes has been successful. Most of the dissolved CO2 has been
removed from Lake Monoun, resulting in the stoppage of the degassing system. How-
ever, the CO2 content in the lake started to increase in recent years due to the continuing
supply of gas from the underlying magma, indicating the necessity of the continuous
removal of gas from the lake. Lake Nyos will attain the same situation in several years
when degassing will stop. Thus, a continuation of scientific monitoring of the lakes is
essential. Since the transfer to Cameroonian scientists of monitoring techniques, includ-
ing analytical equipment necessary for the monitoring, has been effected through the
SATREPS project (Japan’s Official Development Aid), the responsibility is now theirs,
and it is strongly hoped that the lake monitoring, the rehabilitation of displaced people,
and the setting up of an infrastructure for them, etc., will be carried out by the Cameroonian
Government and local scientists.
Lakes Nyos and Monoun Gas Disasters
(Cameroon)—Limnic Eruptions Caused by
Excessive Accumulation of Magmatic CO2
in Crater Lakes
Minoru Kusakabe
Department of Environmental Biology and Chemistry
University of Toyama
3190 Gofuku, Toyama 930-8555, Japan
e-mail: mhk2314@gmail.com
Citation: Kusakabe, M. (2017) Lakes Nyos and Monoun gas disasters (Cameroon)—Limnic erup-
tions caused by excessive accumulation of magmatic CO2 in crater lakes. GEochem. Monogr. Ser.
1, 1–50, doi:10.5047/gems.2017.00101.0001.
1. Introduction
Volatiles in the deep interior of the Earth are brought
to the surface mainly by volcanic activity. In terms of
the present-day global carbon cycle, the CO2 discharge
from subaerial volcanism including the passive dis-
charge from the craters or flanks of volcanoes, is the
major non-anthropogenic contributor to atmospheric
CO2 (e.g., Kerrick, 2001; Gerlach, 2011). The passive
2M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
crater lakes are considered to be the sites of passive
degassing of CO2. On 26th August, 1986, a large
amount of CO2 was suddenly released from Lake Nyos
that asphyxiated 1746 people, and an unaccountable
number of cattle, living near the lake (Sigvaldason,
1989). A very similar gas event took place in August
1984 at Lake Monoun, with 37 casualties (Sigurdsson
et al., 1987). Lake Monoun is located only 100 km
south-east of Lake Nyos (Fig. 1). A term “limnic erup-
tion” was coined by J.-C. Sabroux to describe a gas
outburst from a lake (Halbwachs et al., 2004), and will
be used in this review. Given that this type of gas dis-
aster had not been previously recorded (Sigurdsson,
1987a), the Lakes Monoun and Nyos events attracted
a great deal of attention, not only from the media but
also from a disaster science perspective. At that time,
nobody imagined that the lakes had accumulated so
much lethal gas and that the gas was released into the
atmosphere without any precursor. Subsequent
geochemical investigations revealed that the gas was
CO2 that originated from magma and had accumulated
passively in the deep part of these lakes. The physico-
chemical characteristics of the lakes are unique and
have evolved with time, even after the gas release, due
to the continuing supply of magmatic CO2.
In the present paper, issues related to these gas dis-
charges are reviewed in the following sections; (III)
what happened at the time of the Lakes Nyos and
Monoun gas disasters?; (IV) pre- and syn-degassing
chemical evolution of the lakes; (V) possible causes
of the disasters, the models and the repetitive nature
of a limnic eruption. In relation to the recurrence pre-
vention of a limnic eruption, a bilateral scientific
project between Japan and Cameroon called SATREPS-
NyMo was carried out during 2011 and 2016, and is
outlined in Section 5.
The upper 40 m of Lake Nyos is bounded on the north
by a narrow dam of poorly consolidated pyroclastic
Fig. 1. Location of Lakes Nyos and Monoun (red circles) and volcanoes along the Cameroon Volcanic Line (solid black) in
Cameroon, Central Africa. Modified from figure 1 of Environmental Monitoring and Assessment Journal, Hydrogeochemistry
of surface- and groundwater in the vicinity of Lake Monoun, West Cameroon: Approach from multivariate statistical analysis
and stable isotopic characterization, 2015, Kamtchueng, B. T., Fantong, W. Y., Takounjou, A. F., Tiodjio, E. R., Kusakabe, M.,
Mvondo, J. O., Zhang, J., Ohba, T., Tanyileke, G., Hell, J. V. and Ueda, A. „ Springer International Publishing Switzerland
2015 with permission of Springer.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 3
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Fig. 2. (a) Victims near Lake (Stager and Suau, 1987). Reproduced with permission of Helimission (www.helimission.org).
(b) Dead cow by the lake (photo taken by the author).
rocks. This dam is being affected by back erosion. A
warning was given that the collapse of the dam could
cause a flood that would affect inhabited areas over a
220 km distance (Lockwood et al., 1988). An accurate
estimation of the rate of back erosion of the dam is
critical for the safety of people living downstream.
Thus, the age of the dam formation (or Nyos maar for-
mation) has been hotly debated using different age
determination techniques. Recent progress on the age
of the dam is briefly reviewed in Section 6.
Thirty nine crater lakes including Lakes Nyos and
Monoun and numerous soda springs are located along
the Cameroon Volcanic Line (CVL). An understand-
ing of the origin and the geochemistry of CVL mag-
mas is essential. These subjects are reviewed in Sec-
tion 6, which constitutes the basis on which CO2 accu-
mulation in these lakes is scientifically interpreted. We
also need to understand why CO2 becomes enriched in
magmatic volatiles as they leave the magma. The Lakes
Nyos and Monoun events have stimulated geochemical
interest in other CO2-rich volcanic lakes in the world
for their gas hazard potential. This is reviewed in Sec-
tion 7.
2. Gas disasters at lakes Nyos and Monoun,
Cameroon
2-1. Cameroon: Location and physiography
Cameroon is a country in Central Africa located be-
tween 2–13∞N latitude, and 8–16∞E longitude (Fig. 1).
It is bounded by 6 countries: Chad to the northeast,
Nigeria to the west, Central African Republic to the
east, Equatorial Guinea, Gabon and Congo to the south.
Cameroon can be divided into ten major ecological
regions. These regions are classified under four re-
gional units which are differentiated by their geogra-
phy, climate and vegetation characteristics as follows:
(1) The Sudano-Sahelian zone in the North is composed
of the Mandara mountains, Diamaré plains and the
Benue Valley. (2) The savanna zone is composed of
the Adamawa highlands, the Tikar plain, the low land
savanna of the Center and East regions, and the high-
land of the West and Northwest regions. (3) The tropi-
cal forest zone is composed of the degraded forests of
the Central and Littoral regions, and the tropical rain-
forests of the Southwest and East regions. (4) The
coastal and marine zone spreads along the Gulf of
Guinea. The country’s economy is driven by agro-in-
dustry in the coastal, central and southern zones (Molua
and Lambi, 2006). Because of the above geographic
characteristics, its wide range of climatic types, and
its cultural diversity, Cameroon is often nicknamed
“Africa in miniature”. The population of Cameroon is
estimated to be ~23 million as of January 2015 (http:/
/countrymeters.info/en/Cameroon). According to the
Demographics of Cameroon (http://en.wikipedia.org/
wiki/Demographics_of_Cameroon), the country com-
prises an estimated 250 distinct ethnic groups, which
may be classified into five large regional-cultural di-
visions: (1) the western highlanders (Semi-Bantu or
grassfielders), including the Bamileke, Bamoun, and
many smaller Tikar groups in the Northwest (~38% of
the total population); (2) the coastal tropical forest
peoples, including the Bassa, Duala (or Douala), and
many smaller groups in the Southwest (12%); (3) the
southern tropical forest peoples, including the Beti-
Pahuin with subgroups called Bulu, Fang, Maka, Njem,
and Bakapygmies (18%); (4) the predominantly Islamic
peoples of the northern semi-arid regions (the Sahel)
and central highlands, including the Fulani (or Fulbe)
(14%); and (5) the “Kirdi”, non-Islamic, or recently
4M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Islamic, peoples of the northern desert and central high-
lands (18%). Since people of different ethnic groups
speak different languages, French and English, inher-
ited from colonialism, are used for mutual communi-
cation, although they retain their original languages.
2-2. Cameroon lakes
There are at least 39 lakes of volcanic origin that are
distributed along the CVL (Kling, 1988; Issa et al.,
2014a). Lake Nyos in the Northwest Region of
Cameroon (06∞26¢ N and 10∞17¢ E) is a meromictic
volcanic crater lake with a N-S length of ~2.0 km, E-
W length of ~1.2 km, surface area of 1.58 km2, and a
maximum depth of 209.5 m. It lies in the Oku volcanic
field along the CVL, and was formed by a basaltic
phreato-magmatic eruption (Lockwood and Rubin,
1989). The age of the lake, which has been a topic of
controversy, will be described later (Subsection 6-1).
Lake Monoun in the West region of Cameroon lies at
05∞35¢ N and 10∞35¢ E, and is also a meromictic vol-
canic crater lake with a NEE-SWW length of ~1.6 km,
a maximum NW-SE width of ~0.7 km, a surface area
of 0.31 km2, and a maximum depth of 99 m. It belongs
to the Oku volcanic center along the CVL. The age of
the lake is unknown.
2-3. Lake Nyos disaster: The event and testimo-
nies
Unusual news raced around the world in late August
1986. The first news that reached Japan reported that
40 local people had been killed by a “poisonous” gas
released from a volcanic lake (Lake Nyos) in
Cameroon. The number of casualties increased to
~1200 in a later report. In response to a request for
international support by the Government of the Repub-
lic of Cameroon, the Japan International Cooperation
Agency (JICA) under the Ministry of Foreign Affairs
of Japan sent a Japan Disaster Relief Team (JDR) to
the site. I was asked to join the team as a volcanic gas
expert. It was my first visit to Cameroon. The JDR team
arrived at Douala on 28 August, 1986. A few days later
the team was taken to Lake Nyos by helicopter because
of poor road conditions and heavy rains in the Nyos
area. We were shocked by the terrible scenes we wit-
nessed (Fig. 2), and the reddened surface water of the
lake (Fig. 3), which increased our anxiety concerning
the cause(s) of the disaster. Since the main purpose of
JDR was to provide relief supplies and medical care to
the refugees, we made an initial cursory scientific sur-
vey during this first visit. There was no indication of
the direct involvement of volcanic gases as initially
suggested, for we did not find any trace of acid gases
like SO2, H2S and HCl, which are major components
of high-temperature volcanic gases. Later reports in-
dicated that the cause of the deaths was CO2 gas re-
leased from Lake Nyos on the evening of 21 August,
1986, and that the gas killed 1746 people and ~8000
livestock by asphyxiation (Kling et al., 1987; Kusakabe
et al., 1989). Exactly 2 years prior to the Lake Nyos
disaster, there had occurred a very similar gas release
from Lake Monoun on 15 August, 1984, that killed 37
people also by asphyxiation by CO2 gas released from
the lake. These extremely unusual disasters had never
been recorded before, and therefore constituted a new
type of natural disaster (Sigurdsson, 1987a).
August 21, 1986, was a Thursday, and a market day
at the Nyos village. Local people were selling and ex-
changing their agricultural products and articles for
daily use. At the time when the catastrophe occurred
(8~9 p.m.), people must have been relaxed and chat-
Fig. 3. An aerial view of Lake Nyos taken 10 days after the limnic eruption (photo taken by the author). Debris of vegetation
washed away from the shore was floating on the reddened lake surface.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 5
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
ting at home, or drinking beer outside. Some people
may have already been in bed. Such a peaceful situa-
tion was disrupted by a sudden release of lethal gas
from Lake Nyos. It was indeed a nightmare. It is obvi-
ous that the local people did not understand what hap-
pened. From the testimonies collected later from sur-
vivors by journalists and researchers, the event may
have proceeded as follows. Some people heard faint
rumbles and noises like a car coming from a distance.
They went out of the house to look around, and then
felt a tepid breeze with a smell of rotten eggs or gun
powder. Most people fell down, lost consciousness and
died (Fig. 2). At Nyos village, where 1200 people lived
at that time, only a few people (4~6) survived. The
survivors were stunned to find that they had lost most
of their family members, relatives and friends.
Sigurdsson (1987b) noted that “some survivors of the
disaster attributed it to the wrath of their dead tribal
chief, who, on his deathbed in 1983, ordered that his
best cattle be driven off the sheer cliffs above Lake
Nyos as a sacrifice to the spirit of the lake, Mami-Wa-
ter. But the chief’s family failed to honor his last wish,
and many today believe that the 1986 calamity was an
expression of the chief’s posthumous displeasure”.
Sigurdsson (1987b) also described that, four days be-
fore the lethal event, local herdsmen noticed unusual
bubbling on the lake’s surface, which prompted twenty
five of them to move to a distant village. There were
also unconfirmed reports claiming the emission of
foaming water and vapor from the lake two to three
weeks earlier. At about 4:00 p.m. on August 21, nearby
herdsmen heard strange bubbling sounds and observed
a slight emission of vapor from the lake. At about 8:00
p.m., villagers in Cha, a village about 7 km northwest
of the lake, heard two loud noises, followed by three
rumbles at about 9:00 p.m., when activity built up to
the climactic disaster.
According to Aramaki et al. (1987) who interviewed
Mallam Jae, a local farmer who lived at a place 120 m
higher than the lake surface, the gas explosion took
place at about 8:30 p.m. on August 21 and continued
until 1 a.m. the next day. This account of the time at
which the events took place may be reliable, since
Mallam Jae was wearing a nice wrist watch. Initially,
Mallam Jae heard sounds like a murmur followed by
detonations. He also felt tremors and a smell of gun
powder. The next morning he found the lake quite unu-
sual. Le Guern et al. (1992) published details of inter-
views with some local people who spoke in pidgin
English (which was translated into English) about what
they saw. Observations by local people included: (1)
Minor upwelling of hydrothermal waters from the bot-
tom of Lake Nyos on August 20, one day before the
event. (2) A small explosion that took place at 4:00
p.m. followed by a major explosion between 8:00 and
8:30 p.m. on August 21. (3) A water jet accompanied
by white illuminations. (4) A detonation was heard in
the village at 11:00 p.m. (5) Minor events with an
upwelling of hydrothermal water and gas occurred in
Lakes Nyos and Njupi, a small and shallow lake 2 km
east of Lake Nyos (Chevrier, 1990). (6) White cloud
was seen during the catastrophe on August 21. Based
on these testimonies and observations, Le Guern et al.
(1992) preferred the interpretation that the Lake Nyos
catastrophe was caused by the input of hot hydrother-
mal fluids containing CO2 into the lake and surround-
ing area. Their interpretation seemed to have been in-
fluenced by their experience at Dieng volcano (Indo-
nesia) where CO2 gas, originating from a phreatic erup-
tion of the volcano, killed 142 local people who were
fleeing from the site (Le Guern et al., 1982). How-
ever, there is a view that the anecdotal evidence (such
as “the smell of rotten eggs and gun powder, rumbling
noise heard at distance”, etc.), collected soon after the
disaster through interviews with local survivors by
journalists and scientists, should be interpreted with
care because stories told by local people may have been
tailored to give answers to please the visiting inter-
viewers (Freeth, 1990). The author adopts this view
and believes that the phreatic hypothesis did not have
a firm scientific basis, because this interpretation of
the events was largely based on the testimonies and
anecdotal evidence.
In March 1987, a Cameroon Government and
Fig. 4. Photograph showing a white cloud still remaining
along the valley downstream of Lake Nyos. Nyos village is
seen at the bottom. The photo was taken 2 days after the
eruption by a helicopter pilot carrying a Catholic mission
(supplied by G. Tanyileke).
6M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
UNESCO-sponsored international conference on the
Lake Nyos Disaster was held in Yaoundé, the capital
of Cameroon. More than 200 scientists participated and
presented the results of their initial studies on the geo-
logical, geochemical, physical, medical and socio-an-
thropological aspects of the disaster (Sigvaldason,
1989). Regarding the cause of the gas burst, there was
a sharp confrontation between a group of scientists who
believed that the lake played a key role in the accumu-
lation of the CO2 which was subsequently released (this
interpretation was later named “limnic eruption hypoth-
esis”) and another group of scientists who believed that
the cause of the Nyos catastrophe was due to a vol-
canic (phreatic) eruption from the bottom of the lake
(volcanological or phreatic hypothesis) (Tazieff, 1989;
Barberi et al., 1989; Le Guern et al., 1992). Disagree-
ment between the two scientific views resulted in a
compromise of the resolutions of the Yaoundé Confer-
ence (Sigvaldason, 1989), and encouraged the need for
follow-up investigations, which clearly indicated a
steady supply of magmatic CO2 from the lake bottom
and its accumulation in the lake. This gave strong sup-
port to the limnic eruption hypothesis (Kling et al.,
2005; Kusakabe et al., 2000, 2008). This hypothesis
will be described in detail in Sections 3 and 4.
The gas released from Lake Nyos was almost pure
CO2 (Kling et al., 1987; Kusakabe et al., 1989). Since
CO2 gas is 1.5 times denser than air at room tempera-
ture, and since it may have been cooled due to adi-
abatic expansion when released from the pressurized
deep part of the lake, the density of the gas was likely
to have been significantly greater than that of the am-
bient atmosphere. This would have facilitated its flow
along low-lying areas, such as valleys, before mixing
with air. Costa and Chiodini (2015) simulated the gas
flow, using a computer code TWODEE-2, for 4 differ-
ent scenarios that considered different gas masses and
fluxes from Lake Nyos in 1986. The simulations, indi-
cating how far and fast the cloud dispersed after the
limnic eruption, are useful for making up a hazard map
of the area. Figure 4 is a photograph taken two days
after the eruption by a Helimission helicopter pilot (G.
Tanyileke, pers. commun.) and shows that the white
cloud was still present along a valley downstream of
the lake. Figure 5 (modified from Sigurdsson et al.,
1987a) shows the gas flow path estimated from the
distribution of victims around Lake Nyos. The gas
cloud traveled more than 20 km, asphyxiating people
on its way before dissipating into air. The number of
victims was 1200 at Nyos village, 300 at Cha village
and 52 at Subum village. More than 8000 cattle were
also killed. Survivors were evacuated to 7 resettlement
camps, namely, Kimbi, Buabua, Kumfoutu, Yemge,
Ipalim, Esu and Upkwa (around Lake Wum). As of July
2015, these people were still cut off from the general
population, as neither the national radio, electricity
grid, nor television signals reached them.
Since the victims were asphyxiated almost instantly,
the oxygen concentration in the gas must have been
extremely low compared to normal the atmosphere, or
the CO2 concentration was very high. Table 1 shows
the effect of some gases on human health (Kusakabe
et al., 1989). Mammals, including human beings, live
on a normal atmosphere that contains 21 vol % of O2.
If air is breathed containing less than half of this nor-
mal air concentration of O2, a coma, fainting, cyano-
sis, syncope, respiratory arrest, and ultimately, cardiac
arrest can result. If we breathe air containing high con-
centrations of CO2 (e.g., >10 vol %), a coma, and even-
tually, death can result. Some survivors of the Lake
Nyos disaster were found to suffer from pulmonary
edema, respiratory distress, conjunctivitis, and skin
lesions or “burns” (Baxter et al., 1989). The skin burns
were taken by the phreatic hypothesis supporters as
evidence that the gas was at a high temperature and
contained some acidic, corrosive components, such as
SO2 (which turned to sulfuric acid later) and HCl that
are commonly contained in high-temperature volcanic
gases. This interpretation was highly unlikely, since
vegetation, and clothes on the victims stayed intact and
no appreciable level of sulfur and chlorine components
were found in the lake water (Kusakabe et al., 1989).
The medical interpretation of the skin damage or blis-
ters was that the body’s metabolic rate was drastically
reduced in a state of deep coma, inducing a severely
restricted circulation of blood. As a consequence, the
capillary vessels of the skin lacked circulation, result-
ing in necrosis and the development of skin lesions on
approximately 5% of the patients (Baxter et al., 1989).
Fig. 5. Distribution of localities where victims were found
around Lake Nyos (red circles), and estimated flow paths of
the gas (arrows). Modified from Sigurdsson (1987a).
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 7
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
2-4. Lake Monoun disaster
Lake Monoun experienced a gas burst on 15 August,
1984, that killed 37 people by asphyxiation. A recon-
struction of the event (Sigurdsson et al., 1987) showed
that at almost midnight of that day, people in Njindoun,
a village about 5 km north of the lake, heard a loud
noise in the vicinity of the lake. They informed the
nearby police early next morning. A policeman together
with a medical doctor went to the site where they saw
a whitish, smoke-like cloud that covered the ground to
a height of ~3 m. Since they became nauseous and
dizzy, they left the site and moved to Njindoun vil-
lage. After the cloud dissipated, they came back to the
site and found dead people lying on the road. The vic-
tims had skin lesions or blisters. Clothes were not af-
fected. Domestic and wild animals were also found
dead. Altogether 37 people were killed by this event.
A survivor described the smell of the gas cloud as
“sulfurous like a car battery”. It was found in a later
survey that vegetation at the east end of the lake was
flattened, indicating that the water wave locally reached
up to 5 m high, and that the color of the lake water
changed to a reddish brown. From the above state-
ments, the Lake Monoun event was very similar to the
Lake Nyos event. For this reason, it seems appropriate
to describe and compare, at the same time, the results
of the geochemical surveys made after the gas bursts
at both lakes.
2-5. Unusual geochemical characters of Lakes
Nyos and Monoun
A scientific survey of Lake Monoun was undertaken
by the Office of Foreign Disaster Assistance (USAID)
several months after the gas burst, upon a request by
the Cameroonian Government (Sigurdsson et al.,
1987). They found unusual chemical characteristics.
Waters below 50 m were anoxic, dominated by high
Fe2+ (~200 mg/l) and HCO3– (~1000 mg/1), and su-
persaturated with siderite, a major component of the
crater floor sediments. The unusually high Fe2+ levels
were attributed to the reduction of laterite-derived fer-
ric iron that was gradually brought into the lake as loess
and in river input. Sulfur compounds were below de-
tection limits in both water and gas. Table 2 shows the
chemical composition of Lake Monoun water samples
collected between 27 February and 16 March, 1985
(Sigurdsson et al., 1987). It includes data for samples
collected in 1986 (Kusakabe et al., 1989) and 1993
(Kusakabe et al., 2008). Analysis of the 1985 and 1986
samples showed lower gas and ionic contents than the
original solution. This was interpreted to be due to (i)
loss of CO2 from the waters during retrieval of the
Niskin sampler from the lake, (ii) loss of CO2 from
waters collected in the sample containers, and/or (iii)
oxidation and precipitation of iron prior to the analy-
sis. During the early stages of investigation of the Lakes
Monoun and Nyos disasters, these problems high-
lighted the difficulty of sampling and the analysis of
CO2-rich waters. The data for the 1993 samples were
more reliable (Kusakabe et al., 2008). Based on the
data obtained in 1985, however, Sigurdsson et al.
(1987) had come to the important conclusion that ac-
cumulation of CO2(aq) in the lake was attributed to the
long-term emission of magmatic CO2(aq) from vents
within the crater, which led to a gradual build-up of
CO2(aq) and HCO3– in the lake, i.e., an essential con-
cept of the “limnic eruption hypothesis”. It is noted
that the manuscript of the paper by Sigurdsson et al.
Table 1. Effect of some gases on human health*.
Concentration in atmosphere Stage Response and symptoms
O
2
(%)
21 1 Normal
16-12 2Lowered concentration, headache
14-93Disorientation, unstable gait, headache, nausea, vomiting, facial pallor, somnolence
10-64Coma, fainting, damage of central nervous system, cyanosis, convulsion
<6 5 Syncope, coma, bradypnea, respiratory arrest, cardiac arrest
CO
2
(%)
0.04 1 Normal
1.5 2 Changes in physiological ranges (techypnea etc.)
5Shortness of breath, headache, coma
10 4 Coma in 10-15 min. exposure
>40 5 Sudden death
3
*Reproduced from table 6 of J. Volcanol. Geotherm. Res. 39, Kusakabe, M., Ohsumi, T. and Aramaki, S., The Lake Nyos gas
disaster: chemical and isotopic evidence in waters and dissolved gases from three Cameroonian crater lakes, Nyos, Monoun
and Wum, 167–185, Copyright 1989, with permission from Elsevier.
8M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
(1987) had been prepared prior to the 1987 Interna-
tional Conference on the Lake Nyos disaster in
Yaoundé, so the American team who started an initial
scientific survey at Lake Nyos must have had a gen-
eral idea of the cause of the disaster.
Lake Nyos also has unusual chemical and physical
characteristics similar to Lake Monoun. Dissolved spe-
cies are overwhelmingly dominated by CO2(aq) fol-
lowed by HCO3–, Fe2+, Mg2+, Ca2+, SiO2, NH4+, Na+,
K+ in decreasing order. The concentration of the dis-
solved species in the water column increases with depth
with maximum values reached at the bottom (210 m).
Follow-up studies of Lakes Nyos and Monoun clearly
indicated that the CO2 content in the lakes was increas-
ing at an unusually high rate for a geological phenom-
enon (Evans et al., 1993; Kusakabe et al., 2000). This
situation led scientists working on Lakes Nyos and
Monoun to warn of the possible recurrence of a limnic
eruption in the near future and to recommend the arti-
ficial removal of dissolved gases from the lakes (Freeth
et al., 1990; Tietze, 1992; Kling et al., 1994; Kusakabe
et al., 2000). To achieve this goal, the Nyos-Monoun
Degassing Program (NMDP) was set up by scientists
who were deeply involved in the disaster mitigation
issues of the limnic eruption. After experimental
degassing at Lake Monoun (Halbwachs et al., 1993)
and Lake Nyos (Halbwachs and Sabroux, 2001), a per-
manent degassing apparatus was installed at Lake Nyos
in 2001 and at Lake Monoun in 2003 under the NMDP,
funded by the U.S. Office of Foreign Disaster Assist-
Depth pH Na
+
K
+
NH
4
+
Mg
2+
Ca
2+
Fe
2+
SO
4
2-
Cl SiO
2
HCO
3
-
CO
2(aq)
mmg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l
February 1985*
1
06.9 9 2.2 <0.1 6.0 8.7 <0.02 <1 <1 19 88 26
15 5.8 17 4.7 6.2 22 20 0.03 <1 1.8 40 265 952
61 6.3 25 5.6 15 29 41 200 <1 3.2 35 1050 1086
90 6.4 24 5.7 13 30 42 220 <1 3.4 45 1050 890
90 6.1 24 5.7 15 30 41 170 <1 3.3 50 775 1311
90 6.3 25 5.8 18 30 43 260 <1 3.2 48 805 851
90 6.4 24 5.9 13 30 42 190 <1 3 41 805 680
90 6.0 26 5.5 15 29 42 290 <1 3.4 50 1000 2112
90 5.9 26 5.5 12 30 41 190 <1 3.2 51 885 2357
October 1986*
2
0æ11 3 æ4.2 4 1.7 0.1 0.4 17 57 16
15 æ14 4.2 6 13 10 110 0.3 1.0 40 421 202
25 æ17 5.3 11 17 12 190 0.2 1.4 37 657 542
50 æ17 5.1 17 13 10 340 0.4 2.5 38 1087 2859
75 æ21 7.2 26 22 18 540 0.4 2.5 42 1520 2385
95 æ22 7.2 26 23 19 590 0.2 2.6 44 1660 2922
January 1993*
3
10 6.58 13 3.5 10 8 13 100 0.1 1.1 34 82 4
20 6.46 15 4.3 12 15 21 259 0.1 1.5 58 105 18
30 6.00 18 4.6 18 24 30 464 0.1 1.5 86 1352 1809
40 5.91 19 5.0 19 24 33 505 0.1 1.6 91 1448 2271
50 5.80 21 5.3 22 23 38 533 0.1 1.8 95 1546 3217
55 5.60 23 5.5 28 27 46 641 0.1 2.3 110 1823 6760
75 5.58 24 5.8 28 27 46 646 0.1 2.4 114 1862 6848
90 5.60 24 5.4 30 28 48 682 0.1 2.4 114 1961 6813
95 5.66 25 5.9 37 29 50 804 0.1 2.6 123 2272 6778
100 5.72 26 7.4 39 29 51 918 0.5 2.8 129 2523 6029
-
Table 2. Representative analyses of water samples collected in 1985, 1986 and 1993 from Lake Monoun.
*1Sigurdsson et al. (1987).
*2Reproduced from table 2 of J. Volcanol. Geotherm. Res. 39, Kusakabe, M., Ohsumi, T. and Aramaki, S., The Lake Nyos gas
disaster: chemical and isotopic evidence in waters and dissolved gases from three Cameroonian crater lakes, Nyos, Monoun
and Wum, 167–185, Copyright 1989, with permission from Elsevier.
*3Reproduced from table 1 in Kusakabe et al. (2008).
Numbers in italics were calculated assuming carbonate equilinria.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 9
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
ance (USAID) and the Cameroonian Government. Con-
trolled degassing is continuing successfully at both
lakes. Figure 6 shows the amazingly beautiful
fountains in the initial phase of degassing at Lakes
Nyos and Monoun, a 45 m high fountain at Lake Nyos
(Feb. 2001) and a 8 m high fountain at Lake Monoun
(Jan. 2004). The degassing system and the construc-
tion of the degassing pipes have been described in
Halbwachs et al. (1993, 2004). There was concern that
artificial degassing might trigger another limnic erup-
tion, since degassing could bring deep water to the
surface, which will become cooler due to adiabatic
cooling, and therefore may sink and destabilize the lake
(e.g., Freeth, 1994). However, numerical modeling of
the evolution of CO2 in the lake under different input
conditions (McCord and Schladow, 1998; Kusakabe et
al., 2000; Schmid et al., 2003, 2006) suggested that
destabilization of the water column due to controlled
degassing could not pose an immediate threat from a
“man-made” limnic eruption. However, the possibil-
ity of thermal instability of the water column between
50–70 m, which could become a trigger for a limnic
eruption, was suggested by Schmid et al. (2004), for
they found double-diffusive convection at that depth
range. In agreement with the results of the numerical
modeling, the observed chemical structure of the lakes
after the initiation of the controlled degassing opera-
tion indicated that a stable stratification was estab-
lished, which remained basically the same as the pre-
degassing situations at both lakes (Kling et al., 2005;
Kusakabe et al., 2008).
As stated above, the chemical and physical structure
of Lakes Nyos and Monoun evolved steadily with time
until the early 2000s when degassing started. After the
initiation of gas removal, the lake structure was obvi-
ously affected by degassing. For this reason, the evo-
lution is better described separately as “pre-degassing”
and “syn-degassing”.
3. Pre- and syn-degassing evolution of Lakes Nyos
and Monoun
3-1. Pre-degassing evolution
The first scientific reports on the 1986 Lake Nyos
gas disaster were published by Freeth and Kay (1987),
Kling et al. (1987) and Tietze (1987). Of these, Kling
et al. (1987), which is easily accessible, gave the most
comprehensive results of the initial survey of the dis-
aster, which included the geology of the region, the
geochemistry of water and gas from the lake, and the
pathology of hospitalized people and victims. They
concluded that (i) the gas released was CO2 that had
been stored in the lake’s hypolimnion (bottom layer),
(ii) the victims died of CO2 asphyxiation, (iii) CO2 was
derived from magmatic sources, and (iv) there was no
direct volcanic activity involved. Kusakabe et al.
(1989) reached similar conclusions on the basis of
water chemistry and carbon and noble gas isotopic
compositions of the gases dissolved in Lakes Nyos and
Monoun. They noted that the H2S concentration in the
released gas must have been far below a lethal level, a
point that precluded the phreatic eruption hypothesis
(see above). The same authors also reported the first
petrochemical data of ejecta around Lakes Nyos and
Monoun, which indicated that the lavas were transi-
Fig. 6. Fountains from the degassing pipes. The fountain heights were 45 m at Lake Nyos, February 2001 (a) and 8 m at Lake
Monoun, January 2004 (b). The tapping depth of the pipes was 203 m and 73 m, respectively.
10 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
tional to slightly alkaline in composition (see below).
A detailed temporal variation of the chemical struc-
ture of Lakes Nyos and Monoun since the limnic erup-
tions at both lakes was reported by Kusakabe et al.
(2008). This paper presented the most comprehensive
data set of chemical composition, conductivity, tem-
perature, pH and CO2 profiles obtained from measure-
ments taken almost every 2–3 years from 1986 to 2006,
which enabled an evaluation of the evolution of CO2
in the lakes over a period of about 20 years and which
encompassed pre- and syn-controlled degassing peri-
ods. The chemical structure of the lakes is best repre-
sented by a conductivity profile (Fig. 7). Both lakes
have a similar chemical structure which is character-
ized by four distinct layers. At Lake Monoun, layer I
is the shallowest, is well-mixed, and contains low con-
ductivity water. A sharp chemocline separates layers I
and II at 23 m in January 2003. Layer II extends down
to a 51 m depth, where a second chemocline develops.
A well-mixed layer III continues down to ca. 85 m.
Below this depth, conductivity increases steadily to-
ward the bottom (layer IV). Lake Nyos has basically
the same structure as Lake Monoun in January 2001:
layer I is the shallowest, is well mixed, and contains
low conductivity water. A sharp chemocline at about a
50 m depth separates layers I and II, the latter extend-
ing down to about a 180 m depth. A lower chemocline
develops around this depth, below which a well-mixed
Fig. 7. Profiles of electric conductivity normalized at 25∞C (abbreviated as C25) at Lake Monoun, January 2003 (left) and
Lake Nyos, January 2001 (right). The water column of each lake can be divided into 4 layers, each separated by a chemocline.
Reproduced from Fig. 1 and Fig. 7 of Kusakabe et al. (2008).
Fig. 8. Evolution of pre-degassing temperature profiles at Lake Monoun (a) (1986–2003) and (b) Lake Nyos (1986–2001).
Reproduced from figures 4 and 9 of Kusakabe et al. (2008), but colored.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 11
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
layer III continues down to ~203 m. Below this depth,
conductivity sharply increases towards the bottom
(layer IV).
Pre-degassing temperature variations at Lakes
Monoun (October 1986 to January 2003) and Nyos
(November 1986 to January 2001) are shown in Figs.
8a,b, respectively, (reproduced from Kusakabe et al.,
2008). Temperature profiles at Lake Monoun show a
minimum at the 5–21 m range (layer I); followed by
(i) an increasing temperature to about 23∞C down to
the lower chemocline at depths of 50–63 m (layer II),
(ii) constant values down to around 90 m (layer III),
and (iii) a second increase to >23∞C towards the bot-
tom (layer IV). It is worth noting that the temperature
of the layer III water increased significantly between
1986 and 1999, and that, at the same time, layer III
(thermally homogeneous zone) widened, forming a
“shoulder” at a depth of 51 m. This widening suggests
that warmer water was added to layer IV, and the pro-
files were pushed upward. A simple heat balance indi-
cates that the heat accumulated in layers III and IV
during 15 years (October 1986 to January 2003) is 7.8
¥ 1012 J, supplying heat at an average rate of 5.1 ¥
1011 J/yr (~0.02 MW). The incremental upward move-
ment of the lower thermocline (Fig. 8a) indicates the
addition of water to layer IV, most likely from the bot-
tom. If 4.1 ¥ 108 tons of water having a temperature of
27∞C were added, it would account for the heat accu-
mulation during that period. Diffusive and conductive
heat loss to layer II and above was not taken into con-
sideration in this simple heat balance calculation, thus
giving a minimum heat supply. Note that the rate of
heat and water supply to layers III and IV initially ap-
pears high judging from the change in the temperature
profiles (Fig. 8a).
Similar to Lake Monoun, the temperature of the Lake
Nyos bottom water increased continuously after the
limnic eruption in 1986 (Fig. 8b), indicating a heat in-
put into the lake. The heat input to layers III and IV
was reported to decrease from an initially high value
of 0.93 MW (August 1986 to May 1987) to 0.43 MW
(November 1986 to December 1988, Nojiri et al., 1993)
to 0.32 MW (May 1987 to September 1990, Evans et
al., 1993).
Figure 9 shows the temporal change in the pre-
degassing conductivity profiles at both lakes. As pre-
viously stated, Lake Monoun profiles have a “shoul-
der” between layers II and III. The shoulder became
shallower and sharper, and layer III widened with time
and its conductivity increased (Fig. 9a). Vertical con-
ductivity profiles in layer III suggest that the layer is
well mixed. The rise of the shoulder indicates an addi-
tion of recharge water from the bottom, pushing bot-
tom water upward. This is consistent with the changes
in the temperature (Fig. 8a). By combining the con-
ductivity profiles from October 1986 to January 2003
(15 years) with the bathymetry used in Kling et al.
(2005), we calculated an overall increase of 2.7 ¥ 103
tons of Total Dissolved Solids (TDS) in layers III and
IV. This translates into an average annual TDS input
rate of 1.7 ¥ 102 tons/yr. The sharp conductivity rise
toward the bottom in layer IV may be related to disso-
lution and reduction of particles containing ferric com-
pounds to release Fe2+ and HCO3– in the sediments that
are rich in organic material. The concentration of Fe2+,
HCO3– and NH4+ increased significantly with depth
Fig. 9. Evolution of pre-degassing conductivity profiles at Lake Monoun (a) (1986–2003) and (b) Lake Nyos (1986–2001).
Reproduced from figures 4 and 9 of Kusakabe et al. (2008), but colored.
12 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
only in layer IV whereas the other ions such as Na+,
K+, Mg2+, Ca2+ showed a steady increase with depth
(Kusakabe et al., 2008).
At Lake Nyos, shallow water in November 1986 had
a higher conductivity, even at about 7 m (Fig. 9b), than
that in later years, indicating that deep, TDS-rich wa-
ter was brought to the surface during the limnic erup-
tion, the effect still remaining 3 months after the limnic
eruption. This upper chemocline in November 1986
deepened gradually with time down to 30 m in 1988,
47 m in 1993 and 50 m in 2001. Conductivity profiles
at mid-depths (70–160 m) stayed almost unchanged for
15 years after the eruption, suggesting that transport
of dissolved chemical species through layer II was lim-
ited. The conductivity in layers III and IV (170–210
m) increased notably with time (Fig. 9b). In January
2001, the conductivity profile between 185 m and 202
m became steep, with an associated slight reduction of
earlier high conductivity in layer IV, indicating the ini-
tiation of mixing in the deepest zone. This tendency
had started in 1998, although the 1998 profile is par-
tially obscured behind the 2001 profile in Fig. 9b. From
the depth of 205 m to the bottom, the conductivity in-
creased sharply. This trend is the same as observed at
the bottom water of Lake Monoun. The pre-degassing
increase of the conductivity in layers III and IV from
November 1986 to January 2001 (14 years) corresponds
to an increase of 7700 tons of TDS, with the average
annual input of 540 ton/yr. Initially, the input rate was
relatively high, but later decreased by at least a factor
of two as shown by the close spacing of the conductiv-
ity profiles (Fig. 9b). This temporal trend was similar
to that of the water temperature.
Before describing the temporal variation of CO2 pro-
files in the lakes, it may be worthwhile mentioning the
analytical methods used to determine the dissolved
CO2. As stated before, during the early days of our
observations, we encountered difficulties in measur-
ing CO2 from deep water. The partial pressure of dis-
solved gases in Lake Nyos was so high (~1.1 MPa in
1990, Evans et al., 1993) that we could not use a Niskin
water sampler to collect water and gases, since the lid
of the sampler was forced open before retrieval due to
Fig. 10. Schematic presentation of the “MK sampler”. Re-
printed from figure 2 of J. Volcanol. Geotherm. Res. 97,
Kusakabe, M., Tanyileke, G., McCord, S. A. and Schladow,
S. G., Recent pH and CO2 profiles at Lakes Nyos and
Monoun, Cameroon: implications for the degassing strategy
and its numerical simulation, 241–260, Copyright 2000, with
permission from Elsevier.
Fig. 11. Comparison of the CO2 concentrations at Lake Nyos
measured by the pH method (solid curve) in March 1995,
those obtained by the syringe technique (red open circles,
November 1993) and the cylinder technique (blue open
squares, April 1992, Evans et al., 1994). Dotted curves along
the pH-based CO2 profile indicate possible errors due to an
uncertainty in the pH measurement of ±0.05. Modified from
figure 2 of J. Volcanol. Geotherm. Res. 97, Kusakabe, M.,
Tanyileke, G., McCord, S. A. and Schladow, S. G., Recent
pH and CO2 profiles at Lakes Nyos and Monoun, Cameroon:
implications for the degassing strategy and its numerical
simulation, 241–260, Copyright 2000, with permission from
Elsevier.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 13
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
the exsolution of high-pressure dissolved gases. This
difficulty was partially solved by attaching a gas bag
to the Niskin sampler, which enabled us to collect wa-
ter samples (Kusakabe et al., 1989), but it was still
difficult to measure the dissolved CO2 accurately. To
overcome this difficulty we developed a new method
called the “the MK or syringe method” (Kusakabe et
al., 2000). The sampler in the MK method is schemat-
ically shown in Fig. 10. In this technique, the total dis-
solved carbonate species (H2CO3 + HCO3– + CO32–) at
a given depth is fixed in situ in a 50-ml plastic syringe
containing concentrated (5 M) KOH solution. The to-
tal carbonate concentration in the alkaline solution is
determined later in the laboratory by a classical mi-
cro-diffusion method (Conway, 1958). Subtraction of
the HCO3– concentration and the blank carbonate in
the KOH solution from the total carbonate concentra-
tion gives the H2CO3 (or CO2,aq) concentration.
Since the total carbonate species dissolved in Lakes
Nyos and Monoun are essentially controlled by car-
bonate equilibria, it is possible to determine the H2CO3
(or CO2,aq) concentration from pH values (measured
by CTD) if the HCO3– concentration at a correspond-
ing depth is known (Kusakabe et al., 2000). The HCO3–
concentration is closely related to the electric conduc-
tivity, so we can calculate the concentration of H2CO3
(or CO2,aq) of a water column using the CTD data. It is
a big advantage that we can obtain a continuous CO2
profile, although very careful calibration of the pH
sensor is an important prerequisite. This method is
called “the pH method”. Figure 11 compares CO2 pro-
files at Lake Nyos in 1995 obtained by the pH and sy-
ringe methods. In the figure, the results obtained by
the “cylinder method” are included. In the cylinder
method (Fig. 12a), deep water was sucked into a re-
motely-operated evacuated stainless steel cylinder and
the exsolved total CO2 was later analyzed in the labo-
ratory (Evans et al., 1993). They also introduced an
interesting device called a “gas-probe” (Fig. 12b) with
which the total gas pressure at a given depth of a water
Fig. 12. (a) Cylinder sampler used by Evans et al. (1993). A pre-evacuated cylinder is deployed to a desired depth, and a check
valve is opened to sample water. (b) Gas pressure probe used by Evans et al. (1993). Dissolved gas molecules except water
diffuse through the membrane unit consisting of multiple silicone rubber tubing. The total gas pressure inside the collection
chamber is measured at the surface. Reprinted from figures 3 and 4 of Appl. Geochem. 8, Evans, W. C., Kling, G. W., Tuttle,
M. L., Tanyileke, G. and White, L. D., Gas buildup in Lake Nyos, Cameroon: The recharge process and its consequences,
207–221, Copyright 1993, with permission from Elsevier.
14 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
column was measured. Figure 11 shows the satisfac-
tory agreement between the CO2 concentration ob-
tained by the different methods.
Recently, “the plastic hose method” (Yoshida et al.,
2010) for CO2 determination has also been used. This
is based on a self-gas-lifting principle in a plastic hose
that is deployed into the deep water of the lake. A mix-
ture of gas and water spouting from the hose is sepa-
rated into liquid and gas phases by using a plastic sepa-
rator. The liquid phase accumulates in the separator
and is collected as the water sample, while the dry gas
flows through a volumetric gas meter to measure the
gas volume at the sampling time (Fig. 13). A gas sam-
ple collected directly from the dry gas line is perfectly
free from air contamination which is essential for no-
ble gas analysis (Nagao et al., 2010). A similar method
has been reported by Tassi et al. (2009) for gas collec-
tion from Lake Kivu.
Fig. 13. Analytical system for measuring CO2 concentrations in gassy lakes (copied from Yoshida et al., 2010). Two-phase
flow (gas and water) from a plastic hose deployed to a desired depth in the lake is introduced into a separator and a gas flow
meter. The amount of water accumulated in the separator and the volume of gas that goes through the flow meter in a given
time are measured, from which the CO2 concentration is calculated.
Fig. 14. Evolution of pre-degassing CO2 profiles at (a) Lake Monoun (1986–2003) and (b) Lake Nyos (1986–2001). The
saturation of CO2 in water at 25∞C is shown by a dashed line. Reproduced from figures 4 and 9 of Kusakabe et al. (2008), but
colored. Note that the CO2 concentrations at a depth of ~55 m in 2001–2003 at Lake Monoun are close to saturation.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 15
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
More recently, a new and simple method of measur-
ing the total CO2 (CO2,aq and HCO3–) has been devel-
oped by Saiki et al. (2016). This method is based on a
linear relationship between the total CO2 concentra-
tion and the sound velocity in lake water.
Temporal pre-degassing CO2(aq) variations at Lakes
Monoun and Nyos are shown in Fig. 14. Although no
data were available in 1986 at Lake Monoun, the 1986
profile was estimated from later CO2-conductivity re-
lationships (Kusakabe et al., 2008). Figure 14 shows
that CO2(aq) concentrations in deep lake water were
around 130 mmol/kg in layers III and IV, with the CO2
Date Time aftereruption Total CO2CO2 below layer II CO2 accumulation rate CO2 removal rate
year giga mol giga mol giga-mol/yr giga-mol/yr
Lake Monoun: Pre-degassing
October 1986 2.17 0.38 0.38 ææ
November 1993 9.25 0.53 0.53 ææ
April 1996 11.67 0.59 0.59 ææ
November 1999 15.25 0.60 0.60 ææ
December 2001 17.33 0.61 0.61 ææ
January 2003 18.42 0.61 0.61 0.0084 (1993-2003) æ
Lake Monoun: During-degassing
January 2004 19.42 0.53 0.52 ææ
January 2005 20.42 0.42 0.42 ææ
June 2006 21.92 0.43 0.42 ææ
January 2007 22.42 0.22 0.21 ææ
December 2007 23.33 0.11 0.10 æ0.098 (2003-2007)
January 2009 24.42 0.071 0.055 ææ
January 2011 26.42 0.041 0.036 æ0.005 (2009-2011)
March 2012 27.59 0.066 0.051 ææ
March 2013 28.59 0.074 0.059 ææ
Apr 2014 29.84 0.079 0.065 0.0048 (2011-2014) æ
Lake Nyos: Pre-degassing
November 1986 0.17 13.1 12.9 ææ
December 1988 2.33 13.3 13.3 ææ
November 1993 7.25 13.6 13.6 ææ
April 1998 11.67 14.1 14.0 ææ
November 1999 13.25 14.4 14.0 ææ
January 2001 14.42 14.8 14.6 0.12 (1986-2001) æ
Lake Nyos: During-degassing
December 2001 15.33 14.2 14.0 ææ
January 2003 16.42 13.1 13.1 ææ
January 2004 17.42 13.2 13.0 ææ
January 2005 18.42 12.3 12.6 ææ
January 2006 19.42 11.8 11.7 ææ
January 2007 20.42 11.6 11.4 ææ
January 2009 22.42 11.2 11.1 ææ
January 2011 24.42 10.0 9.7 æ0.46 (2001-2011)
March 2012 25.59 7.8 7.7 ææ
March 2013 26.59 6.6 6.5 ææ
March 2014 27.59 5.9 5.8 æ1.2 (2011-2014)
Table 3. Change with time in CO2 content at Lakes Monoun and Nyos during the last 28 years.
This table was revised from table 1 in Kusakabe et al. (2008).
CO2 removal rate was calculated using the CO2 content below layer II during the period shown in parentheses.
Data after 2011 were supplied by T. Ohba.
16 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
shoulder at a depth of ~63 m in 1986. The CO2(aq) pro-
files evolved with time, especially from 1986 to 1993.
The thickness of layers III and IV expanded with time,
supporting the hypothesis that CO2-rich recharge fluid
was added from the bottom. In December 2001 and
January 2003, the CO2 shoulder at a 58 m depth (157
mmol/kg) was very close to the CO2 saturation con-
centration (Duan and Sun, 2003) at a depth of 50 m.
Considering that the rate at which the shoulder was
rising was about 1 m/yr, saturation at a 58 m depth
could be reached in several years. The formation of
CO2 bubbles which could induce a limnic eruption
(Kozono et al., 2016) would have occurred soon after
2003 at Lake Monoun if no degassing operation had
been undertaken.
Figure 14b shows the temporal variation of CO2(aq)
profiles between November 1986 and January 2001 at
Lake Nyos. The general features of Fig. 14b are sum-
marized as: (i) CO2(aq) concentration was lowest in the
early days after the explosion; (ii) there was little
change with time at mid-depths (~50 m to 150 m); (iii)
the greatest change took place at a depth of >170 m,
where the CO2(aq) concentration at a given depth in-
creased significantly with time; and (iv) the CO2(aq)
concentration at the bottom-most water was almost
constant, near 350 mmol/kg since 1999. The constancy
of the bottom water CO2(aq) concentration was con-
firmed by later syn-degassing measurements. The
change in the bottom water is likely caused by the
gradual addition of recharge fluid having a CO2(aq)
concentration of ~350 mmol/kg. The CO2(aq) content
of the lake was calculated by integrating CO2(aq) pro-
files over the water column below layer II using the
bathymetry in Kling et al. (2005) under the assump-
tion that the horizontal distribution of CO2 was uni-
form, as deduced from the conductivity distribution
(see figure 8 of Kusakabe et al., 2008) and that CO2
loss through the upper chemocline was negligible.
Thus, the CO2 accumulation rate can be regarded as
the CO2 recharge rate. Accumulation of CO2 in layer
II, and in deeper layers, is tabulated in Table 3 for
Lakes Monoun and Nyos. Considering that the CO2(aq)
profile in October 1986 at Lake Monoun was estimated
in an indirect way (Kusakabe et al., 2008), the overall
rate of CO2 accumulation below the upper chemocline
was calculated using the 1993 to 2003 profiles. The
change in CO2(aq) content below layer II in the main
basin for the pre-degassing period (1993 to 2003) was
~80 (=610–530) Mmol, with a CO2 recharge rate of
8.4 Mmol/yr. Almost the same recharge rate of 8.2
Mmol/yr was reported by Kling et al. (2005) using their
own data obtained between 1992 and 2003 for Lake
Monoun. At Lake Nyos, the CO2(aq) content below layer
II steadily increased by ~1.7 Gmol until January 2001,
when permanent degassing started. The increase in the
CO2(aq) content can be translated into the CO2 recharge
Fig. 15. Evolution of syn-degassing CO2 profiles at (a) Lake Monoun (2003–2014) and (b) Lake Nyos (2001–2014). The
saturation of CO2 in water at 25∞C is shown by a dashed line. Recent data were added to figures 5 and 11 of Kusakabe et al.
(2008), and the figures were colored. Note that the CO2 concentrations in 2012, 2013 and 2014 in deep water at Lake Monoun
have increased, indicating a re-buildup of gas. CO2 profiles at Lake Nyos have steadily subsided. The highest CO2 concentra-
tion at the bottom water in 2014 was reduced to ~150 mmol/kg. See text for details.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 17
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
rate, which was 0.12 Gmol CO2/yr between Novem-
ber 1986 and January 2001 (Table 3). Again, the rate
is in good agreement with the value of 0.13 Gmol CO2/
yr given by Kling et al. (2005).
3-2. Syn-degassing evolution of CO2 content and
future prospect
The CO2(aq) profiles between 2001 and 2011 at both
lakes during syn-degassing are shown in Fig. 15. Gen-
erally speaking, degassing went smoothly, as illustrated
by the steady subsidence of the CO2(aq) profiles. This
resulted in a lowering of the fountain height at both
lakes (Fig. 16). The subsidence has continued up to
the present time; however, a buildup of CO2 has re-
sumed recently at Lake Monoun (see below). The over-
all shape of the profiles did not change with degassing,
showing that only bottom water and dissolved CO2(aq)
were removed without causing any effect on the strati-
fication of the lake water. At Lake Monoun (Fig. 15a),
the highest CO2(aq) concentration at the bottom de-
creased to 80 mmol/kg, and the thickness of layer III
reduced to ~20 m in 2009, and further reduced to ~70
mmol/kg and ~15 m, respectively, in 2011 (Kusakabe
et al., 2011). In 2009, two of three degassing pipes
stopped working completely, and the other pipe issued
only a weak bubbly flow (Fig. 16f). Thus, it can be
said that the degassing pipes at Lake Monoun have al-
most lost their gas self-lift capability. Moreover, re-
cent observations (2011–2014) show that CO2 concen-
trations below 80 m and the layer III thickness are in-
creasing (Fig. 15a), clearly indicating that natural CO2
recharge into Lake Monoun still continues. This con-
firms our prediction that CO2 re-buildup is inevitable
if lake degassing stops. On the basis of a geochemical
study on the generation of CO2 in the Nyos mantle,
Aka (2015) has suggested that CO2 will be continu-
ously supplied into the lake for a geologically long time
in the future. This view may also apply to Lake
Monoun. In order to avoid gas re-buildup and to make
the lake continualy safe, Yoshida et al. (2010) sug-
gested continuously removing the bottom water that
contains the CO2 at a significantly high concentration.
The installation of such a bottom water removal sys-
tem was undertaken at Lake Monoun in December 2013
(Yoshida et al., 2016), and details of the system are
Fig. 16. The reduced height of fountains from degassing pipes at Lakes Nyos and Monoun. Three degassing pipes were in
operation at Lake Nyos as of March 2014 (c). Gas self-lift capability was lost in January 2009 at Lake Monoun, leaving a
weak bubbly flow from the neck of the pipe (f).
18 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
described in Section 5.
At Lake Nyos (Fig. 15b), CO2(aq) profiles subsided
steadily until January 2011, resulting in a very thin
layer III by that time. As two more degassing pipes
with a greater diameter (25.7 cm I.D.) were installed
in December 2011–March 2012, the degassing rate was
greatly enhanced, resulting in a rapid decrease of CO2
concentration in deep water in the subsequent years
(2011–2014). We can expect that most of the CO2-rich
bottom water will disappear from Lake Nyos in sev-
eral years from now and that the gas self-lift capabil-
ity will be lost as in Lake Monoun.
Using CO2(aq) profiles and lake bathymetry (Kling
et al., 2005), the amount of CO2(aq) dissolved below
the upper chemocline (layers II, III and IV) was calcu-
lated as a function of time since the limnic eruption at
both lakes (Fig. 17). The amount of dissolved CO2 in
Lake Monoun (Fig. 17a) increased steadily at a rate of
8.4 Mmol/yr, reaching a maximum value of 610 Mmol
in January 2003, shortly before the degassing opera-
tion started. Degassing was effective, reducing the
amount of dissolved CO2 at a mean gas removal rate
of 98 Mmol/yr between January 2003 and December
2007 (see Table 3). This rate is approximately 12 times
greater than the natural recharge rate as shown by the
sharp slope (Fig. 17a). The installation of two addi-
tional pipes in April 2006 accelerated the gas removal
rate. In January 2009, the system had almost lost its
gas self-lifting capability, resulting in a reduction of
gas removal rate to only 5 Mmol/yr in January 2011
(Table 3), although a very weak flow of bubbly water
from one of the three pipes was still visible. At that
time, the amount of CO2(aq) dissolved in deep water
was 55 Mmol, or only 9% of the maximum value ob-
served in 2003 (Kusakabe, 2015). Based on these ob-
servations, it can be said that Lake Monoun has been
made safe. However, the tailing-off of the CO2 con-
tent after 2009 (Fig. 17a) implied that a buildup of
CO2 is inevitable at Lake Monoun if the natural re-
charge of CO2 continues at the previously estimated
Fig. 17. (a) Change with time in the CO2 content at Lakes Monoun (a) and Nyos (b). Modified from figures 6 and 12 of
Kusakabe et al. (2008) to which recent data were added. The blue and red circles denote pre-degassing and syn-degassing
evolution, respectively. Note that the CO2 content at Lake Monoun started to rise after 2011 at a rate of ~4.8 Mmol/yr,
approximately half of the natural CO2 recharge rate of 8.4 Mmol/yr estimated from the pre-degassing data. The degassing rate
at Lake Nyos was accelerated after installation of 3 pipes. The CO2 content at Lake Nyos will attain a minimum in several
years time.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 19
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
rate. Indeed, the re-buildup of CO2 became obvious
after March 2012 (Fig. 17a). Using the data between
2010 and 2014, the rate of gas re-buildup is calculated
to be ~4.8 Mmol/yr which is about half of the CO2
recharge rate of 8.4 Mmol/yr calculated from the pre-
degassing data (Table 3), although we need to accu-
mulate more recent data for a more reliable determi-
nation of the rate of gas re-buildup. The current amount
of total dissolved CO2 in Lake Monoun is 79 Mmol
(as of April 2014) which is ~13% of the maximum pre-
degassing value recorded in 2003. It may take another
~100 years to reach the pre-degassing situation if the
current rate of gas re-buildup remains unchanged. For
these reasons, it is essential to continue monitoring the
lake on a regular basis.
The evolution of CO2 content over time since 1986
at Lake Nyos is shown in Fig. 17b. The gas removal
rate by a single pipe (0.46 Gmol/yr) is about four times
greater than the natural recharge rate of 0.12 Gmol/yr.
At this removal rate, however, it would take another
20 years or so to remove all the gas from the lake. For-
tunately, using funds from the Government of
Cameroon and UNDP, two additional degassing pipes
were installed in early 2011. Since pipes with a greater
diameter (25.7 cm) were used and the water intake
depth was increased to close to the bottom for the ad-
ditional pipes, the rate of gas removal greatly increased
to 1.2 Gmol/y (Table 3). It is hoped that most of the
remaining gas will be removed within the next 5 years
or so. At the last stage of the degassing operation, the
rate of gas removal will decrease due to a lower CO2
concentration at the intake depth. This will lead to gas
re-buildup, as we have seen at Lake Monoun. Thus, a
system to pump up CO2-rich bottom water needs to be
set up after the current degassing system has lost its
gas self-lifting capability.
3-3. Hydrogen, oxygen, carbon, and noble gas iso-
topic signatures
The hydrogen and oxygen isotopic ratios of Lake
Nyos waters were first reported by Kling et al. (1987).
The data for Lakes Nyos, Monoun and Wum (a crater
lake near Lake Nyos) were later added by Kusakabe et
al. (1989) and Nagao et al. (2010). The isotopic deter-
mination was intended to find any input of volcanic
gases into Lake Nyos, for volcanic gases are usually
characterized by high
d
18O signatures, but the data did
not indicate any volcanic input. The
d
D and
d
18O val-
ues from Lake Nyos plot close to the Global Meteoric
Water Line (Craig, 1961; Rozanski et al., 1993) when
combined with the data for rain water, groundwater,
and surface water recently collected in the Lake Nyos
catchment area as shown in Fig. 18, although the lake
Fig. 18.
d
D-
d
18O values of monthly collected rain waters (dark asterisks) and ground waters (circles and triangles) sampled in
the vicinity of Lake Nyos are shown (Kamtchueng et al., 2015a). Those of Lake Nyos waters (Nagao et al., 2010) are also
included. All values are consistent with the global meteoric water line of Rozanski et al. (1993), although the Lake Nyos
waters are plotted slightly upper-right of the cluster. It may suggest that the lake water is not recharged by recent groundwater.
20 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
water is slightly more enriched in heavy isotopes than
the ground and surface waters (Kamtchueng et al.,
2015a).
Noble gas information was used to constrain the ori-
gin of CO2 dissolved in the lakes, for noble gases do
not react with rocks and waters on the way to the
Earth’s surface. Figure 19 shows the relationship be-
tween 3He/4He and 40Ar/36Ar ratios for Lakes Nyos and
Monoun gases in 1999. Coupled with the
d
13C values
of –3.3 ~ –3.4‰ (relative to Vienna Pee Dee Belemnite,
VPDB) for Lake Nyos and –6.8‰ for Lake Monoun,
the helium and argon isotopic ratios suggest a strong
affinity of the dissolved gases with a magmatic source
(e.g., Kusakabe and Sano, 1992). Nagao et al. (2010)
reported more precise data using air-contamination free
samples. The 3He/4He ratios in the gases in the Lake
Nyos deep waters are ~5.7 Ratm, where Ratm is the at-
mospheric ratio of 1.4 ¥ 10–6. The Lake Nyos 3He/4He
ratios are lower than the typical mantle values of 7~9
Ratm for depleted mantle producing Mid-Oceanic Ridge
basalts (MORBs) (Graham, 2002). The reasons why
the 3He/4He ratios of Lake Nyos are lower than the
mantle values are related to the sub-lithospheric struc-
ture beneath the Cameroon Volcanic Line (CVL) as
discussed later (Section 6). Halliday et al. (1988) re-
ported variations in the radiogenic isotopic ratios (Pb,
Nd, and Sr) of volcanic rocks along the CVL, where
the highest 206Pb/204Pb and 208Pb/204Pb ratios were
found at the oceanic and continental boundary. Barfod
et al. (1999), Aka (2000) and Aka et al. (2004) pub-
lished a detailed study of noble gases in basalts and
xenoliths from CVL volcanic rocks, showing a sym-
metrical distribution of 3He/4He ratios along the CVL
(Fig. 20). The lowest 3He/4He ratios (4.5 ¥ 10–6 or ~3
Ratm) were found at Etinde, a small volcano next to
Mt. Cameroon, located at the oceanic and continental
boundary. The 3He/4He ratios become close to the
MORB values (7~9 Ratm) as we go away from the oce-
anic and continental boundary towards both ends (Aka
et al., 2004). This symmetric isotopic variation was
explained as reflecting the geochemical characteristics
of the mantle, or the continental lithosphere underneath
the boundary which is of a HIMU character (Halliday
et al., 1988). HIMU is geochemical jargon for “high-
m
” with
m
defined as the ratio of 238U/204Pb. HIMU
mantle is characterized by an enrichment in U and Th,
the parent elements of radiogenic Pb and He. Melts
derived from this HIMU mantle are postulated to have
been emplaced beneath the oceanic and continental
boundary. Thus, rocks at the oceanic and continental
boundary are high in 206Pb/204Pb and low in 3He/4He
ratios. Mineral separates from rocks around Lake Nyos
(clinopyroxene and amphibole in xenolith) have 3He/
4He ratios of 6.7~7.0 Ratm (Aka et al., 2004), slightly
lower than the typical MORB values, implying a small
degree of the HIMU character of the magma source
beneath Lake Nyos.
The deep water of the lake has even lower values of
Fig. 19. Relationship between 3He/4He (in Ratm) and 40Ar/36Ar ratios of waters in Lakes Nyos and Monoun. Ratm is a 3He/4He
ratio of sample relative to that of air (=1.4 ¥ 10–6). The plots show a mixing of magmatic gases and the atmosphere (green
cross). See text for a discussion. Data from Nagao et al. (2010).
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 21
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
~5.7 Ratm, as was stated above. This low ratio may mean
that He in deep water was originally derived from
magma generated from a slightly HIMU-type mantle,
but acquired radiogenic 4He on the way from the source
magma to the sub-lacustrine fluid reservoir during the
passage of the magmatic fluid through granitic base-
ment rocks. Nagao et al. (2010) reported the distribu-
tion of isotopic ratios of not only He but also Ne, Ar,
Kr, Xe and C in Lakes Nyos and Monoun waters col-
lected at closely separated depths. They stressed the
importance of samples that were free from air-contami-
nation, because noble gas concentrations, especially
those of Ar and Ne, are so low in gases exsolved from
CO2-rich waters compared to air so that any samples
exposed to the atmosphere during sampling, or stor-
age in an improper way, are not good for analysis. Sam-
ples from Lake Nyos (January 2001) were collected
using the “Flute de Pan” which had been deployed by
the French scientific team. This consisted of 11 plastic
hoses having an outside diameter of 15 mm with dif-
ferent intake depths (83–210 m) (Fig. 21). CO2-rich
gas spouting out of a given hose was collected in a
glass bottle using an inverted funnel placed in a bucket.
Although slight contamination of air dissolved in the
water was still suspected to some extent, especially for
heavy noble gases, this sampling method was found to
Fig. 20. Symmetrical distribution of 3He/4He and 206Pb/204Pb ratios of rocks along the CVL as a function of distance from
Annobon. This distribution suggests a contribution of the HIMU mantle for magma genesis in the oceanic and continental
boundary volcanoes (Halliday et al., 1988; Aka et al., 2004). Oceanic sector volcanoes are Annobon (AN), São Tomé (ST) and
Principé (PP). Ocean/continent boundary volcanoes are Bioko (BK), Etinde (ETD) and Mount Cameroon (MC). Continental
sector volcanoes are Manengouba (MB), Bambouto (BT), Oku (OK), and Ngaoundere (ND).
Fig. 21. Sampling water and gas using the “Flute de Pan” (a). Water and gas gushing out of 11 plastic hoses (O.D. of 15 mm)
with different intake depths were collected (b).
22 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
be promising. In the December 2001 sampling, a plas-
tic hose method, essentially the same as the Flute de
Pan method, was adopted. A single plastic hose (12
mm I.D.) was deployed initially to the bottom, followed
by pulling it upward little by little to a desired depth.
Exsolved CO2 gas was directly allowed to pass through
a sampling bottle made of uranium glass that has a low
He diffusivity. With this method, Nagao et al. (2010)
were able to collect air-contamination free samples.
This method was later modified to measure the total
gas concentration on site (Yoshida et al., 2010).
Figures 22a, b illustrate the profiles of He, Ne, Ar,
Kr and Xe in water, measured in 2001, at Lake Monoun
and Lake Nyos, respectively. Except for He, they show
roughly constant concentrations with respect to depths
below 80 m at Lake Nyos, and 50 m at Lake Monoun.
The Ne, Ar, Kr and Xe concentrations are up to sev-
eral times lower than those in air saturated water
Fig. 22. Depth profiles of noble gas concentrations in water (10–6 ccSTP/gwater) measured in 2001 at Lake Monoun (a) and
Lake Nyos (b). Noble gas concentrations in air saturated water (ASW) at 30∞C (table 2 in Kipfer et al., 2002) are also shown
by arrows for comparison. Modified from figures 1 and 2 of Nagao et al. (2010).
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 23
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
(ASW). Depth profiles of the 3He/4He ratio for Lakes
Nyos and Monoun are presented in Fig. 23. The data
published by earlier workers are in the same range
(Kling et al., 1987; Sano et al., 1987, 1990; Kusakabe
and Sano, 1992). Generally speaking, the 3He/4He ra-
tios are almost constant in the depth range of 80–210
m at Lake Nyos, and 40–100 m at Lake Monoun. The
3He/4He ratio approaches the atmospheric value in
waters shallower than 80 m and 40 m for Lakes Nyos
and Monoun, respectively. High 4He/20Ne ratios up to
~1500, as shown in Fig. 22, support the premise of
magmatic gas input to the lake as inferred by high 3He/
4He ratios.
Neon isotopic ratios are presented in Fig. 24. Com-
pared to atmospheric Ne, small excesses of both the
20Ne/22Ne and 21Ne/22Ne ratios are observed. Most data
points for both lakes lie on the MORB line connecting
atmospheric Ne and mantle Ne as reported by
Staudacher and Allègre (1988). The data clearly indi-
cate the presence of mantle Ne in the lakes, and are
consistent with the conclusions of Barfod et al. (1999)
and Aka et al. (2004) that the CVL mantle contains
MORB-like Ne.
Argon isotopic ratios are presented in Fig. 25. The
40Ar/36Ar ratios for all samples are higher than the at-
mospheric value of 296, but much lower than the esti-
mated value of >1650 for Ar in the upper mantle be-
neath the CVL (Barfod et al., 1999). This means that
magmatic fluids containing mantle Ar mixed with at-
mospheric Ar on the way to the surface such as in a
sub-lacustrine fluid reservoir. The contribution of man-
tle Ar to the sub-lacustrine Ar may be less than 20%
assuming that the mantle Ar has a 40Ar/36Ar ratio of
>1650. This is consistent with the conclusion derived
from the Ne signature (Fig. 24), although the contri-
bution of mantle Ne to the sub-lacustrine Ne may be
~6%. At Lake Nyos, the highest 40Ar/36Ar ratio, of
about 600, was found at the bottom (210 m). The 40Ar/
36Ar profile in January 2001 decreased gradually to-
wards the surface approaching the atmospheric ratio,
but it showed a zigzag pattern below the lower
chemocline at ~180 m with a second maximum value
of 480 at 190 m. The zigzag 40Ar/36Ar profile disap-
peared in December 2001 with consistent ratios around
530 below 190 m. This may have resulted from verti-
cal mixing in this depth range caused by degassing,
because water was pumped out by the degassing pipe
from an intake depth of 203 m. A tendency for such
homogenization was also observed in the water tem-
perature and electric conductivity at the correspond-
ing depths (Kusakabe et al., 2008), although they were
less clear than the noble gas profiles. At Lake Monoun,
the 40Ar/36Ar ratios were in a narrow range of about
470 between 60 m and 100 m (bottom) (Fig. 25). The
ratios are lower than those in the deep waters of Lake
Nyos, suggesting that the contribution of atmospheric
Ar to the magma-originating gases at Lake Monoun is
greater than that at Lake Nyos.
The characteristic features of noble gases observed
at Lake Nyos can be summarized as follows (Nagao et
al., 2010): (i) Helium in the lake water derived origi-
nally from the mantle where 3He/4He ratios of ~7 Ratm
Fig. 23. 3He/4He ratios as a function of depth at Lakes Nyos and Monoun in 1999 and 2001. Modified from figure 4 of Nagao
et al. (2010).
24 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Fig. 25. Depth profiles of 40Ar/36Ar in Lakes Nyos and Monoun in 2001. Chemoclines were taken from Kusakabe et al. (2008).
Modified from figure 6 of Nagao et al. (2010).
Fig. 24. 20Ne/22Ne versus 21Ne/22Ne plot for samples collected in 2001 at Lakes Nyos and Monoun. The “mass fractionation
line” indicates the isotopic trend of atmospheric Ne due to mass fractionation. The dashed line heading to MORB represents
a mixing line between atmospheric Ne and Ne in MORB or the upper mantle (Ballentine et al., 2005). Modified from figure
5 of Nagao et al. (2010).
are found in mantle xenoliths (Aka et al., 2004), but,
on its way to the surface, approximately 20% of radio-
genic 4He that accumulated in crustal rocks was ad-
mixed to give ratios of ~5.7 Ratm, probably in the sub-
lacustrine region. (ii) The observed 40Ar/36Ar ratios of
450–550 are also explained by the addition of atmos-
pheric Ar (40Ar/36Ar = 296) carried by groundwater to
mantle-originating Ar (40Ar/36Ar >1650, Barfod et al.,
1999) on the way to the lakes. The most likely source
of Ar to reduce the mantle 40Ar/36Ar ratio is atmos-
pheric Ar-bearing groundwater. (iii) Ne in the lakes
may be a mixture of atmospheric Ne and a small amount
of MORB-like Ne from the mantle. The observed He,
Ne and Ar isotopic ratios in lake waters can be best
explained by mixing between two noble gas reservoirs,
i.e., air dissolved groundwater and the mantle. It is
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 25
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
conceivable that the mantle-derived gases with the
addition of radiogenic 4He from crustal rocks and at-
mospheric Ar and Ne carried by groundwater are fi-
nally homogenized in the sub-lacustrine reservoir. Note
that the contribution of atmospheric He to deep lake
water, if any, is difficult to find, since the He concen-
tration in deep lake water is more than 3 orders of
magnitude higher than that in ASW, whereas the con-
tribution of the other noble gases is more easily dis-
cernible because of their similar concentrations in deep
lake water and ASW (see Fig. 22).
As stated previously, the greatest chemical change
in Lake Nyos took place at depths greater than 180 m.
The CO2 profiles (1986–2001) in the depth range 160–
210 m are enlarged in Fig. 26a. This shows that the
increase of CO2 concentration in the deep water of Lake
Nyos after the 1986 limnic eruption resulted from wid-
ening of CO2-rich water leading to the formation of a
clear lower chemocline at the top of the CO2-rich wa-
ter. The 3He concentration observed in 2001 in the same
depth range was compared with the CO2 profiles (Fig.
26). The 3He profile was obtained from the 4He pro-
file (Fig. 22) and the 3He/4He profile (Fig. 23). It
should be noted that the 3He concentration below 160
m in January 2001 and December 2001 shows a sharp
maximum at around 188 m with a concentration up to
9.1 ¥ 10–10 ccSTP/g-water (December 2001) (Fig. 26b).
The 4He concentrations have a pattern very similar to
those of 3He in the same depth range, because of the
almost constant 3He/4He ratios (Fig. 23), although the
4He concentrations are not graphically shown. Between
190 and 210 m, the 3He concentrations are nearly con-
stant at ~5 ¥ 10–10 ccSTP/g-water. The 3He concentra-
tion gradually decreases as the depth decreases (layer
II).
The C/3He ratios of volcanic fluids have been widely
used to constrain magma sources. The C/3He ratios of
MORB glasses are shown to be fairly constant at 0.20
(±0.05) ¥ 1010, suggesting that the source region of
MORB in the upper mantle has little variation in the
C/3He ratio (Marty and Jambon, 1987). The ratios for
volcanic gases from subduction volcanism, however,
have been found to be significantly greater than the
MORB values, i.e., 0.7 ¥ 1010 ~ 3 ¥ 1010. These high
ratios, coupled with
d
13C values, indicate the existence
of recycled carbon (marine carbonates, slab carbon-
ates and/or organic materials) in subduction zone mag-
mas (Sano and Williams, 1996, and references therein).
Figure 27 shows the C/3He ratios in the depth range
160–210 m in Lake Nyos. The C/3He ratios range from
0.5~1.7 (¥ 1010). These values are higher than the man-
tle values of ~0.2 ¥ 1010. It is interesting to note that
the C/3He ratios in waters below the lower chemocline
are significantly high at around 1.6 ¥ 1010, and sharply
decrease to 0.5 ¥ 1010 above the lower chemocline.
Thus, the behavior of CO2 and 3He are decoupled be-
Fig. 26. (a) “Inflating” CO2 profiles in the deep water of Lake Nyos in the depth range of 160–210 m between 1986 and 2001.
(b) 3He profile in the same depth range observed in 2001. Note the sharp maximum of 3He concentration at 188 m where a
chemocline (dashed line) existed in 2001. Modified from figure 14 of Volcanic Lakes (Dmitri Rouwet, Bruce Christenson,
Franco Tassi, Jean Vandemeulebrouck, eds.), Evolution of CO2 content in Lakes Nyos and Monoun, and sub-lacustrine CO2-
recharge system at Lake Nyos as envisaged from CO2/3He ratios and noble gas signatures, 2015, pp. 427–450, Kusakabe, M.,
„ Springer-Verlag Berlin Heidelberg 2015 with permission of Springer.
26 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Fig. 27. C/ 3He atomic ratios observed in the depth range of 160–210 m at Lake Nyos in 2001. The C/3He ratios were calculated
from the CO2 and 3He profiles shown in Fig. 26. A clear difference is seen across the chemocline (dashed line). Modified from
figure 15 of Volcanic Lakes (Dmitri Rouwet, Bruce Christenson, Franco Tassi, Jean Vandemeulebrouck, eds.), Evolution of
CO2 content in Lakes Nyos and Monoun, and sub-lacustrine CO2-recharge system at Lake Nyos as envisaged from CO2/3He
ratios and noble gas signatures, 2015, pp. 427–450, Kusakabe, M., „ Springer-Verlag Berlin Heidelberg 2015 with permis-
sion of Springer.
Fig. 28. (a) Change with time of the CO2/3He ratio in fumarolic gases from Mammoth Mountain in the Long Valley caldera,
California (1988–1998). (b) Change with time of the He concentration in the same gases. Data of Sorey et al. (1998) were
used.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 27
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
low and above the chemocline. The cause(s) of the
decoupling may be explained by the underplating of
the recharge fluid from the bottom that is character-
ized by different C/3He ratios. By “underplating”, I
meant that the recharge fluid is added to the bottom-
most water from beneath. It is possible that the ratio
was low before the limnic eruption and high after the
limnic eruption. At the time of the limnic eruption, the
lake was not completely mixed, suggesting that deep
water still contained a large fraction of “pre-eruption”
water (Giggenbach, 1990; Tietze, 1992; Evans et al.,
1994) which may have been proportionally higher in
He and lower in CO2 concentrations with the CO2/3He
ratio of ~0.5 ¥ 1010. Recharge fluids entering the lake
after the eruption may have a CO2/3He ratio of ~1.6 ¥
1010. This interpretation implies that the CO2/3He ra-
tio in the recharge fluids may vary with time and has
changed from low to high values with time. A change
with time in the CO2/3He ratio has been observed in
fumarolic gases from Mammoth Mountain in the Long
Valley caldera, California, where “tree-kill” took place
due to an anomalous discharge of magmatic CO2 into
soils (Farrar et al., 1995; Sorey et al., 1998). This
anomalous CO2 discharge was induced by an episode
of shallow dyke intrusion beneath Mammoth Moun-
tain in 1989–1990. The CO2/3He ratios of the fumarolic
gases there changed from ~0.3 ¥ 1010 to 1.6 ¥ 1010 in
about 10 years (Fig. 28). The change was caused by a
trend of decreasing He concentration and little change
in the concentration of CO2, which is a major compo-
nent of the gases. The 3He/4He ratios stayed at around
5.5 Ratm with a few exceptions. These observations
indicate that the above geochemical parameters (CO2/
3He ratio and He concentration) that carry information
about magmatic fluids can change within a geologi-
cally very short period of time, i.e., in the order of 10
Fig. 29. Schematic presentation of the sub-lacustrine fluid reservoir which is encircled by a green circle. The geological cross-
section of Lake Nyos was taken from Lockwood and Rubin (1989). Blue arrows indicate the possible flow of groundwater,
and red arrows indicate a magmatic fluid coming from the magma underneath. Noble gas and carbon isotopic ratios of respec-
tive reservoirs are shown. Modified from figure 16 of Volcanic Lakes (Dmitri Rouwet, Bruce Christenson, Franco Tassi, Jean
Vandemeulebrouck, eds.), Evolution of CO2 content in Lakes Nyos and Monoun, and sub-lacustrine CO2-recharge system at
Lake Nyos as envisaged from CO2/3He ratios and noble gas signatures, 2015, pp. 427–450, Kusakabe, M., „ Springer-Verlag
Berlin Heidelberg 2015 with permission of Springer.
28 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
years, at a single volcanic system. Thus, it is conceiv-
able that the decoupling of CO2 and 3He observed at
Lake Nyos after the limnic eruption was caused by the
addition of “recent” recharge fluids that were charac-
terized by relatively low 3He concentrations. From the
foregoing discussions based on noble gas signatures
and C/3He ratios, we can envisage the sub-lacustrine
CO2-recharge system at Lake Nyos to be as shown in
Fig. 29.
4. Limnic eruption, models, triggers and cyclicity
4-1. Models
Many hypotheses have been put forward to explain
why the limnic eruption occurred. Sigurdsson et al.
(1987) proposed that a landslide slumped into deep
water, pushed CO2-rich water up and induced the 1984
limnic eruption at Lake Monoun. The same idea was
also suggested for the 1986 Lake Nyos event (Kling et
al., 1987). Tietze (1987) suggested supersaturation of
dissolved CO2 just below the shallowest chemocline
(~8 m depth in 1986) to be the main cause of the erup-
tion. The strong density stratification of this layer
worked as a lid for rising gases, inhibiting them from
penetrating this density divide. The supersaturation that
followed led to the exsolution of gases to form a foun-
tain. This process was self-intensified and deeper wa-
ter was steadily degassed in turn. Since the water from
the fountain was cooler than the surface water, it sank
around the fountain, forming a cylindrical “density
wall”. This wall limited lake-wide exsolution of gases,
leaving CO2 dissolved in deep water (>150 m?) intact
during the eruption. Assuming that Lake Nyos was iso-
thermal and fully saturated with CO2, Kanari (1989)
presented a fluid-dynamics model to explain how the
limnic eruption proceeded. In his model, degassing
started from the bottom but was confined to a limited
area at the surface. Circulation of water was confined
in small cells that stacked at various depths. Accord-
ing to this model, stratification within the lake was
hardly affected. Kanari estimated that (i) the released
gas volume (0.68 km3) was the difference between the
saturation and the CO2 profile observed in 1986 by
Kusakabe et al. (1989), (ii) the maximum height of the
gas cloud was 110 m, and (iii) the speed of the gas
cloud running down the valley was 19 m/s. However,
later observations indicated that full CO2 saturation
over the entire lake was unlikely (Kusakabe et al.,
2008).
Obviously, any degassing model depends on a knowl-
edge of the pre-eruption distribution of CO2 in the lake.
Evans et al. (1994) proposed a model based on a linear
pre-eruption relationship between CO2 and TDS (total
dissolved solids) at Lake Nyos using water chemistry,
CTD measurements, gas analyses and tritium profiles
obtained between 1987 and 1992. A linear relationship
between tritium and TDS was interpreted to reflect the
destruction of the pre-existing gradient at mid-depth
during the eruption, suggesting that CO2 exsolved from
deep water. In their model, the upper chemocline was
placed at ~50 m depth, similar to the chemocline depth
observed in January 2001, 14 years after the limnic
eruption, and just prior to the initiation of artificial
degassing (see Fig. 9). Some triggers, such as a com-
bination of seasonal decline in the water column sta-
bility, landslide and/or seiche, pushed water upward at
a layer around the chemocline to the CO2 saturation
depth. Bubble formation then followed and relatively
quiet degassing continued. A local reduction in the
hydrostatic pressure beneath the release area created a
rising column of shallow, slightly gassy water. This
was followed by mixing with pre-release surface wa-
ter (low TDS) to form the surface water that was ob-
served soon after the limnic eruption. The base of the
column became slowly deeper, bringing CO2-rich, more
saline deep water upward. When the base of the col-
umn reached the deeper chemocline, below which CO2
and TDS concentrations were much higher, gas release
became more violent and created wave damage along
the lake shore such as the flattening of vegetation and
the passing of water over an 80-m-high promontory in
the southern part of the lake. The duration of this vio-
lent fountaining was short (<1 min), and the amount
of CO2 released was estimated to be 6.3 Gmoles. This
scenario is consistent with the testimonies of survivors.
Giggenbach (1990) proposed that the gas release at
Lake Nyos was triggered by a climatic factor. The de-
scent of a parcel of unusually cold rain water (18.5∞C)
pushed initially CO2-rich shallow water upward. The
uplift of the CO2-rich water above the saturation depth
induced bubble formation which accelerated upward
movement by a reduction of density, leading to the for-
mation of a convecting water flow that entrained
deeper, more CO2-enriched water, and, finally, to the
limnic eruption. Less-dense degassed waters accumu-
lated at the surface, making it difficult for deeper CO2-
rich water (>100 m) to reach the surface, thus termi-
nating the eruption. Deep water CO2 was therefore left
almost intact. The amount of CO2 released was esti-
mated at 5.4 Gmoles.
In contrast to the previous models for the cause of
the limnic eruption, spontaneous exsolution of dis-
solved gases has been suggested by Kusakabe et al.
(2008) and Kusakabe (2015). In this scenario, atten-
tion was paid to the pre-degassing evolution of dis-
solved CO2 at Lake Monoun (see Fig. 14a) which in-
dicated that CO2(aq) profiles evolved with time and that
CO2-rich layers below the lower chemocline (layers
III and IV) widened due to the continuing recharge of
CO2-charged fluid from beneath. Note that the CO2(aq)
concentration in water below layer III was constant at
~150 mmol/kg. In January 2003, just prior to the ini-
tiation of the degassing operation, the CO2(aq) concen-
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 29
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
tration immediately below the chemocline at the bound-
ary between layers II and III was very close to satura-
tion. If no degassing operation were undertaken, the
saturation of CO2(aq) would have been attained at that
depth in a short time (within several years), and bub-
ble formation would have followed by additional in-
put of the recharge fluid. Thus, at Lake Monoun an-
other limnic eruption could have occurred spontane-
ously within several years after 2001, if any external
trigger had been introduced to this critical situation.
4-2. Spontaneous eruption hypothesis
The evolution of pre-degassing CO2 profiles at Lake
Monoun gives a clue to estimate a pre-eruptive CO2
profile at Lake Nyos. It is conceivable that it was simi-
lar in shape to the 2001–2003 profiles at Lake Monoun
(see Fig. 14a). It is interesting to note that the CO2
profiles at the deep layer of Lake Nyos (>180 m in
1999 and 2001; Fig. 14b) was developing in a way
similar to that observed at Lake Monoun. The thick-
ness of CO2-rich water close to the bottom kept in-
creasing after the 1986 eruption till 2001 due to the
addition of the recharge fluid from beneath. The CO2(aq)
concentration below 195 m reached 350 mmol/kg in
1999. This concentration remained unchanged until
January 2011 down to the bottom (Kusakabe et al.,
2008; Kusakabe, unpublished data). This observation
suggests that the CO2(aq) concentration of the recharge
fluid is constant at ~350 mmol/kg. If no degassing was
undertaken, and if the natural recharge of CO2 contin-
ued as before, the thickness of the bottom CO2-rich
water would have continued to increase, and the top
level of the CO2-rich layer could have eventually
reached saturation at some shallower depth. This specu-
lation is schematically presented in Fig. 30. In this
model, the pre-eruption profile, shown as “Before
1986”, has a shoulder that touches the saturation curve
at a depth of ~110 m. A limnic eruption would take
place spontaneously, releasing the dissolved gases to
the atmosphere, resulting in a CO2 profile shown as
“November 1986” in Fig. 30 (process 1). The observed
evolution of the CO2 concentration between Novem-
ber 1986 and January 2001 is shown as “process 2” in
Fig. 30. If no degassing took place, and if the natural
recharge of CO2 continued as before, the CO2(aq) pro-
file would have shifted upward following “process 3”,
and would eventually have touched the saturation
curve. Saturation is a necessary condition, but may not
be a sufficient condition, for a limnic eruption to take
place. Rising bubbles may re-dissolve in under-satu-
rated water during ascent. However, if sufficient CO2
flux is given, the bubbles can reach the surface, possi-
bly leading to a limnic eruption. Based on the above
model, a numerical approach to the recurrence of a
future limnic eruption was made by Kozono et al.
(2016). They demonstrated that a plume of bubbles
generated from a growing CO2-saturated surface (the
top of the “Before 1986” curve in Fig. 30) can reach
the lake surface with a high flux of CO2, i.e., limnic
eruption, if any external forcing triggers bubble for-
mation at the growing CO2-saturated surface. The trig-
ger may be an instability caused by double diffusive
convection (Schmid et al., 2004), or a seiche near the
CO2-saturated surface where the density gradient is
strong.
If our model is correct, the difference between the
pre- and post-eruption profiles integrated over the lake
volume gives the amount of gas released at the time of
the eruption, which was calculated to be ~14 Gmol or
0.31 km3 (at STP). This value is greater than the esti-
mate of 0.14 km3 by Evans et al. (1994) by a factor of
~2, but significantly smaller than earlier estimates
(0.7~1 km3) by Faivre Pierret et al. (1992), and Kanari
(1989). The estimated amount of CO2 released obvi-
ously depends on the assumptions involved. As long
as the lake receives a continual natural recharge of CO2,
limnic eruptions can occur repetitively (Tietze, 1992),
but may not be regular as described in the model of
Fig. 30. A model of the spontaneous limnic eruption at Lake
Nyos. An assumed pre-eruption CO2 profile is shown by red
small open circles as “Before 1986”. After the eruption, the
CO2 profile turned to the post-eruption profile shown as
“Nov. 1986” (process 1). It evolved to the January 2001 pro-
file (blue) in 15 years (process 2). If the natural recharge
continues, the January 2001 profile may “recover” the pre-
eruption situation (process 3). Modified from figure 9 of
Volcanic Lakes (Dmitri Rouwet, Bruce Christenson, Franco
Tassi, Jean Vandemeulebrouck, eds.), Evolution of CO2 con-
tent in Lakes Nyos and Monoun, and sub-lacustrine CO2-
recharge system at Lake Nyos as envisaged from CO2/3He
ratios and noble gas signatures, 2015, pp. 427–450,
Kusakabe, M., „ Springer-Verlag Berlin Heidelberg 2015
with permission of Springer.
30 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Chau et al. (1996), which considered a possible varia-
tion in the rate of the natural recharge of CO2. How-
ever, if the conceptual model shown in Fig. 30 is cor-
rect, it would take ~100 years to attain the pre-erup-
tion CO2 level shown by the “Before 1986” curve start-
ing from the curve “November 1986” assuming a con-
stant CO2 recharge rate of 0.12 Gmol/year (process 3).
4-3. Repetitive nature of a limnic eruption
The above model implies that the time of repetition
of a limnic eruption is ~100 years. A possibility of cy-
clic gas bursts from lakes which are charged by a gas
influx from the lake bottoms was also pointed out by
Tietze (1992). He argued that dissolution of CO2(aq)
will inevitably create a stratification of the lake be-
cause the density of CO2-containing water is higher
than that of pure water, due to a small partial volume
of dissolved CO2(aq) in water (Ohsumi et al., 1992),
and that the stratification will limit upward gas trans-
port, leading to an accumulation of the gas below the
stratified layers. If limnic eruptions take place repeti-
tively on a timescale of ~100 years, evidence of past
eruptions might be found in geological records and
local documents. Unfortunately, no geological evidence
has been recorded, and no such written documents are
known to exist in the Nyos-Monoun areas. However,
Shanklin (1989, 1992, 2007) published interesting
folklores that are common in the grassfields of west-
ern Cameroon; the Kom story and the Oku story. The
folklores are suggestive of limnic eruptions that took
place in the past. The following paragraphs (shown in
italic) are reproduced from Shanklin (1992).
The Kom Story
Kom people were living at Bamessi (near Lake Nyos)
as guests of the Fon (a ruler is called Fon in the
Grassfields), but the Bamessi Fon was afraid the Kom
were becoming too powerful and he devised a trick to
rid himself of them: he suggested to the Kom Fon that
since their young men were showing signs of their
reigns, they each should build a house and entice the
young men inside, then bar the doors and set the houses
afire. But the wily Bamessi Fon built his house with
two doors and so all the Bamessi men escaped, while
all the Kom men died. Soon the Kom Fon discovered
the trick and vowed revenge. First, he called his sister
to him and told her of his plans. He would hang him-
self and Kom people were not to cut his body down,
nor even go near it; instead, they were to watch and
wait for the appearance of a python track that would
lead them to their new home. Led by the Fon’s sister,
the Kom people followed their Fon’s instructions pre-
cisely. After he hanged himself, his body fluids dripped
down and formed a lake; the Kom people watched.
Maggots from the Fon’s body fell into the lake and
became fish; the Kom people watched. The people of
Bamessi were delighted with the new lake and they in-
formed their Fon, who proclaimed a day when they
would all go into the lake to catch the fish. The day
came and the Kom people watched the people of
Bamessi assemble at the lakeside; then the Bamessi
went into the lake to catch fish for their Fon. Then the
Bamessi went back to catch fish for themselves. At that
point, Kom people say, the lake “exploded”, then sank
and disappeared, taking with it most of the Bamessi
population. Thus was the Kom Fon’s curse fulfilled;
the people of Bamessi were destroyed, leaving the en-
emy Fon with only a few retainers as he had left the
Kom Fon when the two houses were burned. As they
watched from the hills, the python trail appeared to
the Kom people and they turned away to begin the long
journey west, to the area they now occupy.
The Oku Story
At Oku there is a good-sized crater lake and Oku
people say that at one time two groups were settled
beside the lake. On the western slope were the Babanki
or Kijem people and on the eastern slope were the Oku
people. Each had their own Fon. There were many dis-
putes between them, one being a disagreement as to
which group owned Lake Oku. One day a stranger came
and asked the Fon of Kijem for land on which to build
a compound. The Kijem Fon was a disagreeable fel-
low and he refused to give land. The stranger then went
to the Oku Fon, who gave him a building plot. But the
stranger did not like the land that was given, so he
went back to the Oku Fon and asked for a different
plot. The Fon allocated him another, but again the
stranger was not happy, so he returned to the Fon, ask-
ing for a different place. Once again, he was given a
plot, and once again, he returned to complain about
it. Finally the Oku Fon, seeing that the man would not
be satisfied, told him to choose his own land. The man
settled down beside Lake Oku, and, as it is said in
Pidgin English, no one ever knew what he did there.
(The implication is that the man had no visitors be-
cause he was a witch.) When the stranger died, the
Kijem and Oku people went to celebrate his death, the
Kijem people on their side of the lake and the Oku peo-
ple on theirs. Both Fons were called to come into the
lake (presumably by the now-dead stranger) and they
did so, each entering from his side. They were then
taken to the lake bottom and, soon after they disap-
peared, streaks of red (blood) began to appear on the
Oku side. As they watched the red streaks come up, the
Oku people thought their Fon was dead and they be-
gan to mourn for him. At the same time, there appeared
in the distance a Fon dressed in fine new clothes, and
the Kijem people began to cheer, believing their Fon
was being returned to them, having been honored by
the host with precious garments. But, in fact, it was
the Oku Fon who was dressed in fine clothes and the
Kijem Fon who had been slaughtered. The two groups
returned to their homes, wondering what would come
next. Soon after, the waters of Lake Oku left the lake
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 31
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
bed and destroyed most of the homes and people on
the Kijem side; the remnant moved away from the lake,
further west into the nearby Belo Valley. Oku people
still elaborate annual sacrifices to the lake, which
showed by its actions that it wished to belong to the
Oku people. From that day to this, no red streaks have
appeared in the lake.
From the above stories we can find one common
theme that “maleficent” water misbehaves in a spec-
tacular way and sets in motion the migration of ethnic
groups. I believe that this point indicates the occur-
rence of a limnic eruption of Lake Nyos in the past,
although we cannot specify the date(s) and lake(s) of
the past eruptions.
5. SATREPS-NyMo: A project to reduce the risk
of another limnic eruption
The Science and Technology Research Partnership
for Sustainable Development (abbreviated as
SATREPS) is a program for joint research cooperation
between Japan and developing countries for resolving
global issues, e.g., environment, energy, natural disas-
ter prevention and infectious diseases control.
SATREPS is sponsored by the Japan International Co-
operation Agency (JICA) and the Japan Science and
Technology Agency (JST). It was launched in 2008.
The JICA and JST are the organizations under the Min-
istry of Foreign Affairs of Japan, and the Ministry of
Education, Culture, Sports, Science and Technology
of Japan, respectively.
Under the umbrella of the SATREPS, we were able
to obtain funds for a project entitled “Magmatic fluid
supply into Lakes Nyos and Monoun, and the mitiga-
tion of natural disasters through capacity building in
Cameroon”. This started in 2011. The project was nick-
named “SATREPS-NyMo”. It was a 5-year project and
continued until March 2016. The project was headed
by Professors Takeshi Ohba (Tokai University, Japan)
and Minoru Kusakabe (co-leader, University of
Toyama, Japan). The counterpart organization in
Cameroon was the Institute for Geological and Min-
ing Research (IRGM) headed by Dr. Joseph V. Hell,
under the Ministry of Scientific Research and Innova-
tion (MINRESI). The goal of the project was to miti-
gate natural disasters in Cameroon through capacity
building, specifically for issues related to the Lakes
Nyos and Monoun gas disasters. To accomplish the
goal, we planned the following sub-projects: (1) a CO2
discharge system beneath Lakes Nyos and Monoun;
(2) the hydrological regime around the lakes; (3) the
eruptive history of volcanoes along the Cameroon Vol-
canic Line (CVL); (4) the CO2 distribution in Lakes
Nyos, Monoun and other lakes along the CVL; (5) the
setup of an experimental system for removing CO2-
rich deep water to prevent gas re-buildup in Lake
Monoun; and (6) the continuation of geochemical
monitoring of Lakes Nyos and Monoun. During the
project, scientific cooperation between the two coun-
tries was encouraged through the exchange of scien-
tists. Capacity building included scholarships to train
Cameroonian students and technicians in Japan, and
the donation of scientific instruments to IRGM. The
progress of the SATREPS-NyMo can be seen in the
website “http://www.satreps.u-tokai.ac.jp”. The project
went well in terms of scientific achievement. Many
scientific papers were published, e.g., Issa et al. (2013,
2014a, 2014b), Asaah et al. (2014, 2015), Chako
Tchamabé et al. (2013), Fouépé et al. (2013), Fantong
et al. (2013, 2015), Tiodjio et al. (2014, 2015, 2016),
Kamtchueng et al. (2014, 2015a, 2015b), Yoshida et
al. (2016), Ohba et al. (2016), Kozono et al. (2016),
and Saiki et al. (2016).
As described in Section 3, a bottom water removal
system was installed at Lake Monoun in December
2013 to stop re-buildup of CO2 (Yoshida et al., 2016).
The system is shown in Figs. 31, 32. As the degassing
pipes in Lake Monoun had lost their gas self-lift capa-
bility, one of the pipes was utilized to set up a solar
power driven rotary pump, in order to reduce the total
cost of the installation. The intake depth of the pipe is
~99 m, very close to the bottom (100 m). A small ro-
tary water pump with an outer diameter of 74 mm was
placed inside the pipe which had an internal diameter
of 100 mm. Four small solar modules with a total out-
put of 320 W were used as a power source. Although
the system shown in Fig. 32 works only during the
daytime, it is capable of pumping bottom water at an
Fig. 31. Schematic presentation of the deep water removal
system.
32 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
estimated rate of ~100 m3/day. Based on this rate and
the current CO2 concentration of deep water (~90
mmol/kg as of March 2014), the annual removal rate
of CO2 is calculated to be 3.3 Mmol/year. Since this
removal rate is less than half the natural recharge rate
(8.2~8.4 Mmol/year, Kling et al., 2005; Kusakabe et
al., 2008), it would be advisable to install 2 additional
systems at Lake Monoun to equal the natural CO2 re-
charge, in order to reduce the risk of a limnic eruption
in the future. The system is robust and can work for a
long time without complicated maintenance and trans-
portation of fuel, which is an important factor that any
system should have in a remote area like Lakes Nyos
and Monoun in Cameroon.
6. The Cameroon Volcanic Line
6-1. Eruption age of the Nyos maar and potential
collapse of the natural dam
Lakes Nyos and Monoun are maar crater lakes situ-
ated along the Cameroon Volcanic Line (CVL) (Fig.
1). The northern edge of Lake Nyos consists of a 45-
m-wide natural dam (Fig. 33) that holds surface lake
water down to a depth of 40 m. The dam (Fig. 33) is
made of pyroclastic materials deposited at the time of
the volcanic eruption that formed the maar. The upper
unit is moderately consolidated with visible cracks at
the surface, whereas the lower unit is poorly consoli-
dated and looks readily eroded as indicated by a con-
cave structure beneath the upper unit (Lockwood et
al., 1988). Erosion of the lower unit may be facilitated
by seeping water. Lockwood and Rubin (1989) deter-
mined 14C ages of 2 pieces of charcoal found at the
base of the lower unit to be ~400 and ~5100 years BP
(before present). They took the age of 400 years to in-
dicate the age of trees that were growing at the time of
maar formation. The older age was discarded based on
an interpretation that the trees grew in magmatic CO2-
rich atmosphere at the center of the present maar where
the eruption took place. Magmatic CO2 is character-
ized by “dead carbon” (no or very little 14C), and its
incorporation in trees resulted in older ages. The
pyroclastic rocks that form the dam once extended
much farther to the northwest (~600 m), but the lake
water overflowing the spillway has back-eroded these
rocks along the stream bed, leaving only the 45-m-wide
dam at the present time (Fig. 33). An average erosion
rate calculated from these data is 1.5 m/year. At this
rate, the 45-m-wide dam will be eroded away in 30
years, if the age of the dam is correct and the erosion
proceeds at the mean constant rate. It is, however, more
realistic to imagine that the dam collapse will take place
in an irregular and catastrophic way. Figure 34 shows
many joints at the surface of the moderately consoli-
dated upper unit and the seepage of lake water through
the poorly consolidated lower unit. The seepage of
CO2-containing lake water may have chemically eroded
the lower unit in the past, resulting in fall-out of the
lower unit, as suggested by the existence of caves.
There may be an associated breakage of the jointed
upper unit. Thus, the erosion rate may vary irregularly
with time, but it is still alarmingly high. On this basis,
Lockwood et al. (1988) warned that the dam may even-
tually collapse releasing >50 million tons of water and
inducing a catastrophic flood on downstream areas in-
cluding part of Nigeria. This warning was seriously
taken up by the Cameroonian authorities. They asked
support from the United Nations Office for the Coor-
dination of Humanitarian Affairs (OCHA) and the
United Nations Environmental Program (UNEP) for a
detailed survey of the dam. A team of experts from
OCHA and UNEP concluded that a failure of the dam
Fig. 32. Photograph showing the solar power-driven deep
water removal system installed at Lake Monoun. Fig. 33. Photograph showing the 45-m-wide natural dam at
the northern edge of Lake Nyos. The area surrounded by a
green curve is the head of the valley where pyroclastic ma-
terials are said to have been eroded away.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 33
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
is highly likely “to occur within the next 5 years” based
on their geotechnical survey (Joint UNEP/OCHA En-
vironment Unit, 2005). They recommended reinforce-
ment of the dam by cementing fractures and the
unconsolidated part of the dam. As a result, the dam
has been reinforced by engineering methods.
The warning by Lockwood et al. (1988) also drove
geochronological studies of the dam, since the age is
directly related to the erosion rate and thus to the safety
of the dam. The age of the dam has long been debated.
The debate on the age is concisely summarized in Aka
and Yokoyama (2013). Freeth and Rex (2000) proposed
that the age of eruption of the Nyos maar was in ex-
cess of 100,000 years, based on K-Ar dates (Fig. 34)
and evidence from aerial photographs taken in 1963–
1964 that showed no change in the width of the dam
since that time. They concluded that the dam materials
were eroding at a “geologically realistic rate” and that
“there is no reason to suspect that the rate at which it
is currently eroding away is, in itself, sufficient to pose
an immediate threat”. However, the application of the
K-Ar dating method to the basaltic rocks from the Nyos
dam area was criticized by Lockwood and Rubin
(1989), because the Nyos basalts contain fine shards
of K-feldspars which were derived from basement
monzonite (Fig. 34). The K-Ar dating of the rocks con-
taining the shards gives much older ages than their true
age due to the inclusion of K-feldspars with a high ra-
diogenic Ar concentration.
Aka et al. (2008) applied a U-series dating method
to Lake Nyos maar basalts. The basic principles, as-
sumptions and applications of the U-Th dating method
are summarized in Chabaux and Allègre (1994). Aka
et al. (2008) analyzed 12 samples collected from the
Lake Nyos area, including 5 samples of the dam-form-
ing surge deposit and 5 nearby lava flows. They used
XRF and ICP-MS for the analysis of major and trace
element compositions including (238U/232Th), (230Th/
232Th), (226Ra/230Th) and (238U/230Th) ratios. The re-
sults of the Th-Ra disequilibria are reproduced in Fig.
35. The (230Th/232Th) ratios of 10 alkaline rock sam-
ples vary from 0.886 to 1.024, and the (238U/232Th)
ratios vary from 0.716 to 0.880. Data for 26 samples
from the Mt. Cameroon volcano, which has erupted
during the last 100 years, are also included (Yokoyama
et al., 2007). The Lake Nyos and Mt. Cameroon sam-
ples lie closely on a line marked as 238U/230Th = 0.82
with a few exceptions, significantly above the equi-
line which is 238U/230Th = 1.00. This feature indicates
the presence of a 15 to 28% enrichment of 230Th over
238U, suggesting strongly that the Lake Nyos maar for-
mation is younger than ~375 ka which is 5 times the
half-life of 230Th. If the time which has elapsed since
the volcanic eruption is greater than 375 ka, then (230Th/
238U)A (activity ratio) becomes unity, or a secular equi-
librium is established, and no dating can be made (equi-
line in Fig. 35a). Tholeiitic samples, D26 and D27 in
Fig. 35a plot on the line 238U/230Th = 1.00, an indica-
tion that they are in the 238U-230Th radioactive equilib-
rium, giving their formation age older than 375 ka, with
no more information about the age. It is important to
note that the variation in the (230Th/232Th)A and (238U/
232Th)A ratios of the Mt. Cameroon samples (~0.99 and
~0.82, respectively), and the corresponding excess
230Th over 238U (18–24%) were all within the range
for Lake Nyos samples (Fig. 35a). Figure 35b is a
Fig. 34. Cross-section of the natural dam at Lake Nyos. Some explanatory words were added on the cross-section originally
drawn by Lockwood et al. (1988). Figure 3 of Bull. Volcanol., The potential for catastrophic dam failure at Lake Nyos maar,
Cameroon, 50, 1988, 340–349, Lockwood, J. P., Costa, J. E., Tuttle, M. L., Nni, J. and Tebor, S. G., „ Springer-Verlag 1988
with permission of Springer.
34 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
plot of (226Ra/230Th)A vs. (238U/230Th)A for the studied
samples. They were compared to published data for
MORB and OIB (inset). The (226Ra/230Th)A ratios for
the alkaline rock samples range from 1.017 to 1.040
with a mean value of 1.028 ± 0.008. These Nyos data
plot above the (226Ra/230Th)A =1 line (equilibrium), in-
dicating an enrichment of 226Ra compared to 230Th that
was acquired during partial melting of the mantle
source, as is generally observed in oceanic basalts
(Thomas et al., 1999). Similar to 238U-230Th systemat-
ics, the tholeiitic samples are in 226Ra-230Th equilib-
rium. It is highly contrasting that the Mt. Cameroon
data have higher (226Ra/230Th)A ratios (1.09–1.21) than
the Lake Nyos samples (1.01~1.04), although the two
volcanoes are similar in their degree of 238U-230Th dis-
equilibria (Aka et al., 2008). The initial 226Ra/230Th
ratio has to be known to calculate the age of the dam
using the excess 226Ra. Since there are no eruptions of
a known age which have occurred in the Lake Nyos
area, the assumption was made that the initial ratio was
the same (1.15 ± 0.02) as that measured in Mt.
Cameroon lavas that are erupting today (Yokoyama et
al., 2007). Using this assumption, the 226Ra-230Th age
of Lake Nyos was calculated to be 8.75 ± 0.49 ka (Aka
and Yokoyama, 2013) after a careful examination of
the samples. Based on this age, they consider that a
collapse of the Nyos dam from erosion alone is not as
imminent and alarming as has been suggested. How-
ever, making the dam more stable is necessary to com-
pletely eliminate the potential flood hazard.
Stabilization by grouting of the dam has been under-
taken.
Fig. 35. (a) (230Th/232Th)-(238U/232Th) activity ratio diagram for Lake Nyos and Mt. Cameroon samples. Some Lake Nyos
samples are enriched in 230Th compared to 238U by 15–28%. (b) (226Ra/230Th)-(238U/230Th) activity ratio diagram for the samples
showing 2–19% excess 226Ra over 230Th, suggesting a Th-Ra fractionation of <10 ka BP. Reproduced from figure 4 of J.
Volcanol. Geotherm. Res. 176, Aka, F. T., Yokoyama, T., Kusakabe, M., Nakamura, E., Tanyileke, G., Ateba, B., Ngako, V.,
Nnange, J. and Hell, J., U-series dating of Lake Nyos maar basalts, Cameroon (West Africa): Implications for potential
hazards on the Lake Nyos dam, 212–224, Copyright 2008, with permission from Elsevier.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 35
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Fig. 36. Chemical composition of rocks from CVL volcanoes. (a) K2O+Na2O versus SiO2, and (b) Mg number versus SiO2
plots. Reproduced from figure 2 of Asaah et al. (2014) which should be referred to for the abbreviations.
6-2. Origin of the Cameroon Volcanic Line
The Cameroon Volcanic Line (CVL) is an alignment
of Cenozoic volcanoes stretching for 1600 km from
Annobon in the Gulf of Guinea to Biu Plateau in the
continental part of central Africa (Fig. 1). It straddles
both oceanic and continental lithosphere. The CVL can
be grouped into 3 sectors, i.e., the oceanic sector to
the southwest (Annobon, Saõ Tomé, and Principe), the
ocean-continent boundary (Bioko, Etinde and Mt.
Cameroon), and the continental sector (Manengouba,
Bambouto, Oku, Ngaoundéré Plateau, Mandara Moun-
tains and Biu Plateau) to the northeast. The volcanic
islands in the oceanic sector are made up of rocks rang-
ing from nephelinite, basanite and basalt to trachyte
and phonolite (Halliday et al., 1988; Deruelle et al.,
1991). The volcanoes in the ocean-continent bound-
ary are located SW of Mt. Cameroon, and are made of
mostly nephelinitic lavas for Etindé (Nkoumbou et al.,
1995) and basalts and basanites for Bioko and Mt.
Cameroon (Yokoyama et al., 2007; Asaah et al., 2014).
Mt. Cameroon is the only active volcano in the CVL
36 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
with seven eruptions recorded in the last 100 years,
i.e., 1909, 1922, 1954, 1959, 1982, 1999, and 2000 (Suh
et al., 2003). It is a composite volcano made of alka-
line basanitic and basaltic flows interbedded with small
amounts of pyroclastic materials and numerous cinder
cones (Suh et al., 2003; Yokoyama et al., 2007). The
continental sector of the CVL includes Mount
Bambouto and Mount Oku. They are Oligocene to
Quaternary strato volcanoes with lava successions com-
prising a strongly bimodal basalt-trachyte-rhyolite suite
(Marzoli et al., 2000, 2015; Kamgang et al., 2013).
Mt. Manengouba is also in the continental sector and
is a well-preserved stratovolcano whose summit hosts
two concentric calderas with lakes. Lavas range from
basalts to trachytes, quartz trachytes, and rare rhyolites
(Pouclet et al., 2014). The Ngaounderé Plateau in the
northeastern continental part of the CVL consists of
alkaline basalts and basanites capped by trachytes and
phonolitic flows. The Biu Plateau, which is located in
the northern part of the Ngaounderé Plateau, consists
of basaltic flows with a maximum thickness of 250 m.
This plateau is composed of basanite to transitional
basalts (Rankenburg et al., 2005).
Since Lakes Nyos and Monoun are situated on the
CVL, it may be informative to give a brief summary
of the origin of the CVL to understand the characteris-
tics of the Nyos and Monoun volcanoes. The origin of
the CVL has long been a subject of controversy, and
various hypotheses have been proposed. They are sum-
marized by Aka et al. (2004) and more recently by
Asaah et al. (2014), as follows: (1) Reactivation of pre-
existing tectonic structures in the Cenozoic associated
with crustal melting (Gorini and Bryan, 1976; Moreau
et al., 1987; Fairhead, 1988). (2) Membrane stresses
generated by the movement of the African plate away
from the equator (Freeth, 1978). (3) Displacement of
the African plate (Fitton, 1980). (4) Hotspot trail
(Morgan, 1983). (5) Hotline hypotheses (Meyers et al.,
1998). (6) A plate-wide shallow mantle convection
model (Burke, 2001). (7) Edge convection and
lithospheric instability (Reusch et al., 2010). Of the
above, Fitton’s classic hypothesis is still attractive in
that the Benue Trough and the CVL are related to a
common “Y”-shaped hot zone in the asthenosphere
over which the African plate moved during the period
of 110 to 70 Ma (Fitton, 1980). The “Y”-shaped hot
zone was a rift zone that extended from a triple junc-
tion originally located at the Gulf of Guinea that was
underlain by the St. Helena hotspot at the time of the
opening of the south Atlantic. The CVL developed over
this rift zone. In this sense, the magmatism in the CVL
may have been similar to that in the currently active
East African Rift Zone. Asaah et al. (2014) went for
the “hotline” model which invokes multiple plumes
Fig. 37. Trace element patterns for mafic rocks from the oceanic and continental sectors of the CVL. Reproduced from figure
4 of Asaah et al. (2014). The patterns are generally similar to each other and akin to OIB, suggesting an origin from a similar
source. Reproduced from figure 5 of Asaah et al. (2014).
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 37
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Fig. 38. (a) 207Pb/204Pb-206Pb/204Pb diagram, and (b) 143Nd/144Nd-87Sr/86Sr diagram for the CVL lavas. (c) Positive and negative
trends were seen in the 143Nd/144Nd-206Pb/204Pb relationship for the CVL lavas. Reproduced from figures 6, 7 and 9 of Asaah et
al. (2014). EM1 is for Enriched Mantle type 1, EM2 for Enriched Mantle type 2, NHRL for Northern Hemisphere Reference
Line, HIMU for high-
m
(=238U/204Pb ratio), FOZO for Focal Zone, MORB for Mid-Ocean Ridge Basalt, OIB for Ocean Island
Basalt, and DMM for Depleted MORB Mantle.
originating from the same source in the upper mantle,
each of which produced volcanoes independently, as
the model appears to explain the diverse features of
the CVL, i.e., geophysical, structural and geochemical
evidence, including the absence of time-dependent
volcanic activity.
Magmatism of the CVL is characterized by melting
in the garnet lherzolite stability fields (Marzoli et al.,
2000; Yokoyama et al., 2007; Kamgang et al., 2013),
although melting in the spinel lherzolite stability field
38 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
has been reported in the Ngaoundere Plateau (e.g., Lee
et al., 1996; Nkouandou and Temdjim, 2011). In addi-
tion, mixing of both garnet and spinel melting fields
has been reported in Mt. Cameroon (Tsafack et al.,
2009). Both mafic and felsic rocks show chemical fea-
tures consistent with a plume activity outlined by their
ocean island basalt (OIB) characters, and their isotopic
ratios (Mbassa et al., 2012).
6-3. Geochemistry of CVL magmas
Asaah et al. (2014) made a comprehensive review
of the geochemistry of CVL rocks. They compiled the
existing geochemical data of the CVL rocks (580 sam-
ples) consisting of major and trace element composi-
tions and radiogenic (Sr-Nd-Pb) isotope compositions.
Figure 36 shows some chemical characteristics of the
CVL rocks in terms of (a) K2O+Na2O versus SiO2, and
(b) Mg number versus SiO2 plots. The SiO2 contents
show a wide range of variation from 38% (oceanic
CVL) to 79% (continental CVL) reflecting the diverse
rock types. The Mg number (Mg#), defined as Mg# =
MgO/(MgO+FeO)*100, is often used as an index of
the level of evolution of volcanic rocks. It shows dif-
ferent trends from one volcanic center to another. The
CVL rocks from the oceanic and continental sectors
are dominantly alkali basalts and basanites. The Mg#
of mafic samples ranges from 60~69 (least evolved
basalts) to 40~49 (evolved rocks), indicating various
fractional crystallization paths (Fig. 36b). Refer to
Asaah et al. (2014) for further discussion.
Abundance patterns of trace elements are often used
to discuss magma genesis, since they provide
geochemical and geological information through their
unique chemical properties and sensitivity to processes
to which major elements are insensitive. Primitive-
mantle normalized trace element patterns (Palme and
O’Neill, 2003) for mafic rocks from the oceanic and
continental sectors of the CVL are presented in Fig.
37. The patterns are generally similar to each other
(except for the Mt. Etindé samples) and akin to OIB,
suggesting an origin from a similar source. They show
a marked enrichment of light rare earth elements
(LREEs) and a strong fractionation of heavy rare earth
elements (HREEs) relative to LREEs. The most strik-
ing features of Fig. 37 are: (1) the Mt. Etindé samples
have high trace elemental abundances compared to the
other CVL alkaline basalts; (2) relatively high posi-
tive anomalies of Nb, La and Nd; and (3) the occur-
rence of a K-trough. Nearly constant elemental ratios
of incompatible elements in CVL rocks suggest that
magma processes, such as zone refining melting,
magma mixing, and extensive fractionation and replen-
ishment, were not dominant processes during the gen-
eration of the CVL mafic lavas, because the above proc-
esses can efficiently fractionate incompatible elements.
The peculiar features of Mt. Etindé may have resulted
from source materials that are different from the other
CVL lavas, as suggested by the Mg# versus SiO2 trend
(Fig. 36b) and a strong high
m
(HIMU) character there.
The radiogenic isotope (Sr-Nd-Pb) geochemistry of
the CVL rocks is also summarized in Fig. 38, adapted
from Asaah et al. (2014). The Sr-Nd-Pb isotopic com-
positions of the CVL basalts overlap those of OIB. In
the 207Pb/204Pb vs. 206Pb/204Pb diagram (Fig. 38a), the
data plot parallel to the Northern Hemisphere Refer-
ence Line (NHRL) of Hart (1984), and to the right of
the 4.53 Ga geochron. However, some data from the
Oku Volcanic Group (OVG) with low 206Pb/204Pb and
207Pb/204Pb ratios overlap with the MORB end mem-
bers (Atlantic, Indian, and Pacific MORBs). The 143Nd/
144Nd ratios and 87Sr/86Sr ratios of the mafic rocks show
a limited range of variation (Fig. 38b). The 87Sr/86Sr
ratios range from 0.70286 in a sample from the Biu
Plateau to 0.70515 in a sample from Mt. Bambouto.
The 143Nd/144Nd ratios vary from 0.51302 in a sample
from the Biu Plateau to 0.52771 in a sample from Mt.
Cameroon. Some lavas from the Biu Plateau and the
oceanic CVL show relatively low 87Sr/86Sr and high
143Nd/144Nd ratios, implying that they are more primi-
tive than other continental volcanic rocks (Mt.
Bambouto, Mt. Manengouba, and the OVG). Isotope
data for the OVG show a wider spread than those of
the other CVL volcanoes. This difference is conspicu-
ous in the Pb isotopes. In Fig. 38b, a negative correla-
tion is observed between 143Nd/144Nd and 87Sr/86Sr ra-
tios and the correlation slope matches the mantle array
of MORB-OIB samples. From these figures it is sug-
gested that the CVL lavas formed by a dominant con-
tribution of EM2 to the Depleted MORB Mantle
(DMM). The 143Nd/144Nd versus 206Pb/204Pb plots (Fig.
38c) show positive and negative correlations with dif-
ferent slopes, where the role of EM2 becomes domi-
nant over EM1. Mixing with various end members in
different proportions may account for the complex iso-
topic characteristics of the CVL lavas.
Based on trace element and isotope geochemistry, it
has been suggested that these magmas derived from
the sub-lithosphere without interaction with the over-
lying lithosphere (Fitton and Dunlop, 1985). A differ-
ent view, however, was given by Halliday et al. (1990)
that the continent/ocean boundary magmas (Bioko,
Etindé and Mt. Cameroon) are characterized by 206Pb/
204Pb ratios that are higher (more radiogenic) than those
of typical continental and oceanic sector magmas. This
radiogenic feature has also been confirmed by the dis-
tribution of 3He/4He ratios of lavas and mineral sepa-
rates from the CVL rocks showing a clear 3He/4He
valley as already illustrated in Fig. 20 (Aka et al.,
2004). Together with Sr, Nd and O isotopic variations,
Halliday et al. (1988, 1990) suggested that the radio-
genic nature of the 206Pb/204Pb ratios of rocks from the
ocean-continent boundary reflects melt migration from
the St. Helena fossil plume head that took place at 125
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 39
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Ma, and that some of the CVL magmas derive from
the upper metasomatized part of the fossil plume in
the lithospheric mantle (Fig. 39). The degree of trace-
element enrichment (U/Pb, or
m
in this case) varies as
a function of the vertical thickness of the plume head
through which the melts migrated. At the margins
where the flattened plume head is thinnest, the source
regions are dominated by more depleted mantle
(Halliday et al., 1990). The radiogenic nature of 206Pb/
204Pb ratios and the 3He/4He valley observed at the
ocean-continent boundary region can be explained by
the magma genesis affected by the fossil plume head.
6-4. Volatiles in magma
A fundamental question arises as to whether Lake
Nyos magmas are enriched in CO2. Volatile contents
in pre-eruptive magmas have been estimated by vari-
ous techniques. One of the techniques is to analyze melt
inclusions in phenocrysts, since melt is trapped in
growing phenocrysts as melt inclusions in magma and
is quenched to glass at the time of eruption. The
volatiles, mainly H2O and CO2, in the glass inclusions
are determined by microanalytical techniques such as
Fourier transform infra-red spectroscopy (FTIR), la-
ser-Raman spectrometry, and secondary ion mass
spectrometry (SIMS), etc. (Ihinger et al., 1994). An-
other approach is experimental petrology where min-
eral stabilities and assemblages are calibrated under
different (but controlled) conditions, such as tempera-
ture, pressure, and water fugacity. Comparison of ex-
perimental products with natural phenocrystic assem-
blages allows us to constrain the pre-eruptive volatile
contents (Johnson et al., 1994). Unfortunately for us,
however, lavas from the Nyos volcano are mostly
aphyric (Aka et al., 2008) and difficult to use for the
analysis of pre-eruptive volatile contents by the afore-
mentioned techniques. Instead, based on major and
trace elements systematics, Aka (2015) proposed that
the Nyos basalts formed by a small degree (1~2%) of
partial melting of the primitive mantle to which
amphibole and phlogopite had been added by
carbonatitic fluids, and that decarbonation reactions of
the carbonatitic metasomatism are responsible for pro-
ducing the magmatic CO2. However, based on the
geochemical data of the Nyos volcanic rocks, Asaah et
al. (2015) suggest that CVL magmatism is predomi-
nantly of an asthenospheric source with little contri-
bution from the subcontinental lithospheric mantle
(SCLM). The lavas show evidence of enrichment by
metasomatic fluids probably in the Mesozoic (e.g.,
Halliday et al., 1990; Aka, 2015; Asaah et al., 2015).
The metasomatism affected the SCLM, inducing hy-
drous minerals like amphibole and phlogopite that are
not stable in the asthenosphere. Asaah et al. (2015)
suggest that the metasomatic fluids crystallized as small
pockets or veins in the SCLM. An ultimate source of
CO2 in the Nyos magma may derive from the
decarbonation of such crystallized metasomatic fluids.
It is unlikely that the CVL magmas, including the Nyos
magma, have abnormal CO2 in their mantle source.
Lake Nyos and Lake Monoun volcanoes are located
in the Oku and Bambouto volcanic centers, respec-
tively, in the middle of CVL (Fig. 1). Lake Nyos is a
maar lake created by a phreato-magmatic eruption.
There are some other maar lakes near Lake Nyos, i.e.,
Oku, Elum, Nyi, Wum and Enep, but only Lake Nyos
contains a large amount of dissolved CO2. According
to Lockwood and Rubin (1989) who described the ge-
ology of the Nyos volcano, eruption sequences are sum-
Fig. 39. Schematic presentation of a model for source enrichment in high
m
elements following plume emplacement at 125 Ma
beneath the ocean/continent boundary of CVL (St. Helena). Reprinted by permission from Macmillan Publishers Ltd: Nature
347, 523–528, Halliday, A. N., Davidson, J. P., Holden, P., DeWolf, C., Lee, D.-C. and Fitton, J. G., Trace-element fractionation
in plumes and the origin of HIMU mantle beneath the Cameroon line, Copyright 1990.
40 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
marized as shown below. The formation of the Nyos
maar is directly related to the ascent of alkali basalt
magma. The first lava reached the surface in a rela-
tively gentle, fire-fountaining fashion, depositing
scoria and fluid bombs over a wide area around the
present north end of Lake Nyos. This phase was fol-
lowed by an explosive and violent eruption due to vola-
tile expansion. The violence of the activity increased
rapidly, however, and basalt is only included as shat-
tered fragments in upper parts of the pyroclastic sec-
tion. Neither the basalt flow nor the associated scoria
at the base of the pyroclastic section were found to
contain ultramafic xenoliths, suggesting that mantle
rocks were only transported to the surface during the
later more explosive phases of the eruption. The depth
of the explosive activity may have gradually increased
during the eruption, and the initial explosion crater
gradually widened, which resulted in the formation of
the maar crater, or the present Lake Nyos. It can be
imagined that the magma subsided after the eruption,
but the release of CO2-rich volatiles from the magma
continues until today.
Magma is generally generated by a partial melting
of rocks in the lower crust or upper mantle. Mantle
rocks, mainly comprised of peridotite, exist as a solid,
because the geothermal gradient within the Earth is
generally below the solidus of mantle rocks. It has been
hypothesized that part of solid mantle, if heated lo-
cally, can ascend as a diapir and cross the solidus where
partial melting starts to take place. Volatile materials
such as H2O and CO2, if they coexist with the rocks,
reduce the solidus temperature and facilitate a partial
melting of the rocks, or magma genesis. Thus, the co-
existence of volatiles is important for partial melting.
Metasomatic fluids may have affected the primitive
mantle beneath Lake Nyos and the fluids produced by
the decarbonation of metasomatized mantle facilitated
partial melting (Aka, 2015). Once a melt is formed, it
rises through the mantle due to its lower density (higher
buoyancy) and approaches the surface of the Earth to
form magma. The melt may remain as a magma reser-
voir in the shallow part of the crust. When the magma
further ascends, crystallization in the magma begins
because of the reduction of temperature and pressure.
Magma contains various volatile materials, such as
H2O, CO2, S, Cl, etc. Since the volatile materials will
not all be incorporated into crystals (or minerals), they
tend to be concentrated as fluids in the magma as it
rises and cools. A volcanic eruption is often facilitated
by magma ascent driven by a lowered density due to
the accumulation and expansion of bubbles of the
volatiles in the magma.
The chemical composition and concentration of mag-
matic volatiles have been estimated through the analy-
sis of high temperature volcanic gases, chilled glassy
margins of lava that has extruded onto the bottom of
the deep ocean, and glass inclusions in phenocrysts of
volcanic rocks. Volatiles in magma are almost com-
pletely discharged into the atmosphere at the time of
volcanic eruption. For this reason, the chemical analy-
sis of high-temperature volcanic gases, if collected and
analyzed properly, can give the volatile composition
(not concentration) in magma. Table 4 shows the con-
centration of H2O, CO2, S and Cl in some types of
magma, and the composition of volcanic gases that is
expected from the degassing of each type of magma
(Shinohara, 2003). The concentration of the magmatic
volatiles in magma is highly variable depending on its
type. Water concentration in Mid-Oceanic Ridge Ba-
salt (MORB) is low (0.1~0.5 wt%), whereas that of
subduction zone magma is more than an order of mag-
nitude higher (1~5 wt%). The concentration of CO2 in
Shinohara (2003) and Giggenbach (1996).
Concentration in magma Expected gas compostion
H
2
OCO
2
SCl H
2
OCO
2
SHCl
wt% ppm ppm ppm mmol/mol mmol/mol mmol/mol mmol/mol
Glassy margin of MORB
Eastern Pacific Ocean 0.12 1630 690 50 526 292 170 11
Atlantic Ocean 0.20 1320 1170 60 621 167 203 9
Melt inclusions from hotspot basalts
Kilauea, ocean floor 0.46 3100 1050 æ712 196 91 æ
Kilauea, summit 0.23 800 1300 80 677 96 215 12
Melt inclusions from subduction zone volcanoes
Basalt 1. 0 >1000 1000 1000 871 3 6 49 44
Andesite 3. 0 >1000 400 3000 933 13 7 47
Rhyolite 5.0 >1000 100 2000 971 8 1 2 0
Table 4. Concentration of volatiles in magma and expected gas composition from the magma.
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 41
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
MORB magma is 1000~3000 ppm, which is generally
higher than that of subduction zone magma. The cal-
culated composition of gases exsolved from magma is
also shown in Table 4. Such calculated gas composi-
tions are in general agreement with the observed gas
compositions (not shown, but can be found in Allard,
1983; Gerlach, 1983; Shinohara, 2003). In gases from
hotspot basaltic volcanoes such as Kilauea (Hawaii),
the water content is lower than that from subduction
volcanoes, whereas the CO2 content is significantly
higher in hotspot and mid-oceanic volcanoes, reflect-
ing its higher concentration and low solubility in ba-
saltic melts.
Solubilities of H2O and CO2 in silicate melts have
been experimentally determined as shown in Fig. 40
(Holloway and Blank, 1994, and references therein).
The solubility depends on the temperature, pressure
and the chemistry of melts. It increases as the partial
pressure of the volatile species in question increases,
and decreases as the temperature of the melt increases.
Generally speaking, water is approximately an order
of magnitude more soluble than CO2. Water dissolves
slightly more in silicic melts than in basaltic melts,
whereas CO2 dissolves more in basaltic than in silicic
melts (Fig. 40). Figure 41 illustrates the solubility of
CO2 and H2O in basaltic melts at 1200∞C as a function
of the total pressure of the volatiles. The non-linear
relationship of this binary system in the melts comes
from the non-ideal mixing properties of these species
(Holloway and Blank, 1994). Using Fig. 41, it can be
envisaged how the volatile composition in the melt
changes as the decompression proceeds. For example,
at point A of Fig. 41, where CO2 = 540 ppm and H2O =
1.6 wt%, the melt is saturated with the coexisting fluid
of which the mole fraction of H2O equals 0.2 and that
of CO2 is 0.8. This implies that the fluid coexisting
with the basaltic melt is extremely rich in CO2. As the
magma ascends, or the confining pressure is reduced,
the fluid exsolves, or degassing takes place. If
degassing proceeds in a closed system, the fluid com-
position in the melt will follow the thin dotted line
depending on the co-existing H2O concentration as
shown in Fig. 41. If degassing takes place in an open
system, the melt composition may follow a different
path, as indicated by the thick long dashed line, since
the CO2-rich fluid leaves the magma when the system
becomes open due to the low CO2 solubility in the
melts, making the remaining magma progressively
CO2-poor, while the H2O concentration decreases only
a little. As long as the magma keeps open-system
degassing, CO2-rich fluid is continuously released from
the magma. This solubility-controlled behavior of CO2
in basaltic magma may explain a CO2-rich nature of
fluids separated from the magma. The ultimate source
of CO2 in the Nyos magma may derive from the
decarbonation of crystallized metasomatic fluids in the
Fig. 41. Solubilities of CO2 and H2O in basaltic melts at
1200∞C as a function of the total pressure of the volatiles
(Holloway and Blank, 1994; Shinohara, 2003).
Fig. 40. Solubilities of H2O and CO2 in silicate melts (Holloway and Blank, 1994; Shinohara, 2003).
42 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
subcontinental lithosphere (Aka, 2015; Asaah et al.,
2015). The permanent supply of such CO2 is likely to
be responsible for the high concentration of CO2 gas
in the fluids feeding into Lakes Nyos and Monoun.
7. Other CO2-rich volcanic lakes in the world
Of 714 volcanoes in the world, 86 volcanoes host
lakes (Pasternak and Varekamp, 1997). Information on
the volcanic lakes is now available from the VOLADA
(2013) database (https://vhub.org/resources/2822).
Although Lakes Monoun and Nyos in Cameroon be-
came notoriously famous because of the gas disasters
in the mid-1980s, other CO2-rich volcanic lakes exist
in the world, e.g., Lake Kivu (Democratic Republic of
the Congo and Rwanda, see below for references),
Laacher See (Germany) (Aeschbach-Hertig et al.,
1996), Lake Van (Anatolia in eastern Turkey) (Kipfer
et al., 1994), Lago Albano and the two Monticchio
lakes (Italy) (Anzidei et al., 2008; Caracausi et al.,
2009), Laguna Hule and Rio Cuarto (Costa Rica)
(Alvarado et al., 2011), and Lac Pavin (France)
(Aeschbach-Hertig et al., 1999).
Of these lakes, Lake Kivu has been known to con-
tain a high concentration of CO2 and CH4 in its deep
water since well before the Lake Nyos event (e.g.,
Deuser et al., 1973; Tietze et al., 1980). Carbon diox-
ide dissolved in the lake is basically of magmatic ori-
gin, a situation similar to that in Lakes Nyos and
Monoun in Cameroon, although the magmatic CO2 is
mixed with a variable proportion of biogenic CO2. The
lake is located along the East African Rift on the bor-
der between the Democratic Republic of the Congo
(DRC) and Rwanda. The lake area is tectonically and
volcanically active as part of the East African Rift Sys-
tem. Because of the high gas concentrations in the lake
and the large population around it, Lake Kivu has a
potential risk of a gas disaster caused by a limnic erup-
tion which may be triggered by a possible volcanic
eruption at the lake bottom (Schmid et al., 2005) or a
plunge of lava flows from Nyiragongo, the nearest ac-
tive volcano (only 20 km NE to the lake). Indeed, the
2002 eruption of the volcano generated lava from flank
fissures flowed into the city of Goma, the provincial
capital, resulting in destruction of local structures and
the evacuation of local people, and these lava flows
eventually ran into the lake. Fortunately, no limnic
eruption was induced at that time (Tedesco et al., 2007).
Detailed gas and water chemistry of Lake Kivu and
the surrounding region has been published by several
authors (Tietze et al., 1980; Tassi et al., 2009; Schmid
et al., 2005). The lake has 5 basins, each of which is
characterized by a different chemistry, CO2 profile, and
biology. The main basin (>250 m) contains the highest
CO2 concentration with a horizontal heterogeneity.
Although the highest CO2 concentration in the main
basin is far from saturation at any depth, the CO2 con-
centration at Kabuno Bay (a small basin on the north-
western end of Lake Kivu) is relatively close to satu-
ration. Since Kabuno Bay is shallower than the main
basin and is characterized by the highest input of CO2-
rich magmatic fluid, the bay is considered to be poten-
tially most hazardous in terms of the possibility of
limnic eruption. Continuous monitoring is recom-
mended (Tassi et al., 2009). The concentration of dis-
solved CH4 is highest (~17 mmol/L), approx. 12% of
dissolved CO2 at the bottom of the main basin. The
gas is produced by the bacterial reduction of CO2 and
acetate fermentation (Schoell et al., 1988). It is im-
portant to note that microbial activity contributes to
the gas chemistry in deep, stratified and anaerobic
lakes, as recently found also at Lakes Nyos and
Monoun (Tiodjio et al., 2014, 2016). Carbon isotopic
ratios (
d
13C) of CO2 dissolved in the main basin of the
Lake Kivu range from –7 to –6‰ (relative to VPDB),
suggesting also a large contribution from mantle-origi-
nating CO2. Those at Kabuno Bay (–11 ~ –13‰), how-
ever, are significantly lower than the values for the
main basin, probably reflecting the interaction of mag-
matic fluids with organic-rich sedimentary materials
that underlie volcanic rocks derived from nearby
Nyamulagira and Nyiragongo volcanoes (Tassi et al.,
2009). The 3He/4He ratio of Kabuno Bay water is 5.5
Ratm, indicating a large contribution of a magmatic
component regardless of the low
d
13C values, whereas
the ratio ranges from 2.1–2.6 Ratm in the main Kivu
basin water. Fumarolic gases collected at the summit
crater of the Nyiragongo volcano may best represent
the 13C/12C and 3He/4He ratios of magmatic end-mem-
bers in the fluids that are supplied to Lake Kivu and
its surroundings (Tedesco et al., 2010). These authors
observed typical mantle values of
d
13C = –3.5 ~ –4‰
and 3He/4He ratios up to 8.7 Ratm for the fumarolic
gases. The influence of this magmatic signature be-
comes smaller, and the crustal components increase,
as we move southward (toward Lake Kivu). The C/
3He ratio of ~30 ¥ 1010 was observed for summit
fumarolic gases. This high value probably reflects the
high CO2 solubility in the Nyiragongo magma which
is foiditic (alkaline), different from typical MORB
magmas. High C/3He ratios up to 36 ¥ 1010 were mea-
sured for the main basin water of Lake Kivu. Although
these ratios are close to the Nyiragongo magmatic
value, it is more likely that the addition of CO2 in lo-
cal groundwater that has interacted with organic mate-
rials enhanced the lake’s C/3He ratio, as suggested by
d
13C values. These observations show that magmatic
fluids interact with surrounding materials in varying
degrees, and that the gas geochemistry of this area is
controlled by the local tectonic-geologic settings
(Tedesco et al., 2010).
Lake Mashu is a small, dimictic (mixing twice a year)
caldera lake in Hokkaido, Japan, with a surface area
of 19 km2 and a maximum depth of 211 m. A hot spring
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 43
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
has been identified at the bottom of the lake. The chemi-
cal characteristics of the lake were given by Nojiri et
al. (1990). Based on noble gas data of the lake water
collected at various time, points and depths, Igarashi
et al. (1992) estimated a 3He/4He ratio of 6.7 Ratm for
helium supplied from the lake bottom through the hot
spring, suggesting the addition of mantle helium to the
lake from the underlying magma. The accumulation of
mantle helium between two overturns (spring and au-
tumn) was estimated to be 9.2 ¥ 107 atoms cm–2 s–1
(4He) and 8.7 ¥ 102 atoms cm–2 s
–1 (3He) using the
helium profiles and a one-dimensional diffusion model.
Since the CO2 supply rate was estimated to be 5.3 ¥
108 mol/y (Nojiri et al., 1993), a high C/3He ratio of
18 ¥ 1010 can be calculated for the supply fluid. The
C/3He ratio, similar to that estimated for the Alban Hills
volcanic district (see below), is two orders of magni-
tude greater than the MORB value. Igarashi et al.
(1992) attributed this high value to the enrichment of
CO2 in the source magma beneath Lake Mashu. Al-
though the CO2 supply rate of 5.3 ¥ 108 mol/y is greater
than that for Lake Nyos (1.2 ¥ 108 mol/y, Kusakabe et
al., 2008) by a factor of ca. 4, the dimictic nature of
the lake does not allow an excessive accumulation of
CO2 in it, which is fortunate from the limnic eruption
perspective.
Laacher See is also a 53-m-deep holomictic (com-
plete vertical mixing once a year) maar lake in the East
Eifel volcanic district in Germany, where the discharge
of CO2 gas from the lake has been observed for years.
Helium and neon isotopes dissolved in the lake were
measured twice (spring and early autumn) in 1991 by
Aeschbach-Hertig et al. (1996) with the aim of esti-
mating the helium flux from the lake bottom, since
gases supplied from the bottom were considered to
accumulate in the lake during summer stratification.
Both the He concentration and 3He/4He ratios increased
with depth, and the rate of increase was more clearly
observed in early autumn. The 3He/4He ratio of the
incoming He was estimated to be 5.4 Ratm, suggesting
a large contribution of magmatic He with a minor
crustral contribution. Using the amount of He stored
during summer stratification and a one-dimensional
vertical mixing model, the 4He flux into the lake was
estimated to be 10 ¥ 108 atoms/cm2 s–1 with a 3He/4He
ratio of 5.3 Ratm. Since gas samples from the lake were
>99% CO2, a C/3He ratio of 8.6 ¥ 109 was calculated.
Combining the 3He flux of 7.4 ¥ 103 atoms cm–2 s–1, a
CO2 flux into Laacher See was estimated to be 3.3
mmol cm–2 y–1 (Aeschbach-Hertig et al., 1996). This
is equivalent to an annual release of 1.1 ¥ 108 mol CO2
to the atmosphere. Even if this value represents the
annual recharge of CO2 to the lake, the holomictic na-
ture does not allow the accumulation of CO2 as was
the case in Lakes Nyos and Monoun.
Lake Van in Anatolia, eastern Turkey, was formed
during the Pleistocene in a tectonic depression with its
outlet blocked by lava flows from the nearby Nemrut
volcano. Lake Nemrut is one of the caldera lakes near
Lake Van. The injection of He, derived from depleted
mantle with 3He/4He ratio of 7.4 Ratm, into Lakes
Nemrut and Van was documented by Kipfer et al.
(1994). It is likely that CO2 is also supplied to the lakes,
but unfortunately there is no mention of CO2 in the
lakes in Kipfer et al. (1994).
Alban Hills in the volcanic area near Rome, Italy,
has been characterized by high emissions of CO2 from
a pressurized CO2-rich aquifer, and small-scale gas
outbursts from the aquifer have been recorded
(Carapezza and Tarchini, 2007). Carbon isotopic ra-
tios were reported to be in a limited range around
+1.3‰ (relative to VPDB), which suggests the contri-
bution of decomposed marine carbonates as the source
of CO2. The 3He/4He ratio of He in the associated gas
was 1.9 Ratm, very low compared to MORB and sub-
duction volcanic gas values, but still suggestive of a
magmatic affiliation. The C/3He ratio of gases collected
from a nearby well is 2.3 ¥ 1011, 2 orders of magnitude
greater than typical MORB values. This value is con-
sistent with a high contribution of CO2 that was most
likely derived from the thermal decarbonation of lime-
stone involved in magma genesis at the Alban Hills
volcanic district. Historical evidence has shown that
Lake Albano, a 160-m-deep crater lake located in the
center of the district, experienced lahars associated with
water overflow (Carapezza and Tarchini, 2007). The
present water and gas chemistry of the lake indicates
that dissolved CO2 concentration increases with depth
in anoxic hypolimnion (>80 m). However, the total gas
pressure calculated from the CO2 concentration is far
below the hydrostatic pressure at all depths, suggest-
ing that a gas hazard at the lake is unlikely, unless CO2
from the pressurized aquifer is suddenly injected into
the lake (Carapezza et al., 2008). The Monticchio cra-
ter lakes in Southern Italy are also receiving passive
magmatic CO2, and the potential risk of a Nyos-type
gas hazard has been described (Caracausi et al., 2009).
8. Concluding remarks
This review mainly summarizes the author’s achieve-
ments in work and related matters on the Lakes Nyos
and Monoun gas disasters that took place in the mid-
1980s in Cameroon. At that time, nobody knew that
lakes could accumulate so much CO2 gas and then sud-
denly release it to induce such disasters. The Lake Nyos
and Monoun events had a strong impact on scientists
working on gas emissions from the interior of the Earth.
This impact especially boosted volcanic lake studies.
Soon after the 1986 gas burst at Lake Nyos, scientists
working on the initial phase of their research created a
small informal group “The International Working
Group on Crater Lakes (IWGCL)” to exchange scien-
tific information about the Lake Nyos gas disaster, to
44 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
coordinate follow-up field trips planned by those who
were interested in the subject, and to organize scien-
tific meetings as a forum for further discussions. The
scope of IWGCL was later expanded to include not
only studies of gassy lakes in Cameroon but also those
of other volcanic lakes in general. The new objectives
were to obtain information on the activity and
degassing state of shallow magmatic bodies so that
forecasting volcanic eruptions and the mitigation of
volcanic lake-related hazards could be achieved. Ex-
pansion of the scope of IWGCL naturally meant a
greater number of scientists, and resulted in acquiring
a formal IAVCEI status as the Commission on Volcanic
Lakes (CVL) in 1993. I was satisfied by these organi-
zational developments as the leader of IWGCL and
CVL in those early days. The CVL has organized sci-
entific meetings every 2–3 years, reports of which can
be found in the website “http://www.ulb.ac.be/sciences/
cvl/”. On a personal note, and as a scientist who has
worked on Lakes Nyos- and Monoun-related disaster
reduction issues for close to 30 years, I was particu-
larly happy to know that the CVL-9 meeting took place
in Yaoundé, Cameroon, in March 2016, to commemo-
rate the 30th anniversary of the Lake Nyos gas disas-
ter.
During my career, I have acquired experiences of
working in national and international scientific com-
munities, and, consequently, have made friendships
with many wonderful scientists worldwide. Such ex-
periences have led me to obtain research funding that
has made it possible for me to continue to work in
Cameroon for ~30 years. As described in Section 5 of
this article, a typical example was the success in get-
ting support from JICA and JST for the SATREPS-
NyMo project. The resolution of solving various prob-
lems associated with the Lakes Nyos and Monoun gas
disasters, such as the continuation of scientific moni-
toring of the lakes, the monitoring of the reinforced
natural dam at Lake Nyos, the rehabilitation and set-
ting up of an infrastructure for the displaced people,
etc., are obviously domestic issues for which the
Cameroonian Government and scientists should, in
principle, take responsibility. But the reality is differ-
ent; the economic insufficiency of Cameroon has hin-
dered the principle. The main goal of the SATREPS-
NyMo project is to mitigate natural disasters in
Cameroon through capacity building, specifically for
issues related to the Lakes Nyos and Monoun gas dis-
asters. The risks of the recurrence of limnic eruptions
can be defused if proper and timely actions are taken.
The SATREPS-NyMo capacity building included the
donation of some analytical instruments necessary to
help Cameroonian scientists achieve the project’s goals.
Also included was the training of young Cameroonian
scientists and technicians in Japan, so that, after they
get back home, they can play an important role in the
field of mitigation of natural disasters. Unfortunately,
the gas content at Lake Monoun has recently been
found to be increasing due to the continuing gas sup-
ply from the underlying magma, the duration of which
is much longer than the span of human life. It is al-
most certain that the same situation will occur at Lake
Nyos within several years when the gas self-lift capa-
bility is lost. Now is the turn for Cameroonian scien-
tists and technicians to work toward defusing the new
risks of the increasing gas content in the lakes, for they
have acquired the needed knowledge and techniques. I
hope the safety of the lakes is secured and that the sur-
rounding populations can return to their ancestral roots
and go about their daily lives without the fear of fur-
ther gas disasters.
Acknowledgments
This article is a review of scientific achievements con-
cerning the Lakes and Monoun Nyos gas disasters and re-
lated subjects. The review could not have been made with-
out the cooperation of many colleagues and friends who
worked together with me in the field. I express my sincere
thanks to: Y. Yoshida, T. Ohba, K. Nagao, G. Tanyileke, F.
T. Aka, Issa, Y. W. Fantong, J. V. Hell, G. W. Kling, W. C.
Evans, D. Rouwet, and many others, who worked together
in the field and have provided me with important scientific
information. Special thanks go to T. Ohba who kindly sup-
plied recent data (unpublished) on the CO2 concentrations
in the lakes used in Fig. 15. Fieldwork since in the period
1986–2006 was mostly supported by the Grant-in-Aid for
Scientific Research from JSPS (Japan Society of Promotion
of Science). Recent fieldwork (2011–2015) has been sup-
ported by the SATREPS-NyMo project. Logistic support
from IRGM and its technicians is appreciated. The Embassy
of Japan and the JICA office in Yaoundé are acknowledged
for their help while I was in Cameroon.
K. Oshida of TerraPub is acknowledged for giving me the
chance to write this review. I also thank Y. Matsuhisa who
made constructive comments on an early version of the
manuscript. W. C. Evans, D. Rouwet and F. Aka are also
thanked for their comments that helped improve the manu-
script. The English of the final version was improved by D.
Larner who kindly checked the manuscript in a very careful
manner and suggested the corrections.
References
Aeschbach-Hertig, W., Kipfer, R., Hofer, M., Imboden, D.
M., Wieler, R. and Signer, P. (1996) Quantification of gas
fluxes from the subcontinental mantle: The example of
Laacher See, a maar lake in Germany. Geochim.
Cosmochim. Acta 60, 31–41.
Aeschbach-Hertig, W., Hofer, M., Kipfer, R., Imboden, D.
M. and Wieler, R. (1999) Accumulation of mantle gases
in a permanently stratified volcanic lake (Lac Pavin,
France). Geochim. Cosmochim. Acta 63, 3357–3372.
Aka, F. T. (2000) Noble gas systematics and K-Ar chronol-
ogy: Implications for the Cameroon Volcanic Line, West
Africa. Ph.D. thesis, Okayama University, Japan.
Aka, F. T. (2015) Depth of melt segregation below the Nyos
maar-diatreme volcano (Cameroon, West Africa): Major-
trace element evidence and their bearing on the origin of
CO2 in Lake Nyos. Volcanic Lakes (Rouwet, D.,
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 45
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Christenson, B., Tassi, F. and Vandemeulebrouck, J., eds.),
Springer-Heidelberg.
Aka, F. T. and Yokoyama, T. (2013) Current status of the
debate about the age of Lake Nyos dam (Cameroon) and
its bearing on potential flood hazards. Nat. Hazards 65,
875–885, doi:10.1007/s11069-012-0401-4.
Aka, F. T., Nagao, K., Kusakabe, M., Sumino, H., Tanyileke,
G., Ateba, B. and Hell, J. (2004) Symmetrical helium iso-
tope distribution on the Cameroon Volcanic Line, West
Africa. Chem. Geol. 203, 205–223.
Aka, F. T., Yokoyama, T., Kusakabe, M., Nakamura, E.,
Tanyileke, G., Ateba, B., Ngako, V., Nnange, J. and Hell,
J. (2008) U-series dating of Lake Nyos maar basalts,
Cameroon (West Africa): Implications for potential haz-
ards on the Lake Nyos dam. J. Volcanol. Geotherm. Res.
176, 212–224.
Allard, P. (1983) The origin of hydrogen, carbon, sulfur, ni-
trogen and rare gases in volcanic exhalations: Evidence
from isotopic geochemistry. Forecasting Volcanic Events
(Tazieff, H. and Sabrrorux, J.-C., eds.), Chapt. 25, 337–
386, Developments in Volcanology 1, Elsevier, Amster-
dam.
Alvarado, G. E., Soto, G. J., Salani, F. M., Ruiz, P. and
Hurtado de Mendoza, L. (2011) The formation and evolu-
tion of Hule and Rio Cuarto maars, Costa Rica. J. Volcanol.
Geotherm. Res. 201, 342–356.
Anzidei, M., Carapezza, M. L., Esposito, A., Giordano, G.,
Lelli, M. and Tarchini, L. (2008) The Albano Maar Lake
high resolution bathymetry and dissolved CO2 budget
(Colli Albani volcano, Italy): Constrains to hazard evalu-
ation. J. Volcanol. Geotherm. Res. 171, 258–268.
Aramaki, S., Ohsumi, T., Kanari, S., Kusakabe, M. and
Hirabayashi, J. (1987), Letha gas burst from Lake Nyos,
Cameroon, August 1986. Kazan (Bulletin of the Volcano-
logical Society of Japan) 32, 17–24 (in Japanese).
Asaah, A. N. E., Yokoyama, T., Aka, F. T., Usui, T.,
Wirmvem, M. J., Tchamabe, B. C., Ohba, T., Tanyileke,
G. and Hell. J. V. (2014) A comparative review of
petrogenetic processes beneath the Cameroon Volcanic
Line. Geoscience Frontiers (http://dx.doi.org/10.1016/
j.gsf.2014.04.012).
Asaah, A. N. E., Yokoyama, T., Aka, F. T., Usui, T., Kuritani,
T., Wirmvem, M. J., Iwamori, H., Fozing, E. M., Tamen,
J., Mofor, G. J., Ohba, T., Tanyileke, G. and Hell, J. V.
(2015) Geochemistry of lavas from maar-bearing volca-
noes in the Oku Volcanic Group of the Cameroon Vol-
canic Line. Chem. Geol. 406, 55–69.
Ballentine, C. J., Marty, B., Sherwood Lollar, B. and Cassidy,
M. (2005) Neon isotopes constrain convection and vola-
tile origin in the Earth’s mantle. Nature 433, 33–38.
Barberi, F., Chelini, W., Marinelli, G. and Martini, M. (1989)
The gas cloud of Lake Nyos (Cameroon, 1986). J.
Volcanol. Geotherm. Res. 39, 125–134.
Barfod, D. N., Ballentine, C. J., Halliday, A. N. and Fitton,
J. G. (1999) Noble gases in the Cameroon Line and the
He, Ne, and Ar isotopic composition of high-A (HIMU)
mantle. J. Geophys. Res. 104, 29509–29527.
Baxter, P. J., Kapila, M. and Mfonfu, D. (1989) Lake Nyos
disaster, Cameroon, 1986: the medical effects of large scale
emission of carbon dioxide? British Med. Jour. 298, 1437–
1441.
Burke, K. (2001) Origin of the Cameroon Line of Volcano-
capped swells. J. Geol. 109, 349–362.
Caracausi, A., Nuccio, P. M., Favara, R., Nicolosi, M. and
Paternoster, M. (2009) Gas hazard assessment at the
Monticchio crater lakes of Mt. Vulture, a volcano in South-
ern Italy. Terra Nova 21, 83–87, doi:10.1111/j.1365-
3121.2008.00858.x.
Carapezza, M. L. and Tarchini, L. (2007) Accidental gas
emission from shallow pressurized aquifers at Alban Hills
volcano (Rome, Italy): Geochemical evidence of magmatic
degassing? J. Volcanol. Geotherm. Res. 165, 5–16.
Carapezza, M. L., Lelli, M. and Tarchini, L. (2008)
Geochemistry of the Albano and Nemi crater lakes in the
volcanic district of Alban Hills (Rome, Italy). J. Volcanol.
Geotherm. Res. 178, 297–304.
Chabaux, F. and Allègre, C. J. (1994) 238U-230Th-226Ra dis-
equilibria in volcanics: a new insight into melting con-
straints. Earth Planet. Sci. Lett. 126, 61–74.
Chako Tchamabé, B., Youmen, D., Owona, S., Ohba, T.,
Németh, K., Ngapna, M. N., Asaah, A. N. E., Aka, F. T.,
Tanyileke, G. and Hell, J. V. (2013) Eruptive history of
the Barombi Mbo Maar, Cameroon Volcanic Line, Cen-
tral Africa: Constraints from volcanic facies analysis. Cent.
Eur. J. Geosci. 5(4), 480–496, doi:10.2478/s13533-012-
0147-2.
Chau, H. F., Kwok, P. K. and Mak, L. (1996) A model of gas
buildup and release in crater lakes. J. Geophys. Res.
101(B12), 28,253–28,263.
Chevrier, R. M. (1990) Lake Nyos: phenomenology of the
explosive event of December 30, 1986. J. Volcanol.
Geotherm. Res. 42, 387–390.
Conway, E. J. (1958) Micro Diffusion Analysis and Volu-
metric Error. Macmillan, New York.
Costa, A. and Chiodini, G. (2015) Modelling air dispersion
of CO2 from limnic eruptions. Volcanic Lakes (Rouwet,
D., Christenson, B., Tassi, F. and Vandemeulebrouck, J.,
eds.), 451–465, Springer-Verlag, Berlin, Heidelberg,
doi:10.1007/978-3-642-36833-2_19.
Craig, H. (1961) Isotopic variations in meteoric water. Sci-
ence 133, 1702–1703.
Deruelle, B., Moreau, C., Nkoubou, C., Kambou, R., Lis-
som, J., Njonfang, E., Ghogomu, R. T. and Nono, A. (1991)
The Cameroon line: a review. Magmatism in Extensional
Structural Settings: The Phanerozoic African Plate
(Kampunzu, A. B. and Lubala, R. T., eds.), 275–327,
Springer, Berlin.
Deuser, W. G., Degens, E. T., Harwey, G. R. and Rubin, M.
(1973) Methane in Lake Kivu: New data bearing on its
origin. Science 181, 51–53.
Duan, Z. and Sun, R. (2003) An improved model calculat-
ing CO2 solubility in pure water and aqueous NaCl solu-
tions from 273 to 533 K and from 0 to 2000 bar. Chem.
Geol. 193, 257–271.
Evans, W. C., Kling, G. W., Tuttle, M. L., Tanyileke, G. and
White, L. D. (1993) Gas buildup in Lake Nyos, Cameroon:
The recharge process and its consequences. Appl.
Geochem. 8, 207–221.
Evans, W. C., White, L. D., Tuttle, M. L., Kling, G. W.,
Tanyileke, G. and Michel, R. L. (1994) Six years of change
at Lake Nyos, Cameroon, yield clues to the past and cau-
tions for the future. Geochem. J. 28, 139–162.
Fairhead, J. D. (1988) Mesozoic plate tectonic reconstruc-
tions of central South Atlantic Ocean: the role of West
46 M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
and Central Rift system. Tectonophysics 155, 181–191.
Faivre Pierret, R. X., Berne, P., Roussel, C. and Le Guern,
F. (1992) The Lake Nyos disaster: model calculations for
the flow of carbon dioxide. J. Volcanol. Geotherm. Res.
51, 161–170.
Fantong, W. Y., Fouepe, A. T., Issa, Djomou, S. L. B.,
Banseka, H. S., Anazawa, K., Adelana, S. M. A., Mendjo,
J. W., Aka, F. T., Ohba, T., Hell, J. V. and Nkeng, G. E.
(2013) Temporal pollution by nitrate (NO3) and discharge
of springs in shallow crystalline aquifers: Case of Akok
Ndoue catchment, Yaounde (Cameroon). Afr. J. Environ.
Sci. Technol. 7(50), 175–191.
Fantong, W. Y., Kamtchueng, B. T., Yamaguchi, K., Ueda,
A., Issa, Ntchantcho, R., Mengnjo, M. J., Kusakabe, M.,
Ohba, T., Zhang, J., Aka, F. T., Tanyileke, G. and Hell, J.
V. (2015) Characteristics of chemical weathering and wa-
ter-rock interaction in Lake Nyos dam (Cameroon): Im-
plications for vulnerability to failure and re-enforcement.
J. Afr. Earth Sci. 101, 42–55.
Farrar, C. D., Sorey, M. L., Evans, W. C., Howle, J. F., Kerr,
B. D., Kennedy, B. M., King, C.-Y. and Southon, J. R.
(1995) Forest-killing diffuse CO2 emission at Mammoth
Mountain as a sign of magmatic unrest. Nature 376, 675–
678.
Fitton, J. D. (1980) The Benue Trough and the Cameroon
line—A migrating rift system in West Africa. Earth Planet.
Sci. Lett. 51, 132–138.
Fitton, J. G. and Dunlop, H. M. (1985) The Cameroon Line,
West-Africa, and its bearing on the origin of Oceanic and
Continental alkali basalt. Earth Planet. Sci. Lett. 72, 23–
38.
Fouépé, T. A., Kuitcha, D., Fantong, W. Y., Ewodo, M. G.,
Haris, H. K., Issa and Ohba, T. (2013) Assessing ground-
water nitrate pollution in Yaoundé, Cameroon: Modelling
approach. World App. Sci. J. 23(3), 333–344, doi:10.5829/
idosi.wasj.2013.23.03.321.
Freeth, S. J. (1978) A model for tectonic activity in West
Africa and the Gulf of Guinea during the last 90 m.y. based
on membrane tectonics. Geol. Rundsch 67, 675–688.
Freeth, S. J. (1990) The anecdotal evidence, did it help or
hinder investigation of the Lake Nyos gas disaster? J.
Volcanol. Geotherm. Res. 42, 373–380.
Freeth, S. J. (1994) Lake Nyos: can another disaster be
avoided? Geochem. J. 28, 163–172.
Freeth, S. J. and Kay, R. L. F. (1987) The Lake Nyos gas
disaster. Nature 325, 104–105.
Freeth, S. J. and Rex, D. C. (2000) Constraints on the age of
Lake Nyos, Cameroon. J. Volcanol. Geotherm. Res. 97,
261–269.
Freeth, S. J., Kling, G. W., Kusakabe, M., Maley, J., Tchoua
and Tietze, K. (1990) Conclusions from Lake Nyos disas-
ter. Nature 348, 201.
Gerlach, T. M. (1983) Intrinsic chemical variations in high
temperature volcanic gases from basaltic lavas. Forecast-
ing Volcanic Events (Tazieff, H. and Sabroux, J. C., eds.),
Chapt. 24, 323–336, Elsevier, Amsterdam.
Gerlach, T. M. (2011) Volcanic versus anthropogenic car-
bon dioxide. EOS 92(24), 14.
Giggenbach, W. F. (1990) Water and gas chemistry of Lake
Nyos and its bearing on the eruptive process. J. Volcanol.
Geotherm. Res. 42, 337–362.
Giggenbach, W. F. (1996) Chemical composition of volcanic
gases. Monitoring and Mitigation of Volcanic Hazards,
221–256, Springer.
Gorini, M. A. and Bryan, G. M. (1976) The tectonic fabric
of Equatorial Atlantic and adjoining continental margins:
Gulf of Guinea to northeastern Brazil. Anais da Academia
Brasileira de Ciências 48, 10–119.
Graham, D. (2002) Noble gas isotope geochemistry of mid-
oceanic ridge and ocean island basalts: characterization
of mantle source reservoirs. Noble Gases in Geochemistry
and Cosmochemistry (Porcelli, D., Ballentine, C. J. and
Wieler, R., eds.), Rev. Mineral. Geochem. 47, 247–317,
Geochem. Soc. Mineral. Soc. America, Washington, D.C.
Halbwachs, M. and Sabroux, J.-C. (2001) Removing CO2
from Lake Nyos in Cameroon. Science 292, 436.
Halbwachs, M., Grangeon, J., Sabroux, J.-C. and Villevielle,
A. (1993) Purge par auto-siphon du gaz carbonique dissous
dans le lac Monoun (Cameroun): premiers resultats
experimentaux. C.R. Acad. Sci., Paris 316, Series II, 483–
489.
Halbwachs, M., Sabroux, J.-C., Grangeon, J., Kayser, J. G.,
Tochon-Danguy, J.-C., Felix, A., Beard, J.-C., Villevieille,
A., Vitter, G., Richon, B., Wuest, A. and Hell, J. (2004)
Degassing the “Killer Lakes” Nyos and Monoun,
Cameroon. EOS 85(30), 281–288.
Halliday, A. N., Dickin, A. P., Fallick, A. E. and Fitton, J. G.
(1988) Mantle dynamics: a Nd, Sr, Pb and O isotopic study
of the Cameroon line volcanic chain. J. Petrol. 29, 181–
211.
Halliday, A. N., Davidson, J. P., Holden, P., DeWolf, C., Lee,
D.-C. and Fitton, J. G. (1990) Trace-element fractionation
in plumes and the origin of HIMU mantle beneath the
Cameroon line. Nature 347, 523–528.
Hart, S. R. (1984) A large-scale isotope anomaly in the South-
ern Hemisphere mantle. Nature 309, 753–757.
Holloway, J. R. and Blank, J. G. (1994) Application of ex-
perimental results to C-O-H species in natural melts.
Volatiles in Magmas, Rev. Mineralogy 30, 187–230, Min-
eral Soc. Amer.
Igarashi, G., Ozima, M., Ishibashi, J., Gamo, T., Sakai, H.,
Nojiri, Y. and Kawai, T. (1992) Mantle helium flux from
the bottom of Lake Mashu, Japan. Earth Planet. Sci. Lett.
108, 11–18.
Ihinger, P. D., Hervig, R. L. and McMillan, P. F. (1994) Ana-
lytical methods for volatiles in glasses. Volatiles in Mag-
mas, Rev. Mineralogy 30, 67–121, Mineral Soc. Amer.
Issa, Ohba, T., Fantong, W., Fouepe, A., Chako Tchamabé,
B., Yoshida, Y., Kusakabe, M., Sigha, N., Tsunogai, U.,
Oginuma, Y., Tanyileke, G., Satake, H. and Hell, J. V.
(2013) Contribution of methane to total gas pressure in
deep waters at lakes Nyos and Monoun (Cameroon, West
Africa). Geochem. J. 44, 349–362.
Issa, Fantong, W. Y., Aka, F. T., Ohba, T., Chako Tchamabé,
B., Rouwet, D., Yoshida, Y., Gbetnkom Mouliom, A.,
Sighomnoun, D., Sigha, N., Kusakabe, M., Tanyileke, G.
and Hell, J. V. (2014a)
d
18O and
d
D variation in some
volcanic lakes along the Cameroon Volcanic Line (West-
Africa): Generating an isotopic baseline data for volca-
noes monitoring/surveillance in Cameroon. J. Limnol. 74,
95–113, doi:10.4081/jlimnol.2014.966.
Issa, Ohba, T., Chako Tchamabé, B., Padrón, E., Hernández,
P., Eneke Takem, E. G., Barrancos, J., Sighomnoun, D.,
Ooki, S., Sigha, N., Kusakabe, M. and Yoshida, Y. (2014b)
M. Kusakabe / GEochem. Monogr. Ser. 1: 1–50, 2017 47
doi:10.5047/gems.2017.00101.0001 © 2017 TERRAPUB, Tokyo. All rights reserved.
Gas emission from some Diffuse Diffusion Structures
(DDSs) of the Cameroon Volcanic Line (CVL): Implica-
tion for the prevention of CO2 related hazards. J. Volcanol.
Geotherm. Res. 283, 82–93, doi:10.1016/
j.jvolgeores.2014.07.001.
Johnson, M. C., Anderson, A. T., Jr. and Rutherford, M. J.
(1994) Pre-eruptive volatile contents of magmas. Volatiles
in Magmas, Rev. Mineralogy 30, 281–330, Mineral Soc.
Amer.
Joint UNEP/OCHA Environment Unit (2005) Lake Nyos
Dam Assessment. Embedded in http://www.reliefweb.int/
library/documents/2005/ocha-cmr
Kamgang, P., Chazot, G., Njonfang, E., Ngongang, N. B. T.
and Tchoua, F. M. (2013) Mantle sources and m