Content uploaded by Gladys Melián
Author content
All content in this area was uploaded by Gladys Melián on Aug 16, 2016
Content may be subject to copyright.
Surface CO
2
emission and rising bubble plumes from degassing
of crater lakes in Sa
˜o Miguel Island, Azores
GLADYS MELIA
´N1,2,3*, LUI
´S SOMOZA4, ELEAZAR PADRO
´N1,2,3, NEMESIO
M. PE
´REZ1,2,3, PEDRO A. HERNA
´NDEZ1,2,3, HIROCHIKA SUMINO5, VICTOR
H. FORJAZ6& ZILDA FRANC¸A
6
1
Environmental Research Division, Institute of Technology and Renewable Energies
(ITER), 38611 Granadilla de Abona, Tenerife, Canary Islands, Spain
2
Instituto Volcanolo
´gico de Canarias (INVOLCAN), 38400 Puerto de la Cruz,
Tenerife, Canary Islands, Spain
3
Agencia Insular de la Energı
´a de Tenerife (AIET), 38611 Granadilla de Abona,
Tenerife, Canary Islands, Spain
4
Marine Geological Resources Division, Geological Survey of Spain (IGME),
Rios Rosas 23, Madrid, Spain
5
Department of Basic Science, Graduate School of Arts and Science, The University of Tokyo,
3-8-1 Komaba, Tokyo, Japan
6
Observato
´rio Vulcanolo
´gico e Geote
´rmico dos Ac¸ores (OVGA),
9560-414 Lagoa, Ac¸ores, Portugal
*Corresponding author (e-mail: gladys@iter.es)
Abstract: We report the first detailed study on the types and distributions of active subaqueous
fumaroles and surface diffuse CO
2
degassing in the three main volcanic lakes of Sa
˜o Miguel Island
(Sete Cidades, Fogo and Furnas), Azores archipelago, Portugal. The results of the surveys, carried
out in May 2011 using a floating accumulation chamber and a dual beam 50 and 200 kHz echo
sounder, revealed a very low surface CO
2
degassing at the three lakes, in the range of 32–
608 kg d
21
. However, dense subaqueous degassing plumes were found in the north of the Furnas
crater lake (7.5 –9 plumes per 100 m
2
), and moderate-density degassing in the Fogo (1.5 –2 plumes
per 100 m
2
) and Sete Cidades crater lakes (1 –1.5 plumes per 100 m
2
). The echo sounder detected
hydroacoustic signatures interpreted as acoustic flares, ‘puffing’ bubble plumes or walls of bubbles
associated with numerous subaqueous fumaroles. The recorded echograms show that the bubbles
rise at average speeds of between 19 and 30 cm s
21
at the bottom, with frequencies of release from
1–2 to 31 s. Most subaqueous fumaroles disappear due to the dissolution of CO
2
before reaching
the lake surface. These dissolution processes are enhanced by the pH range observed in the three
volcanic lakes (c. 7–9). Observed dissolved CO
2
values indicate that the pressure of this gas in the
three lakes remained much lower than the hydrostatic pressure and the risk of a limnic eruption is
therefore negligible. We suggest that the rising levels of CO
2
from the subaqueous bubbles could
constitute a critical fuel for subsurface phytoplankton layers, interpreted as horizontal acoustic lay-
ers with high backscattering values. The highest density of subaqueous bubbling correlates with
recent submerged secondary craters formed around the caldera rims of the three Late Quaternary
stratovolcano complexes of Sa
˜o Miguel Island. Our results emphasize the need to perform regular
surface degassing studies as an important volcanic surveillance tool in the Azores archipelago.
Most of the world’s lakes are net sources of CO
2
to
the atmosphere (Cole et al. 1994). This process is
much more evident in volcanic lakes because they
are major sites of emission and condensation for
the volatile elements produced by magmatic activ-
ity. Recent estimates of global CO
2
emissions
from volcanic lakes (c. 117 Mt a
21
) show a similar
value to recent estimated submarine emissions and
are equivalent to 30% of the global degassing of
subaerial volcanism (Pe
´rez et al. 2011). The main
hazard in volcanic lakes is the accumulation of mag-
matic CO
2
that may result in the sudden release of
enormous amounts of dissolved CO
2
gas trapped
at the bottom of lakes. This process is well known
to the scientific community after the Lake Monoun
(1984) and Lake Nyos (1986) disasters, both in
From:Ohba, T., Capaccioni,B.&Caudron, C. (eds) Geochemistry and Geophysics of Active Volcanic Lakes.
Geological Society, London, Special Publications, 437, http://doi.org/10.1144/SP437.14
#2016 The Author(s). Published by The Geological Society of London. All rights reserved.
For permissions: http://www.geolsoc.org.uk/permissions. Publishing disclaimer: www.geolsoc.org.uk/pub_ethics
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
Cameroon (Sigurdsson et al. 1987; Le Guern & Sig-
valdason 1989; Kling et al. 2005; Kusakabe et al.
2008). Degassing at volcanic lakes occurs by an
advective mechanism, in the form of ascending bub-
bles from active subaqueous fumaroles, and by
molecular diffusion through the water –air interface.
The latter is called diffuse degassing and represents
an important proportion of the total CO
2
emission
from volcanic lakes (Pe
´rez et al. 2011). The magni-
tude of the amount of CO
2
diffusion through the
air–water interface depends on the concentration
gradient between the air and the surface water.
Echo sounding methods have been used mainly
on volcanic lakes for bathymetric purposes in
Kawah Ijen Crater Lake, East Java, Indonesia, for
example (Takano et al. 2004), or to map underwater
volcanic structures in the Yellowstone Lake, Wyo-
ming, USA (Morgan et al. 2003). Owing to the
high contrast in the acoustic impedance of water and
the free gas phase, subaqueous fumaroles strongly
scatter the acoustic energy in water (Greinert et al.
2010). Echo sounding techniques also provided use-
ful information on degassing rates and the CO
2
dynamics of the Kelud volcanic lake, East Java,
Indonesia (Caudron et al. 2012).
The Azores archipelago has 88 shallow lakes,
most of which are associated with craters or subsi-
dence calderas. This work presents the first results
of diffuse CO
2
degassing and echo sounding surveys
carried out in May 2011 at the Furnas, Fogo and Sete
Cidades volcanic lakes. These are the main volcanic
craters of Sa
˜o Miguel Island, located in the calderas
of major stratovolcanoes where pre-historical and
historical eruptions (between AD 1439 – 44 and
AD 1630) have occurred. The main goals of this
work are: (1) to study the spatial distribution of dif-
fuse CO
2
degassing; (2) to estimate the total output
of CO
2
emitted into the atmosphere; (3) to analyse
the types and distributions of active subaqueous
fumaroles identified by means of a dual beam 50
and 200 kHz echo sounder (ES); and (4) to study
the possible accumulation of CO
2
in the deep waters
of the three main volcanic crater lakes of the Sa
˜o
Miguel Island (Azores archipelago).
Geological background
The Azores archipelago consists of nine volcanic
islands situated in the North Atlantic Ocean c.
1360 km west of continental Portugal. The geologi-
cal setting of the Azores region is dominated by the
role of the American, Eurasian and African litho-
spheric plate boundaries (Miranda et al. 2014).
The most important tectonic structures recognized
in the area are the Mid-Atlantic Ridge and the Ter-
ceira Rift. Together they are the main source of the
seismic and volcanic activity recorded in the region
(Lourenc¸o et al. 1998). Sa
˜o Miguel (Fig. 1) is the
largest and most populated island of the Azores
archipelago and is located in the eastern part of
the Terceira Rift. The island is formed by several
volcanic edifices situated along a general east –
west axis and is crossed by NW– SE, NE – SW,
WNW– ESE and east – west regional tectonic struc-
tures (Trota 1998). Recent volcanism of the Sa
˜o
Miguel Island is related to three main volcanic cen-
tres: Sete Cidades, Fogo and Furnas (Forjaz 1984;
Moore 1991). During the last 5000 years, the activ-
ity of these volcanic centres has resulted in at least
57 eruptive processes, with five eruptions during
the past 500 years (Moore 1991).
Volcano-hydrothermal fluid discharges in Sa
˜o
Miguel are evident from the existence of low-tem-
perature fumaroles (95–1008C), steaming ground,
thermal springs, cold CO
2
-rich springs and areas
of diffuse degassing soils (Viveiros et al. 2010).
There are several thermal water discharges and
cold CO
2
-enriched springs on Sa
˜o Miguel Island
that discharge mainly from perched water bodies
and have a clear mantle CO
2
signature (Cruz &
Franc¸a 2006).
The Sete Cidades volcano (350 m above sea-
level; a.s.l.) is composed of a semicircular crater
that is slightly elongated in the NW – SE direction
(5.2 ×4.8 km) with a total area of 18.5 km
2
. Within
the Sete Cidade crater, two separated crater lakes are
aligned from SSW to NNE: Lagoa Azul (Blue Lake)
to the north and Lagoa Verde (Green Lake) to the
south (Fig. 1). Lagoa Azul lake is slightly elongated
along the NW direction (2.7 ×2.1 km) covering a
total extension of 3.6 km
2
. Lagoa Verde is also an
elongate lake (5.2 ×4.8 km), covering a total area
of 0.7 km
2
. In this way, the two crater lakes cover
around 25% of the Sete Cidades volcano crater.
The caldera of the Sete Cidades volcano formed
approximately 22 kyr after the eruption of several
cubic kilometres of trachyte pumice. Six vents are
present on the floor of Sete Cidades caldera in a
roughly circular pattern. Other vents may be sub-
merged beneath Lagoa Azul and Lagoa Verde.
The youngest Sete Cidades intracaldera eruption
formed the Caldeira Seca pumice ring. It has a radio-
carbon age of 500 +100 years BP (Shotton & Wil-
liams 1971) and correlates with a reported eruption
at the time of the Portuguese discovery of the island
in the middle of the fifteenth century. Only two
warm springs are known, near Mosteiros on the
northwestern coast and at Ponta da Ferraria on the
western coast (Fig. 1). The latter has a temperature
of about 508C (Moore 1991).
Furnas volcano, a stratovolcano about 800 m
a.s.l, occupies the east-central part of Sa
˜o Miguel.
The Furnas crater lake is located within the Fur-
nas caldera, the third major volcano of the Sa
˜o
Miguel Island (Fig. 1). It is an elongated lake
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
(2.0 ×1.5 km) covering a total extension of
1.9 km
2
. The geographical distribution of the
water types in the Furnas caldera reveals a WSW –
ENE trend in volcanic CO
2
flux (Cruz et al.
1999). Furnas volcano releases much more CO
2
in a non-visible or diffuse way (1100 t d
21
; Viveiros
et al. 2010) than is released by water springs
(4556 t a
21
; Cruz et al. 1999), without taking into
account the contributions from fumaroles, bubbling
pools, cold CO
2
-rich springs or the lake. The
Lagoa das Furnas is located at the western side of
the inner crater. During the last 5 kyr at least 17
trachytic eruptions have taken place at Furnas vol-
cano, including ten eruptions within the inner cal-
dera that formed a trachyte and pumice dome
(Moore 1991). Hot springs are prominent at Furnas
volcano, the distribution of which is shown by Zbys-
zewski et al. (1958). Those on the northern shore of
Lagoa das Furnas are probably related to the nearby
caldera-bounding fault. Hot springs in the village
of Furnas are probably associated with the Furnas
crater or with radial fractures that formed during
the emplacement of the trachyte domes (Booth
et al. 1978).
The Lagoa do Fogo (Fire Lake) is a crater lake
located in the inner part of the Agua de Pau volcano,
a stratovolcano with a prominent 3.0 ×2.5 km cal-
dera with a total area of 7.5 km
2
and walls as high as
300 m (Fig. 1). At least five eruptions of trachyte
pumice have occurred from vents mostly within
the inner caldera during the past 5 kyr (Booth
et al. 1978). There have been two recent eruptions
reported close to the Lagoa do Fogo (Moore
1990). The AD 1563 eruption formed low spatter
ramparts on top of Queimado, a relatively old tra-
chyte dome about 6 km WNW of Lagoa do Fogo
(Moore 1990). In AD 1652, a strombolian eruption
built a cinder cone and extruded flows about
10 km west of Lagoa do Fogo (Weston 1964). Sev-
eral hot springs, with temperatures commonly near
boiling, are located on Agua de Pau, mainly on its
northwestern flank (Zbyszewski et al. 1958). The
hot springs suggest that hot rock or magma associ-
ated with the Late Pleistocene and Holocene erup-
tions is close to the surface (Moore 1991). Fogo
volcano has developed an active geothermal system
that is currently being exploited on its northern flank
to generate over 100 GWh of electricity using geo-
thermal power plants (Wallenstein et al. 2007).
Methods, analysis and data processing
In May 2011 surface diffuse CO
2
efflux and echo
sounder surveys were performed at Furnas, Fogo
and Sete Cidades volcanic lakes, Sa
˜o Miguel Island.
Fig. 1. Map of Sa
˜o Miguel Island, Azores, showing the location of the Sete Cidades, Fogo and Furnas
volcanic lakes.
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
Diffuse CO
2
emission was measured at 102, 80 and
167 sampling sites selected at Furnas, Fogo and Sete
Cidades, respectively, to obtain an almost homoge-
neous pattern distribution following the accumu-
lation chamber method (Parkinson 1981), with the
chamber placed on a flotation device (Huttunen
et al. 2003; Pe
´rez et al. 2011). At each sampling
site pH, temperature and conductivity were mea-
sured at 15 cm depth from the water surface by
means of an Oakton pH/CON 300 meter. Spatial
distribution maps were constructed using sequential
Gaussian simulation (sGs) provided by the sgsim
program (Deutsch & Journel 1998; Cardellini
et al. 2003). The final maps were constructed as
an average of 100 equiprobable realizations per-
formed over a grid of 11 358, 4836 and 3967
squared cells (20 ×20 m) for Sete Cidades, Furnas
and Fogo lakes, respectively, following spherical
variogram models that honour the experimental
variograms.
The ES surveys were carried out by means of a
Lowrance HDS-5 equipped with a dual frequency
(50 and 200 kHz) transducer. Boat speeds range
on average between 0.5 and 3 kn. Digital data from
the ES were processed by means of the Sonar
Viewer 2.1.2 software. Extracted bathymetric data
were converted from the Lowrance-type Mercator
projection to Universal Transverse Mercator using
ArcGIS 10.2. The bathymetric grid and 3D models
of the crater lakes were generated using the Fleder-
maus IVS software. We mainly used the 50 kHz
sampling frequency for the analysis of the data,
because data analysed at 200 kHz missed bubbles
that were detected at 50 kHz, as demonstrated in
Caudron et al. (2012). The density map of the acous-
tic plumes was constructed using the Kernel density
procedure with ArcGIS 10.2. This procedure calcu-
lates the number of plumes per unit of area by using
a kernel function to fit a smoothly tapered surface to
each grid cell (i.e. 10 m). Thus the density of the
acoustic plumes is defined as the number of plumes
per unit of area (e.g. 150 per km
2
or 1.5 per 100 m
2
).
To study the possible water stratification and
CO
2
accumulation in the three volcanic lakes five
vertical profiles of dissolved gas samples were car-
ried out (two in Furnas, one in Fogo and two in
Sete Cidades). The vertical profiles were completed
at the deepest point, located after a short echo
sounding survey. A 2.2 l WaterMark horizontal
PVC water bottle was used to collect water samples
at different depths. For each water sample, chemi-
cal analyses were carried out in the laboratory.
The concentrations of Cl
2
and SO
4
22
were analysed
by means of a Dionex DX-500 ion chromatograph
and HCO
3
2
(the main carbon species) was analysed
by automatic titration with a Metrohm 716 DMS
Titrino. The accuracy of the analysis was estima-
ted at c. 2–3%. Dissolved He, N
2
,CO
2(aq)
and O
2
concentrations were analysed following the method
of Capasso & Inguaggiato (1998), with pure Ne as
the host gas using a quadrupole mass spectrometer
(QMS) Pfeiffer Omnistar 422. The detection limit
for He was estimated to be about 0.05 ppmv,
whereas the analytical accuracy was c. 2.5% and
5% for the main gas components and minor gas
compounds, respectively. Additionally, at the deep-
est point of each profile, the isotopic composition
of dissolved helium was also analysed. Dissolved
gases in these samples were extracted using the
method of Padro
´net al. (2013) and analysed to
determine the concentrations of helium and neon
and the helium isotopic ratios following the method
of Sumino et al. (2001). The results are expressed in
terms of R
A
, where R
A
is the atmospheric
3
He/
4
He
ratio (R
A
¼1.384 ×10
26
; Clarke et al. 1976).
Results and discussion
Sete Cidades crater lakes
Spatial distributions. Figure 2 depicts the spatial
distributions of diffuse CO
2
emission (a), water
pH (b) and temperature (c) values. Low diffuse
CO
2
efflux values were measured at the water sur-
face of Sete Cidades, ranging between non-detected
values (,0.5 g m
22
d
21
, 66% of the data) and
4.1 g m
22
d
21
. The highest values were measured
mainly in the NW coastal areas of Lagoa Azul
(Fig. 2a). The diffuse CO
2
emission, computed as
608 +74 kg d
21
released over an area of 4.5 km
2
,
represents one of the lowest normalized CO
2
emis-
sion rates (135 kg km
22
d
21
) reported for a volca-
nic lake (Pe
´rez et al. 2011). Surface water pH was
slightly basic, ranging between 7.1 and 9.1 with an
average value of 8.1. A slight acidification (pH c.
7.6) can be observed in the surface waters in the
NW coastal area of Lagoa Azul, which suggests a
relatively good spatial correlation between CO
2
emission and pH (Fig. 2b). The water temperature
ranged between 13.38C and 15.98C with an average
value of 15.18C, whereas air temperature was c.
178C. Warmer waters (.158C) were observed in
the middle of Lagoa Azul (Fig. 2c). Water conduc-
tivity showed an almost constant value around
113 mScm
21
without any significant spatial vari-
ability with a relative standard deviation (RSD)
of 0.4%.
Vertical profiles. Two vertical profiles were per-
formed in Sete Cidades crater lake, one in Lagoa
Azul and the other in Lagoa Verde, reaching 21
and 19 m depth, respectively (white stars in Fig.
2). Both profiles showed a similar descending trend
in the pH value from c. 9toc. 7 (Fig. 3a), which
indicates that dissolved CO
2
is mainly present as
HCO
3
2
and CO
2(aq)
. As the upper water layers in
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
both profiles showed the most basic values (c. 9),
eutrophication processes cannot be excluded. The
descending trend in the pH is probably due to
the dissolution of CO
2
that is being released from
the bottom of the lake. This process is more evident
in Lagoa Azul. Figure 3b and f do not show any sig-
nificant trend in either the amount of HCO
3
2
or the
total CO
2
dissolved in both Lagoa Verde and Lagoa
Azul. Bicarbonate values ranged between 30.6 and
39.3 mg l
21
(Fig. 3b). Total dissolved CO
2
(anionic
CO
2
as HCO
3
2
plus CO
2(aq)
) ranged between 36 and
53 mg l
21
(Fig. 3f). Slightly lower total dissolved
CO
2
values (c. 20– 30 mg l
21
) were reported by
Cruz et al. (2006); however, there is no record of
any increase in the CO
2
degassing rate over the
period from 2003 to 2011.
At the sites where the vertical profiles were per-
formed, water temperature descended gradually
with depth from c. 168C at the surface to c. 138C
at the bottom for Lagoa Verde and c. 168C at the
surface to c. 158C at the bottom for Lagoa Azul
(Fig. 3c), which indicates the existence of a dense
stratification and rules out the possible existence
of hot springs at depth at these sites. A thermocline
was not present in the Sete Cidades lake. The deep-
est waters of higher N
2
/O
2
ratios (higher than the
N
2
/O
2
ratio in air-saturated water, ASW) were mea-
sured at the bottom of both Lagoa Verde and Lagoa
Azul, suggesting an addition of endogenous or bio-
genic nitrogen (e.g. benthic nitrogen production).
The
3
He/
4
He ratio analysed in the water of
Lagoa Azul at 21 m depth (the deepest point)
showed a value similar to the atmospheric ratio
(0.95 +0.04R
A
), whereas in Lagoa Verde (at
19 m depth) the result was 0.67 +0.04R
A
.
Echo sounding results
Lagoa Azul crater lake. Figure 4a shows the mor-
phology of the Sete Cidades crater lake. The bathy-
metry of the bottom of Lagoa Azul is split into two
different parts (Fig. 4b). The southern part is com-
posed of a small circular depression that is 450 m
in diameter and 6 m deep, with flank slopes ranging
between 38and 48. The northern part is composed of
a stepped platform ranging from 5 to 22 m deep that
is bound to the north by a flat basin 23– 24 m deep
(Fig. 4b). The platform and the deep basin are con-
nected by an 8 –98slope. The deepest basin of Lagoa
Azul, reaching 25 m deep is located close to the
northern limit of the main crater of the Sete Cidades
volcanic system (Figs 1 & 4b).
Features of bubbles in echograms. The locations
of the observed bubble plumes at the bottom of the
lake are indicated by red dots in Figures 4 and 5a,
b. Most of the bubble plumes identified in the
Lagoa Azul crater lake are located within the deep
basin and along the steep slope that connects the
shallow platform (Fig. 4b). Except for the shallow
areas, the northern part of Lagoa Azul exhibits fre-
quent vertical plumes of bubbles (Fig. 5a, b). The
bottom of the deep basin shows very concentrated
zones that we termed walls of bubbles (Fig. 5c).
These walls or clouds, sourced from the bottom of
the lake, form columns that are 190– 225 m wide
(Fig. 5c). The clouds that are only observed on the
Fig. 2. Spatial distribution of (a) surface CO
2
emission, (b) water pH and (c) water temperature values at the Sete
Cidades volcanic lake. The black dots and white stars indicate the sampling sites and locations of the vertical
profiles, respectively.
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
Fig. 3. (m–r) Fogo volcanic lakes. (f) The variation in total dissolved CO
2
with depth in the Sete Cidades, Furnas and Fogo volcanic lakes, Sa
˜o Miguel. Lagoa Azul, closed
squares; Lagoa Verde, open squares; Furnas N-profile, closed circles; Furnas S-profile, open circles; Fogo, open triangles.
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
50 kHz echogram have medium scattering values
with numerous high-backscattering crescent-shaped
hyperbolas within the clouds when viewed at high
resolution. High-scattering crescent shapes that are
observed between the walls of bubbles are possibly
related to a reduction in the speed of the boat from
3.0 to 0.5 kn that expands the shape of the tar-
gets and forms crescents. Most of the high-density
walls are sourced from the flanks of the steep slopes
(Fig. 5c), reflecting the possibility that active diffuse
degassing is related to fault structures within the cra-
ter lake. A horizontal and strongly stratified 2 – 3 m
thick layer with very high backscatter values
extends over the subsurface of the deep basin of
Lagoa Azul at water depths ranging between 0.5
and 3.0 m (Fig. 5c).
Towards the border of the deep basin of Lagoa
Azul the hydroacoustic manifestations of bubbling
are identified as continuous streams of bubbles
(e.g. Fig. 6a) and intermittent bursts of bubbles
(e.g. Fig. 6b, c). The 50 kHz echograms of these
bubble plumes illustrate the intermittency of the
degassing process or puffing (Fig. 6b, c). Another
characteristic of these bubble plumes is that they
are also observed in the 200 kHz echogram as thin-
ner plumes. Normally, data at 200 kHz miss bubbles
that are detected at 50 kHz. The detection of these
plumes at higher frequencies (with shorter wave-
lengths), reducing the range for observing targets,
might indicate that they are formed by bubbles
with larger diameters (Caudron et al. 2012).
Density and distribution of plumes. The maxi-
mum density of the plumes identified at Lagoa
Azul is located in the north of the lake (Fig. 5a).
The calculated density of the plumes reaches its
highest values of 150 plumes per km
2
in the deepest
basin of Lagoa Azul (Fig. 5b). The total area of the
lake floor of Lagoa Azul that is affected by bubbling
plumes is 2.4 km
2
(Fig. 5b). This area, with a high
density of bubbling plumes, coincides with an
increase in the lake surface water temperature up
to 168C.
Lagoa Verde crater lake. The morphology of
Lagoa Verde (Fig. 4c) is constructed of an asymmet-
rical elongate (1.2 ×0.78 km) basin that is 18 –
21 m deep at its southern side. The western flank of
this basin is constructed of a stepped platform with
water depths ranging from 4 to 18 m. In contrast,
the eastern flank of the deep basin does not show
this stepped morphology. The maximum slope of
this crater lake averages 3– 48on the eastern flank
of the deep basin. The bubble plumes identified in
Lagoa Verde are located along the deep basin and
on the steep eastern flank of this basin (Fig. 5b).
Features of bubbles in echograms. The hydroa-
coustic manifestations of degassing in the Lagoa
Verde crater lake mainly constitute bursts of bub-
bles and isolated flares (Fig. 5d). The bursts of bub-
bles are similar to those observed in Lagoa Azul.
These are composed of ascending trains of bubble
plumes with their tops at approximately at 10, 7
and 6 m below the surface (Fig. 6d). These bubble
plumes are sourced from 12 m deep and disappear
at 6 m below the surface, indicating a rapid process
of bubble dissolution in the water column. This
process gives rise to high-backscatter walls above
the bubble plumes and beneath the subsurface
Fig. 4. 3D bathymetric image of (a) the Sete Cidades
crater lake, (b) Lagoa Azul and (c) Lagoa Verde. The
red dots show the location of bubble plumes.
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
horizontal layer. At these degassing sites the hori-
zontal layer is thickened and ‘bright’ spots (with
very high backscatter values) are identified along
the horizontal layer at the intersection with the dis-
appearing rising bubble plumes (Fig. 6d). Isolated
vertical plumes are also identified as acoustic
Fig. 5. Maps of the (a) location and (b) density of the bubble plumes within the Sete Cidades crater lake.
Echograms (50 kHz) of (c) Lagoa Azul and (d) Lagoa Verde portraying well-defined bubble plumes rising from the
bottom of the lake. h, denotes acoustic hyperbola. In the colour scales in (c) and (d) green to yellow reflects high
concentrations of bubbles (high scattering) and blue reflects low concentrations (low scattering).
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
columns that link the bottom of the lake to the
horizontal layer (Fig. 5d). In contrast to the bubble
plumes, the flares are downward tapering cone
shapes 20 m apart and 11– 12 m in height. One of
the observed differences between the bubble plumes
and isolated flares is that the latter originate from
depressions in the lake.
Density and distribution of plumes. The calcu-
lated density of the plumes ranges between 125
and 150 plumes per km
2
with the maximum density
found in the deepest basin of Lagoa Verde (Fig. 5b).
The total area of the lake floor of Lagoa Verde that is
affected by bubbling acoustic plumes is estimated to
be 0.9 km
2
(Fig. 5b). As occurred in Lagoa Azul,
the pattern of the density maps points to degassing
mostly throughout peripheral secondary craters
within the main caldera complex (Fig. 5b). A third
zone of moderate-rate degassing (ranging between
50 and 75 plumes per km
2
) is located between the
two lakes. This could be related to a small crater
within the Sete Cidades caldera at the intersection
with the NW– SE Mosteiro Graben (Figs 1 & 5b).
This is a pronounced structure on the northwestern
flank of the caldera that is interpreted to be the sub-
aerial segment of the Terceira Rift (Queiroz & Gas-
par 1998).
Furnas crater lake
Spatial distributions. Figure 7 depicts the spatial
distribution of diffuse CO
2
emission values (a),
water pH (b) and temperature (c). As was observed
in Sete Cidades, low diffuse CO
2
efflux values were
measured at the water surface of the Furnas crater
lake, ranging between non-detected values (,0.5
gm
22
d
21
, 82% of the data) and 13.9 g m
22
d
21
.
Although diffuse CO
2
effluxes were low, two con-
siderable degassing sites can be identified (Fig.
7a): the shoreline near the Caldeira das Furnas hot
springs (which occurs in the Furnas lake fumarolic
field and has temperatures approaching the boiling
point of the gas-rich water at surface conditions;
Cruz et al. 1999) and a site located slightly north
of the centre of the lake. Both sites were also char-
acterized by the lowest surface pH values (Fig.
7b). The Furnas lake fumarolic field is an anomalous
diffuse CO
2
degassing structure with CO
2
emission
values of more than 7 kg m
22
d
21
(Viveiros et al.
2010). The highest diffuse CO
2
emission values
measured in the north of the lake are spatially corre-
lated with this diffuse CO
2
degassing structure. The
diffuse CO
2
emission was estimated at 32 +
11 kg d
21
released over an area of 1.9 km
2
, which
represents a lower normalized CO
2
emission rate
Fig. 6. (a) Echogram (50 kHz) used to calculate the speed of the rising bubbles according to the depth (vertical
axis) and time (horizontal axis) for each bubble line in the Lagoa Azul crater lake. The red and yellow dots
indicate that bubbles rise at 19 and 28 cm s
21
, respectively. (b), (c) Echograms (at 50 kHz) illustrating the
intermittency of the degassing process (puffing) in the Lagoa Azul crater lake. Bursts of bubbles are emitted every
1–2 s. (d), (e) Echograms (at 50 kHz) and interpretation, respectively, used to calculate the speed of the rising
bubbles and the frequency of intermittent bubbling in Lagoa Verde. Bursts of bubbles rise at velocities of
30 cm s
21
close to the bottom, reducing their velocity to 15 cm s
21
as they ascend. The reduction in the velocities
is related to the decrease in the bubble diameters as they dissipate. Bubble plumes disappear as they rise through
the water column due to fast dissolution. The frequency of bubble bursts is obtained by calculating the time
between two bubbles at the same water depth. In this case, the time between bubble bursts is 31 s. See the text for
further explanation.
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
(17 kg km
22
d
21
) than that measured in Sete
Cidades in this work. Water temperature ranged
between 13.98C and 15.68C with an average value
of 14.98C, whereas air temperature was c. 178C.
No significant changes in surface water temperature
were observed in the Furnas crater lake (Fig. 7c).
The spatial distribution of water conductivity is
not shown because the values did not show any sig-
nificant spatial variability, with an almost constant
value of c. 150 mScm
21
with an RSD of 1.3%.
Vertical profiles. Two profiles were performed in
the Furnas crater lake, a southern profile (S-profile)
in the deepest part (pink star in Fig. 7) and a northern
profile (N-profile) close to the shoreline near the
Caldeira das Furnas hot springs (white star in
Fig. 7), reaching 9 and 5 m depth, respectively. No
significant trend was observed in the pH of the
waters (Fig. 3g). Water temperature data for both
profiles indicate that the lake was not appreciably
stratified during the study. The S-profile tempera-
ture values decreased with depth from c. 15.68Cat
the surface to 14.68C at the bottom (Fig. 3i). This
behaviour was not present in the N-profile, where
the water temperature showed an almost constant
value of c. 158C (Fig. 3i). The increase in the
HCO
3
2
concentration with depth is more evident in
the N-profile, where the higher CO
2
degassing
rates were measured (Fig. 3h). Total dissolved CO
2
ranged between 40 and 58 mg l
21
(Fig. 3f). Cruz
et al. (2006) reported an almost constant dissolved
CO
2
value of c. 40 mg l
21
.
SO
4
22
and Cl
2
concentrations were studied only
in the S-profile. Although the concentrations of
these anions were almost constant throughout the
profile (Fig. 3j, k), they reached the highest values
at the bottom of the lake. Furnas showed the highest
N
2
/O
2
ratios in the deepest waters, suggesting an
addition of endogenous or biogenic benthic nitrogen
production. The deepest waters of Furnas showed
also discrete enrichments in He (Fig. 3l). Assuming
that the He concentrations in the Furnas crater lake
can be described by the concentration profiles
observed (Fig. 3l), it is possible to estimate the He
emissions at the water surface by applying a pure
diffusive model. According to Fick’s law, the gas
flux at the surface of the water is driven by the con-
centration gradients along the depth and the diffu-
sivity of the gas in water:
Fi=−Di(w)
∂Ci(w)
∂z(1)
where F
i
is the flux of gas i(in kg m
22
s
21
), D
i(w)
is
the diffusion coefficient in water (in m
2
s
21
), C
i(w)
is the concentration of iin water (in kg m
23
) and z
is the depth (in m). In the case of He, we used a
value for D
He(w)
at the average temperature of Fur-
nas Lake waters (158C) of c. 5.45 ×10
29
m
2
s
21
.
This value was calculated as an interpolation of the
values between 108and 208C reported by Verhallen
et al. (1984). The observed concentrations of He
at different depths in both the N- and S-profiles
Fig. 7. Spatial distribution of (a) surface CO
2
emission, (b) water pH and (c) temperature values at the Furnas
volcanic lake. The black dots and stars (white: N-profile and pink: S-profile) indicate the sampling sites and
locations of the vertical profiles, respectively.
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
were fitted with the following exponential decay
function:
[He]w=1.58 ×10−8e−(z/1.9) +4.13 10−7
[He]w=1.02 ×10−7e−(z/4.7) +5.45 10−8(2)
for N- and S-profile, respectively. The derivative
with respect to depth at z¼0 (the water surface)
allows for the estimation of He emissions through
the water surface (1.92 km
2
), in the range of 8 –
20 mg d
21
with a normalized emission value v. the
area of 4.2 –10.4 mg km
22
d
21
. The
3
He/
4
He ratio
analysed at 9 m depth (for the S-profile) was
1.12 +0.04R
A
.
Echo sounding results. The maximum depth of the
Furnas crater lake is 12 m, found at the centre of
the lake (Fig. 8a). The slope of the lake flanks ranges
from 98to 128, whereas at its centre the deepest
basin shows slopes of less than 0.28.
Features of bubbles in echograms. Subaqueous
bubbling in the Furnas crater lake is identified as
hydroacoustic plumes in the water column; i.e.
strong backscatter signals that are 8 – 10 m in height
with a flare-like shape that is rooted to the lake floor
(Figs 8b & 9). Flares are mainly observed on the
slope of the northern flank of the lake at water
depths ranging between 5 and 12 m and are located
close to the field of subaerial hot springs called Cal-
deiras das Furnas (Fig. 8a). Normally these flares
reach the uppermost horizontal layer of the high
backscatter zone located between 3.0 and 0.5 m
deep (Fig. 8b). Most of the flares are sourced depres-
sions or holes in the floor of the lake (Fig. 8b), show-
ing the same morphology as the nearby Caldeiras
das Furnas (Fig. 8b). The subaqueous furnas are
characterized by depressions in the lake floor that
are 1– 3 m deep and 40 –50 m wide and are partially
filled by recent sediments (Fig. 8b).
Density and distribution of plumes. Flares are
mainly concentrated in the northern basin of the Fur-
nas crater lake around the area where the Caldeiras
da Furnas are found (Fig. 8a). The density map
shows that the highest values of 750 – 900 plumes
per km
2
are found in this area (Fig. 9a). However,
the area of the lake floor that is affected by this
high density of plumes is only 0.1 km
2
. Other
areas with a moderate plume density ranging from
350 to 400 plumes per km
2
are also observed in
the centre of the Furnas crater lake (Fig. 9a). Figure
9b shows 3D imaging of the flares rising from the
bottom of the Furnas crater lake for profiles C1
and C2 identified in Figure 9a. At regional scale,
the main density of flares identified at Furnas crater
lake is located along the border of the caldera rim
of the stratovolcanic complex.
Fogo crater lake
Spatial distributions. Figure 10 shows the spatial
distribution of diffuse CO
2
emission values (a),
water pH (b) and temperature (c) at the surface of
the Fogo crater lake. Low diffuse CO
2
effluxes
are also shown, ranging between non-detected
values (,0.5 g m
22
d
21
, 63% of the data) and
3.8 g m
22
d
21
. The highest values of diffuse CO
2
emission (.2gm
22
d
21
) were located in the SE
of the lake (Fig. 10a). Furthermore, the spatial
variability of the surface water pH and temperature
was also very low: i.e. water pH and temperature
fluctuated within the ranges of 7.0 – 7.3 and 12.4 –
12.98C, respectively. For this reason, a different col-
our scale was used for their spatial distribution
(Fig. 10b, c) with respect to that used for the Sete
Cidades and Furnas crater lakes. The highest CO
2
Fig. 8. (a) Location of the bubble plumes within the Furnas crater lake. (b) Echogram (at 50 kHz) showing
well-defined acoustic flares at the northern side of the Furnas crater lake. Some plumes are rooted in depressions
(furnas) at the bottom of the lake.
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
emission (c. 4gm
22
d
21
) located in the SE of the
lake correlated with the water temperature values
(c. 138C). This is the reason why this locality was
selected for the vertical profile (white star in
Fig. 10). The diffuse CO
2
emission was estimated
to be 161 +14 kg d
21
released over an area of
1.59 km
2
, which represents a normalized CO
2
emission rate of 101 kg km
22
d
21
. Again, we do
not show the spatial distribution of water con-
ductivity due to the absence of considerable spatial
variability, with an almost constant value of
c. 42 mScm
21
with an RSD of 1.0%.
Vertical profiles. The profile performed in the SE of
the Fogo crater lake reached 26 m depth (white star
in Fig. 10). No significant trends were observed
in any of the parameters analysed in the profile
(Fig. 3m–r). The total dissolved CO
2
values were
the lowest of the three volcanic lakes studied in
this work, and ranged between 7 and 8 mg l
21
(Fig. 3e), similar to values reported by Cruz et al.
(2006). No evidence of the accumulation of CO
2
at the bottom of the lake was observed. It is worth
noting that although there were no vertical trends
that allowed for the application of a diffusion
model to estimate the emission of He, important
concentrations of this gas (up to 0.4 cm
3
l
21
at stan-
dard temperature and pressure (STP)) were mea-
sured at different depths (Fig. 3r). The
3
He/
4
He
ratio analysed at 20 m depth was 0.84 +0.03R
A
.
Echo sounding results
Features of bubbles in echograms. Two main
types of hydroacoustic features related to bubbling
have been identified in the Fogo crater lake (Fig.
11a): (i) scattered acoustic hyperbolas at various
depths and (ii) strong backscatter signal columns
that are rooted to the lake floor. Two large hydroa-
coustic plumes have been identified in the southern
basin of the Fogo crater lake (Fig. 11b). These
Fig. 9. (a) Density maps of the bubble plumes in the northern part of the Furnas crater lake. (b) 3D image of the
flares rising from the bottom of the Furnas crater lake.
Fig. 10. Spatial distribution of (a) surface CO
2
emission, (b) water pH and (c) temperature values at
the Fogo volcanic lake. The black dots and white stars
indicate the sampling sites and locations of the vertical
profiles, respectively.
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
plumes are characterized by very high backscatter
values, forming 10 m high columns that originate
at the lake floor and terminate 3.0 m below the
lake surface (Fig. 11b). The plumes are connected
by a strong horizontal backscatter signal 1.0–
2.0 m below the surface of the lake.
Density and distribution of plumes. The map
of the density of acoustic anomalies of the Fogo cra-
ter lake shows that the highest density of plumes
(150–200 per km
2
) is located in the northern part
of the lake (Fig. 11c). This high-density bubbling
spot covers a small zone of 0.158 km
2
. At the
same time, a zone of moderate plume density (aver-
aging 75– 100 per km
2
) is also identified at the cen-
tral and deepest part of the lake at water depths of 12
to 28 m (Fig. 11c). The map of the density of the
plumes thus shows a NW– SE trend, which suggests
that the plumes could be related to degasification
along faults with this orientation (Fig. 11c). The
location of the two bubbling plumes identified
in the southern basin of Fogo lake (Fig. 11b) coin-
cides with the zone of increased diffuse CO
2
efflux
(.2gm
2
d
21
; Fig. 10a) and also with a slight
increase in the temperature of the water surface
(.138C, Fig. 10c).
Subaqueous degasification characteristics
of the Sa
˜o Miguel lakes
Several types of hydroacoustic feature are identified
in the three studied crater lakes. We interpret these
features in terms of sequences that reflect different
types of bubbling by subaqueous degasification
(Fig. 12). The two main end-members are: (i) con-
tinuous streams of bubbles (flares and clouds)
sourced from focused or diffuse sources and (ii)
intermittent (crescent-shaped) bursts of bubbles.
The flare-like plumes identified in Furnas are related
to continuous streams of bubbles escaping from a
focused source (Fig. 12a). The columns identified
in the southern basin of Fogo are interpreted as
continuous streams of bubbles but from a diffuse
source (Fig. 12b). The cloud of bubbles originating
from the deep basin of Lagoa Azul is interpreted
as a continuous streams of bubbles with a diffuse
source (Fig. 12c). These continuous streams of
bubbles coincide with areas where a slight increase
in surface temperature is observed, as seen in the
northern part of the Furnas crater lake (Fig. 7c),
the southern basin of the Fogo crater lake (Fig.
10c) and the northern deep basin of Lagoa Azul
(Fig. 2c). We suggest that these continuous plumes
might be formed not only by degassing but also by
the involvement of hydrothermal waters. Otherwise,
intermittent degassing is identified as bursts of
bubbles.
The velocity of rising bubbles can only be
obtained from echograms in Sete Cidades crater
lakes. In the case of continuous bubbling, the rising
speed can be directly measured by picking the time
difference at two distinct depths (yellow and red cir-
cles in Fig. 6a) along the same line defined by the
rising bubbles. The different slopes of the lines
defined by the rising bubbles indicate distinct ris-
ing speeds (Fig. 6a). The progressive bending of a
bubble-line indicates a decrease in speed due to bub-
ble shrinkage (Greinert et al. 2006). The velocities
Fig. 11. (a) Location of the bubble plumes and
bathymetry of the Fogo crater lake. (b) Echogram (at
50 kHz) showing well-defined vertical bubble plumes
linked by a horizontal layer beneath the surface of the
lake in the southern basin of the Fogo crater lake.
(c) Density maps of the bubble plumes in the Fogo
crater lake.
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
Fig. 12. The sequence of the acoustic features and their relationships with degassing processes in the studied crater lakes. The end-members are continuous v. intermittent
degassing and focused v. diffused sources. Further explanations are in the text.
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
of the bubbles from continuous degassing in the
deep basin of Lagoa Azul at 23 m depth range
from 19 to 28 cm s
21
(Fig. 6a). This range of veloc-
ities falls within the typical range reported for rising
bubbles (20 – 30 cm s
21
) with diameters between 5
and 20 mm, respectively (McGinnis et al. 2006).
The splitting and deceleration observed in bubble-
line echograms (Fig. 6a) up through the water col-
umn is interpreted as the result of large bubbles
breaking into small ones or the effect of spatial
divergence of a bubbles in close proximity that had
been released from the bottom almost simultane-
ously but could not be distinguished by the acoustic
system (multiple targets) as proposed Ostrovsky
(2003). When bubbles are not constantly released,
the echograms show intermittent bursts of bubbles
identified as crescent shapes (Fig. 6b –d). The occur-
rence of bubble bursts at the same water depth
allows for the determination of the frequencies at
which the bubbles are released from the lake bot-
tom. From calculations using the echograms,
we observe that the main difference in degassing
observed between Lagoa Azul and Lagoa Verde is
the frequency of release (F
r
) of the bubble bursts.
The frequency of release in Lagoa Verde (F
r
¼
31 s) is much slower than that observed in Lagoa
Azul (F
r
¼1.0–1.6 s) (Fig. 6). In the other crater
lakes (Fogo and Furnas) the crescent-shaped hyper-
bolas are much more smaller, which precludes the
calculation of F
r
.
Echo sounding data revealed a high density of
degassing acoustic plumes in the northern part of
the Furnas crater lake (750 – 900 per km
2
). In con-
trast, similar and moderate values of the density of
the bubble plumes are calculated both in the deep
basin of Lagoa Azul (150 per km
2
) and Lagoa
Verde (125–150 per km
2
), the northern and south-
ern parts of the Sete Cidades crater lake and Fogo
crater lake (150 –200 km
2
). The greatest area of
lake bottom that is affected by degassing corre-
sponds to 2.4 km
2
in the northern Lagoa Azul,
whereas only 0.9 km
2
in the southern Lagoa Verde
is affected. In the Fogo crater lake, the main concen-
tration of bubbling occurs over a very reduced area
(0.158 km
2
) in the southern basin. The lake bottom
area that is affected by bubble plumes in the Furnas
crater lake covers an small area of 0.1 km
2
, mainly
concentrated in the northern part around Caldeira
das Furnas.
The meaning of the horizontal acoustic layer
observed in the three crater lakes (Fig. 5c, d) is
not explained by changes in the geochemical or
physical signatures along the vertical profiles (Fig.
3). We suggest that this high-backscatter layer
could have a biological origin related to the level
of total dissolution of CO
2
at the top of the degasifi-
cation plumes. In this way, the increased levels of
CO
2
from the bubble plumes could trigger the
intensification of phytoplankton blooms at this
level of the water column, as proposed by Verspa-
gen et al. (2014). The eutrophication of the Furnas
and Sete Cidades crater lakes by large blooms of
cyanobacteria and microcystins has been recog-
nized since the 1980s (e.g. Santos et al. 2005). We
thus propose that the eutrophication, by acting
as a ‘biological mask’, might explain the low CO
2
emission rates measured in the ‘latent’ crater
lakes. On the contrary, the observed columns of
submarine degassing bubble plumes of CO
2
may
trigger phytoplankton blooms at horizontal levels
(e.g. Fig. 11b).
Hydrochemistry of the Sa
˜o Miguel lakes
We used the Cl– SO
4
–HCO
3
ternary diagram to
classify the Sete Cidades, Furnas and Fogo volcanic
lake waters on the basis of the major anion concen-
trations (Giggenbach 1991). The Cl– SO
4
–HCO
3
ternary diagram shown in Figure 13 includes the
data from Cruz et al. (2006). With the exception
of the Fogo data, similar observations were made
by Cruz et al. (2006); the dominant anion in the
lake waters was HCO
3
2
. Although these waters
are located within a crater, they plot in the typi-
cal range defined for peripheral waters by Giggen-
bach (1991).
In Figure 14, we show the
3
He/
4
He v.
4
He/
20
Ne
diagram. The fact that the
3
He/
4
He and
4
He/
20
Ne
ratios of the Sete Cidades and Fogo lakes lie on
the mixing line between the atmospheric and crus-
tal end-members suggests that dissolved He is
mainly atmospheric, with a possible minor addition
of crustal (radiogenic) He. These results indicate
negligible mantle He degassing at these lakes. At
Furnas dissolved He showed a slight mantle origin,
as it plots close to the air–mid-ocean ridge basalt
(air– MORB) mixing curve (Fig. 14). The He
emission values estimated for Furnas (4.2 – 10.4
mg km
22
d
21
) are much lower than the soil diffuse
He emission values reported in other volcanic
systems such as Pico do Fogo, Cape Verde (2.8 ×
10
7
mg km
22
d
21
; Dionis et al. 2015); Cumbre
Vieja volcano, Spain (0.8–1.7 ×10
5
mg km
22
d
21
;
Padro
´net al. 2012) and El Hierro Island, Spain
(0.3–1.4 ×10
5
mg km
22
d
21
; Padro
´net al. 2013).
The variation is mainly due to the differences
between the diffusion coefficient of He in water
(5.45 ×10
29
m
2
s
21
) and that for air (7 ×
10
25
m
2
s
21
). The absence of a clear concentration
gradient prevented the theoretical estimation of He
emissions using Fick’s law in the Fogo and Sete
Cidades crater lakes.
The highest diffuse CO
2
emission rates coin-
cided spatially with the highest density of bubbles
in the Furnas volcanic lake. The total CO
2
emissions
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
were estimated from the floating accumulation
chamber method and were found to range from
32 to 608 kg d
21
, representing some of the lowest
CO
2
emission rates so far reported for a volcanic
lakes (Pe
´rez et al. 2011). In the case of Furnas, the
emission value contrasts with the rest of the Furnas
volcano, which is considered one of the highest
emitters of CO
2
with a normalizing emission rate
of c. 186 000 kg km
22
d
21
(Viveiros et al. 2010).
Andrade et al. (2016) recently reported a much
higher CO
2
emission rate from the water surface
of Furnas (c. 321 000 and c. 28 000 kg km
22
d
21
for surveys performed in October– November 2013
and March 2014, respectively). This significant
Fig. 13. Classification of the waters from the Sete Cidades, Furnas and Fogo volcanic lakes on the basis of the
relative Cl
2
,SO
4
22
and HCO
3
2
content. Data from Cruz et al. (2006) are displayed as black symbols.
Fig. 14.
3
He/
4
He v.
4
He/
20
Ne diagram for dissolved gas from the bottom of the Lagoa Azul, Lagoa Verde, Furnas
(N- and S-profiles) and Fogo crater lakes. The grey curves indicate the calculated binary mixing lines between
ASW, crustal (
3
He/
4
He ¼0.01R
A
) and mantle (
3
He/
4
He ¼8+1R
A
, Graham 2002) end-member compositions,
which have assumed
4
He/
20
Ne ratios of 1 ×10
6
and 1 ×10
3
, respectively.
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
increase does not seem to be caused by the mono-
mictic character of the Furnas crater lake and should
be attributed to changes in the volcanic activity of
the Furnas volcanic system.
These results suggest that the CO
2
bubble
plumes in the three volcanic lakes studied here dis-
solve almost completely and very rapidly, as has
been observed in other volcanic lakes (Caudron
et al. 2012). Echo sounding results confirm that
most of the subaqueous fumaroles fade out by disso-
lution of CO
2
before reaching the lake surface. The
pH range observed in the three volcanic lakes is
c. 7–9 and almost all total dissolved aqueous CO
2
undergoes ionization in HCO
3
2
and CO
3
22
. Thus dis-
solution processes also minimize the molecular
diffusion of CO
2
through the water– air interface,
considerably reducing the CO
2
emission to the
atmosphere. Emission rate values, normalized per
unit area, were very similar for Fogo and Sete
Cidades (101 – 134 kg m
22
d
21
) and higher than
that of Furnas (17 kg m
22
d
21
). This one order of
magnitude difference may be due to the persis-
tent advective discharge of volcano-hydrothermal
gases through fumaroles and bubbling pools in Cal-
deira das Furnas, which represents a preferred path
for degassing. As the isotopic composition of CO
2
was not analysed in this study, biogenic CO
2
emis-
sion cannot be ruled out and prevents differentiation
between a biogenic source and deep degassing.
Although most of these observed acoustic
bubbles are related to CO
2
degasification, the occur-
rence of hydrothermal waters in the vents is cannot
be ruled out. The upward migration of water due to
thermal heating in some areas (such as the deep
basin of Lagoa Azul, southern Fogo and northern
Furnas) could explain the relative anomalies of the
lake surface temperatures and the types of acous-
tic features. Therefore, the area with the highest
density of bubbling in Lagoa Azul (Sete Cidades)
coincides (Fig. 4b) with the positive temperature
anomaly mapped on the surface waters of the lake
(Fig. 2c). The northern area of the Furnas crater
lake shows spots of anomalous temperatures and
relatively low pH close to Caldeiras das Furnas
(Fig. 7). Furthermore, in this area flares are sourced
from carved depressions in the lake bottom, suggest-
ing the movement of hydrothermal waters might
be responsible for the formation of these erosional
features (Fig. 8).
Degassing at the three crater lakes at present has
a close relationship with recent secondary craters
that formed around the caldera rims of the three
Late Quaternary stratovolcano complexes of Sa
˜o
Miguel (e.g. Moore 1990): i.e. Sete Cidades, Fogo
and Furnas. This is indicated by the dense popula-
tion of subcircular acoustic plumes mapped within
the lakes. Therefore, the shape of the highest density
area of bubble plumes identified at Lagoa Azul and
Lagoa Verde is formed in association with ring cra-
ters in a roughly circular pattern within the caldera
rim of Sete Cidades (Fig. 5a). The same occurs in
the Furnas crater lake, where the main degassing
activity is probably associated with a ring crater
along the northern rim of the caldera of the Furnas
stratovolcano (Fig. 8a).
Conclusions
In this work we analysed the subaqueous degassing
from the bottom of three crater lakes by means of a
dual beam 50 and 200 kHz echo sounder as part of a
multidisciplinary study. Data from the echo sounder
allowed for the detection of a wide range in the
intensity of subaqueous degassing that we have
classified into two categories: continuous streams
to intermittent bursts of bubbles and focused to
diffuse degassing. Continuous streams of bubbles
released from focused areas of the lake floor were
identified as hydroacoustic flare-like shapes with
high backscatter signals that rise to the near surface
of the lake. Examples of these focused degassing
flares up to 10 m high have been detected in the
northern basin of the Furnas crater lake close to sub-
aerial hot springs and in the southern basin of Fogo.
Continuous degassing sourced from diffuse areas
was identified as clouds of bubbles. This type of
bubbling was observed mainly from the deep basin
of Lagoa Azul, rising up to 23 m into the water col-
umn. The average speed at which the bubble streams
rise ranged from 19 –28 cm s
21
, and large bubbles
were shown to broke up into small ones and decel-
erate on their way up from the bottom. Intermittent
bursts of single/multiple bubbles (puffing) were
identified as large crescent-like hyperbolas rising
into the water column.
The F
r
value ranged from 1 to 2 s in Lagoa Azul
to 31 s in Lagoa Verde of the Sete Cidades lake.
The echograms show the bubble(s) ascend at speeds
of 30 cm s
21
near the lake floor, decelerating to
15 cm s
21
at mid-height in the water column and
dissipating by dissolution at c. 2 m below the lake
surface. The degassing processes are mainly
focused on secondary craters that formed near the
caldera rims of the three Late Quaternary stratovol-
cano calderas of Sa
˜o Miguel. Therefore, the highest
density of subaqueous fumaroles we have mapped is
c. 7.5– 9 plumes per 100 m
2
within the northernmost
basin of the Furnas caldera rim related to subaerial
hot springs. Otherwise, we measured a moderate
density of subaqueous fumaroles (c. 1.5– 2 plumes
per 100 m
2
) within the Fogo lake near the caldera
rim of the Agua de Pau stratovolcano and c. 1 – 1.5
plumes per 100 m
2
within the deep basin of Lagoa
Azul located near the caldera rim of the Sete
Cidades stratovolcano.
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
Although there is an evident gas discharge from
the bottom of the three lakes, the observed dis-
solved CO
2
values indicate that the pressure of
this gas in the three lakes remains much lower
than the hydrostatic pressure and the risk of lim-
nic eruption is negligible. The acoustic horizontal
plumes that appear in the three crater lakes at 1 –
2 m beneath the lake surfaces are related to subaqu-
eous fumaroles but cannot be explained by changes
in the geochemical or physical signatures along
the vertical profiles. This high-backscatter acoustic
layer could have a biological origin related to CO
2
dissolution of the rising bubbles. Therefore, the
CO
2
from the dissolution of rising bubbles (a pro-
cess that is enhanced by the pH range of c. 7–9)
could be a factor that triggers the intensification
of phytoplankton blooms at the subsurface of the
lake as proposed by Verspagen et al. (2014). Eutro-
phication might act as a biological mask and thus
might explain the low surface CO
2
degassing
measured at the three lakes, in the range of 32 –
608 kg d
21
. Following this assumption, the elimi-
nation of this biological mask may reveal a drastic
increase of CO
2
released to the atmosphere from
this type of quiescent crater lakes (according the
classification of Varekamp et al. (2000). Our results
emphasize the need to perform regular surface
degassing and hydroacoustic studies as an impor-
tant volcanic surveillance tool in the Azores archi-
pelago. These studies are of special interest in the
case of Furnas due to the recent significant increase
in the emission rate of CO
2
reported by Andrade
et al. (2016).
This work was supported by the following projects: (i)
MAKAVOL (MAC/3/C161), co-financed by the Euro-
pean Union Transnational Cooperation Programme MAC
2007-2013; (ii) SUBVENT (CGL2012-39524-C02-02)
financed by the R +D+I Spanish National Plan 2008–
2011; and (iii) CO2BASAVOL (SolSubC200801000385)
financed by the Canary Islands Agency for Research, Inno-
vation and Information Society (ACIISI) of the Govern-
ment of the Canary Islands, as well as the Observato
´rio
Vulcanolo
´gico e Geote
´rmico dos Ac¸ores (OVGA) and
the Cabildo Insular de Tenerife. The authors are grateful
to Helen Robinson for her help with writing in English.
References
Andrade, C., Viveiros, F., Cruz, J.V., Coutinho,R.&
Silva, C. 2016. Estimation of the CO
2
flux from Furnas
volcanic Lake (Sa
˜o Miguel, Azores). Journal of Volca-
nology and Geothermal Research,315, 51– 64.
Booth, B., Croasdale,R.&Walker, G.P.L. 1978. A
quantitative study of five thousand years of volcanism
on Sa
˜o Miguel, Azores. Royal Society of London Phil-
osophical Transactions,288, 271– 319.
Capasso,G.&Inguaggiato, S. 1998. A simple method
for the determination of dissolved gases in natural
waters. An application to thermal waters from Vulcano
Island. Applied Geochemistry,13, 631– 642.
Cardellini, C., Chiodini,G.&Frondini, F. 2003.
Application of stochastic simulation to CO
2
flux from
soil: mapping and quantification of gas release. Journal
of Geophysical Research: Solid Earth,108, 2425,
http://doi.org/10.1029/2002JB002165
Caudron, C., Mazot,A.&Bernard, A. 2012. Carbon
dioxide dynamics in Kelud volcanic lake. Journal of
Geophysical Research Solid Earth,117, B05102,
http://doi.org/10.1029/2011JB008806
Clarke, W.B., Jenkins, W.B. & Top , Z. 1976. Determina-
tion of tritium by mass spectrometric measurement of
3
He. The International Journal of Applied Radiation
and Isotopes,27, 515– 522, http:// doi.org/10.1016/
0020-708X(76)90082-X
Cole, J.J., Caraco, N.F., Kling, G.W. & Kratz, T.K.
1994. Carbon dioxide supersaturation in the surface
waters of lakes. Science,265, 1568–1570.
Cruz, J.V. & Franc¸a, Z. 2006. Hydrogeochemistry of
thermal and mineral water springs of the Azores archi-
pelago (Portugal). Journal of Volcanology and Geo-
thermal Research,151, 382– 398.
Cruz, J.V., Coutinho, R.M., Carvalho, M.R., Oskars-
son,N.&Gislason, S.R. 1999. Chemistry of waters
from Furnas volcano, Sa
˜o Miguel, Azores: fluxes of vol-
canic carbon dioxide and leached material. Journal of
Volcanology and Geothermal Research,92, 151– 167.
Cruz, J.V., Antunes, P., Amaral, C., Franc¸a,Z.&
Nunes, J.C. 2006. Volcanic lakes of the Azores archi-
pelago (Portugal): geological setting and geochemical
characterization. Journal of Volcanology and Geother-
mal Research,156, 135– 157.
Deutsch,C.&Journel, A. 1998. GSLIB: Geostatistical
Software Library and User’s Guide. Oxford University
Press, New York.
Dionis, S., Melia
´n,G.
et al
. 2015. Diffuse volcanic gas
emission and thermal energy release from the summit
crater of Pico do Fogo, Cape Verde. Bulletin of Volca-
nology,77, 1– 13, http:// doi.org/10.1007/s00445-
014-0897-4
Forjaz, V.H. 1984. Sa
˜o Miguel Volcanostratigraphic
Sketch. Geosciences Department, University of the
Azores, Ponta Delgada [in Portuguese].
Giggenbach, W.F. 1991. Chemical techniques in geother-
mal exploration. In:D’Amore, F. (ed.) Application of
Geochemistry in Geothermal Reservoir Development.
UNITAR/UNDP Publication, Rome, 119– 142.
Graham, D.W. 2002. Noble gas isotope geochemistry
of Mid-Ocean Ridge and Ocean Island Basalts:
characterization of mantle source reservoirs. In:
Porcelli, D., Ballentine, C.J. & Wieler, R. (eds)
Noble Gases in Geochemistry and Cosmochemistry:
Reviews in Mineralogy and Geochemistry, 47. Geo-
chemical Society and Mineralogical Society of Amer-
ica, Washington D.C., 247– 317.
Greinert,J.,Artemov,Y.,Egorov,V.,De Batist,M.&
McGinnis, D. 2006. 1300-m-high rising bubbles from
mud volcanoes at 2080m in the Black Sea: hydroacoustic
characteristics and temporal variability. Earth and Plan-
etary Science Letters,244, 1– 15, http://doi.org/10.
1016/j.epsl.2006.02.011
Greinert, J., McGinnis, D.F., Naudts, L., Linke,P.&
De Batist, M. 2010. Atmospheric methane flux
CO
2
DEGASSING, SAO MIGUEL VOLCANIC LAKES
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
from bubbling seeps: spatially extrapolated quantifica-
tion from a Black Sea shelf area. Journal of Geophys-
ical Research,115, C01002, http://doi.org/10.1029/
2009JC005381
Huttunen, J.T., Alm,J.
et al
. 2003. Fluxes of methane,
carbon dioxide and nitrous oxide in boreal lakes and
potential anthropogenic effects on the aquatic green-
house gas emissions. Chemosphere,52, 609– 621.
Kling, G.W., Evans, W.C., Tanyileke, G., Kusakabe,
M., Ohba, T., Yoshida,Y.&Hell, J.V. 2005. Degas-
sing Lakes Nyos and Monoun: defusing certain disas-
ter. Proceedings of the National Academy of Sciences
of the United States of America,102, 14185– 14190,
http://doi.org/10.1073/pnas.0502274102
Kusakabe, M., Tanyileke, G.Z., McCord, S.A. &
Schladow, S.G. 2008. Evolution of CO
2
in Lakes
Monoun and Nyos, Cameroon, before and during con-
trolled degassing. Geochemical Journal,42, 93– 118.
Le Guern,F.&Sigvaldason, G.E. 1989. The Lake
Nyos event and natural CO
2
degassing I. Journal of
Volcanology and Geothermal Research,39, 95–275.
Lourenc¸o, N., Miranda, J.M., Luis, J.F., Ribeiro, A.,
Mendes Victor, L.A., Madeira,J.&Needham,
H.D. 1998. Morpho-tectonic analysis of the Azores vol-
canic plateau from a new bathymetric compilation of
the area. Marine Geophysical Research,20, 141– 156.
McGinnis, D.F., Greinert, J., Artemov, Y., Beaubien,
S.E. & Wu
¨est, A. 2006. Fate of rising methane bubbles
in stratified waters: how much methane reaches the
atmosphere? Journal of Geophysical Research,111,
C09007, http://doi.org/10.1029/2005JC003183
Miranda, J.M., Luis, J.F., Lourenc¸o,N.&Goslin,J.
2014. Distributed deformation close to the Azores
Triple ‘Point’. Marine Geology,355, 27– 35.
Moore, R.B. 1990. Volcanic geology and eruption fre-
quency, Sa
˜o Miguel, Azores. Bulletin of Volcanology,
52, 602– 514.
Moore, R.B. 1991. Geology of the three quaternary stra-
tovolcanoes on Sa
˜o Miguel, Azores. US Geological
Survey Bulletin 1900.
Morgan, L.A., Shanks III, W.C.
et al
. 2003. Explora-
tion and discovery in Yellowstone Lake: results from
high-resolution sonar imaging, seismic reflection pro-
filing and submersible studies. Journal of Volcanology
and Geothermal Research,122, 221– 242, http:// doi.
org/10.1016/S0377-0273(02)00503-6
Ostrovsky, I. 2003. Methane bubbles in Lake Kinneret:
quantification and temporal and spatial heterogeneity.
Limnology and Oceanography,48, 1030– 1036.
Padro
´n, E., Pe
´rez, N.M.
et al
. 2012. Helium emission
at Cumbre Vieja volcano, La Palma, Canary Islands.
Chemical Geology,312– 313, 138 – 147, http:// doi.
org/10.1016/j.chemgeo.2012.04.018
Padro
´n, E., Pe
´rez, N.M.
et al
. 2013. Diffusive he-
lium emissions as a precursory sign of volcanic
unrest. Geology,41, 539– 542, http://doi.org/10.
1130/G34027.1
Parkinson, K.J. 1981. An improved method for measur-
ing soil respiration in the field. Journal of Applied
Ecology,18, 221– 228.
Pe
´rez, N.M., Herna
´ndez, P.A.
et al
. 2011. Global CO
2
emission from volcanic lakes. Geology,39, 235–238.
Queiroz,G.&Gaspar, J.L. 1998. The geology of
Sete Cidades volcano, S. Miguel island, Azores. In:
EC Advanced Study Course, Volcanic Hazard Assess-
ment, Monitoring & Risk Mitigation, Ponta Delgada,
Sa
˜o Miguel.
Santos, M.C.R., Medeiros Pacheco, D.M., Santana,F.
&Muelle, H. 2005. Cyanobacteria blooms in Sete-
Cidades lake (S. Miguel Island – Azores). Algological
Studies 117 (Cyanobacterial Research 6), 393– 406.
Sigurdsson, H., Devine, J.D., Tchoua, F.M., Presser,
T.S., Pringle, M.K.W. & Evans, W.C. 1987. Origin
of the lethal gas burst from Lake Monoun, Cameroun.
Journal of Volcanology and Geothermal Research,31,
1 –16, http:// doi.org/10.1016/0377-0273(87)90002-3
Shotton, F.W. & Williams, R.E.G. 1971. Birmingham
University radiocarbon dates, V. Radiocarbon,13,
141–156.
Sumino, H., Nagao,K.&Notsu, K. 2001. Highly sensi-
tive and precise measurement of helium isotopes using
a mass spectrometer with double collector system.
Journal of the Mass Spectrometry Society of Japan,
49, 61– 68.
Takano, B., Suzukia,K.
et al
. 2004. Bathymetric
and geochemical investigation of Kawah Ijen Crater
Lake, East Java, Indonesia. Journal of Volcanology
and Geothermal Research,135, 299– 329.
Trota, A.N. 1998. Main alignments in Sa
˜o Miguel island
obtained from aerial photo-interpretation. In:Volcanic
Hazard Assessment, Monitoring & Risk Mitigation.EC
Advanced Study Course, Ponta Delgada, Sa
˜o Miguel.
Varekamp, J.C., Pasternack, G.B. & Rowe, G.L., JR.
2000. Volcanic lake systematics II. Chemical con-
straints. Journal of Volcanology and Geothermal
Research,97, 161– 179.
Verhallen, P.T.H.M., Oomen, L.J.P., Elsen, A.J.J.M. &
Kruger, A.J. 1984. The diffusion coefficients of
helium, hydrogen, oxygen and nitrogen in water from
the permeability of a stagnant liquid layer in the
quasi-steady state. Chemical Engineering Science,
39, 1535– 1541.
Verspagen, J.M.H., Van de Waal, D.B., Finke, J.F.,
Visser, P.M., Van Donk,E.&Huisman, J. 2014. Ris-
ing CO
2
levels will intensify phytoplankton blooms
in eutrophic and hypertrophic lakes. PLoS ONE,9,
e104325,http://doi.org/10.1371/journal.pone.0104325
Viveiros, F., Cardellini, C., Ferreira, T., Caliro, S.,
Chiodini,G.&Silva, C. 2010. Soil CO
2
emissions
at Furnas volcano, Sa
˜o Miguel Island, Azores archipel-
ago: volcano monitoring perspectives, geomorpho-
logic studies, and land use planning application.
Journal of Geophysical Research: Solid Earth,115,
http://doi.org/10.1029/2010JB007555
Wallenstein, N., Duncan, A., Chester,D.&Mar-
ques, R. 2007. Fogo Volcano (Sa
˜o Miguel, Azores):
a hazardous edifice. Ge
´omorphologie: Relief, Proces-
sus, Environnement,3/2007, http://geomorphologie.
revues.org/2853, http://doi.org/10.4000/geomorpho
logie.2853
Weston, F.S. 1964. List of recorded volcanic eruptions in
the Azores with brief reports. Boletim do Museu e Lab-
oratorio Mineralogico e Geologico da Faculdade de
Ciencias,10, 3– 18.
Zbyszewski, G., De Medeiros, A.C., De Almeida, F.M.
&Da Veiga Ferreira, O. 1958. Carta Geologica de
Portugal na escala de 1:50 000 (Sa
˜o Miguel). Servicos
Geologicos de Portugal, Lisbon.
G. MELIA
´NET AL.
by guest on August 13, 2016http://sp.lyellcollection.org/Downloaded from
