Hydrologic and geochemical survey of the lake “Specchio di Venere” (Pantelleria island, Southern Italy)

ORIGINAL ARTICLE

Hydrologic and geochemical survey of the lake‘‘Specchio di Venere’’ (Pantelleria island, Southern Italy)

A. Aiuppa Æ W. D’Alessandro Æ S. Gurrieri ÆP. Madonia Æ F. Parello

Received: 24 January 2007 / Accepted: 23 February 2007 / Published online: 16 March 2007

� Springer-Verlag 2007

Abstract Hydrological and geochemical studies per-

formed on Lake Specchio di Venere on Pantelleria island

(Italy) indicate that this endorheic basin has been formed

through upwelling of the water table, and that it is con-

tinuously fed by the thermal springs situated on its shores.

The lake is periodically stratified both thermally and in

salinity, albeit this stratification is rather unstable over

time, since meteorological events such as strong rain or

wind can determine the mixing of its waters. Periodical

analyses of the lake water chemistry show large variations

of the salt content due to the yearly evaporation-rain

dilution cycle. These processes are also responsible for the

saline stratification during steady meteorological condi-

tions. The mineralogical characterisation of the bottom

sediments shows the almost exclusive presence of neofor-

mation minerals, mainly carbonates, formed in response to

the pH gradient between spring- (pH � 6) and lake-waters

(pH � 9). Finally, the CO2 partial pressures in the lake

water slightly exceeding the atmospheric one, are due to

the large amounts of CO2 brought to the lake through the

bubbling free gas phase of the thermal springs. Neverthe-

less the high pH value of the lake water, its small volume

and its periodical mixing prevent dangerous built up of this

gas.

Keywords Pantelleria island � Volcanic lake � Gas hazard

Introduction

Volcanic lakes are a common feature of many active vol-

canoes (Varekamp et al. 2000). They range in composition

from very dilute, meteoric-water dominated to hyper-acid

brines, reflecting variations in the composition and flux of

volcanic or hydrothermal fluids into the lake, with super-

imposed dilution and evaporation effects (Varekamp et al.

2000; Marini et al. 2003). These lakes are known to be

potentially dangerous, since accumulation of CO2 and CH4

in the deep waters, followed by overturning after an

external cause (e.g., earthquakes, landslides, cold rains,

strong winds), may give rise to catastrophic gas exsolution.

Events of this kind occurred in the 1980s from the

Cameroonian Lakes Nyos and Monoun, causing about

1,800 casualties (Sigurdsson et al. 1987; Le Guern and

Sigvaldason 1990). In Italy, gas accumulation takes place

in the deep waters of Lake Piccolo of Monticchio, Vulture

volcano (Chiodini et al. 2000) and of Lake Albano (Cioni

et al. 2003).

Lake ‘‘Specchio di Venere’’ (also known as ‘‘Bagno

dell’Acqua’’) is an endorheic saline lake within a calderic

depression on Pantelleria (Fig. 1), a quiescent volcanic

island in the Sicily Channel, between Tunisia and Sicily.

The island is for its entirety covered by volcanic products

(Mahood and Hildreth 1986) of both effusive (mainly

basalts) and explosive activity (trachytes to peralkaline

rhyolites). The volcanic system of Pantelleria is still active,

with a probable resumption of eruptive activity within the

next 2000 years (Civetta et al. 1988). At present, volcanic

activity is limited to low temperature fumarolic emissions

and thermal springs characterized by temperatures up to

A. Aiuppa � F. Parello

Dipartimento CFTA, Universita di Palermo, via Archirafi 36,

90123 Palermo, Italy

A. Aiuppa � W. D’Alessandro (&) � S. Gurrieri �P. Madonia

Istituto Nazionale di Geofisica e Vulcanologia,

Sezione di Palermo, via U. La Malfa 153,

90146 Palermo, Italy

e-mail: [email protected]

123

Environ Geol (2007) 53:903–913

DOI 10.1007/s00254-007-0702-1

90�C (Parello et al. 2000). In 1992, four exploratory wells

ascertained the presence of an exploitable 1,500–2,000 m

deep geothermal field at temperatures higher than 250�C

(Squarci et al. 1994). Furthermore, a recent study evaluated

the huge output of magmatic CO2 at Pantelleria, with one

of the most degassing areas of the island being the sur-

roundings of the Lake (Favara et al. 2000).

The aim of this work is to review hydrological, geo-

chemical and mineralogical aspects of the lake ‘‘Specchio

di Venere’’. An hydrological balance of the lake is pre-

sented, and the results of two extended surveys of the lake

waters are shown in the attempt to investigate the three-

dimensional distribution of major dissolved ions and

chemico-physical parameters. The relationships between

the lake and feeding thermal springs, and between the

dissolved species and the mineralogy of its sediments, were

also studied. The temporal evolution of the chemical and

chemical–physical parameters of the lake waters has been

studied through both monthly samplings during a period of

12 months and through continuous temperature registration

at four different depths during a period of 3 months. The

whole acquired dataset is used to evaluate the potential gas

hazard in the lake’s area.

Study area and methods

Lake description

Lake ‘‘Specchio di Venere’’ (also known as ‘‘Bagno

dell’Acqua’’) is an endorheic saline lake located inside a

calderic depression of Pantelleria Island (Fig. 1). The

calderic depression was named ‘‘Caldera Cinque Denti’’

by Mahood and Hildreth (1986) and was formed during the

‘‘Green Tuff’’ ignimbrite eruption. The products erupted

during this explosive event covered the island completely

and have been dated at about 50,000 years BP. (Civetta

et al. 1984). Sodarhyolitic and sodatrachytic lavas and

domes 16,000 years old also crop out on the shores of the

lake (Civetta et al. 1984).

The SW part of the lake is characterized by a quite

intense exhaling hydrothermal activity, long recognized

and documented in the literature (Foerstner 1881; Wash-

ington 1913; Rittmann 1967). Exhaling gases mainly

consist of CO2 (98%) and low percentages of N2, O2, Ar,

CH4, H2 and He (D’Alessandro et al. 1994; Parello et al.

2000). The lake has a sub-circular shape and is about

450 m long and 350 m wide. In a bathymetric survey,

X8

X7

X6

X5

X4

X3

X2

X1

X0

Z13Z12

Z11Z10

Z9Z8

Z7Z6

Z5Z4

Z3Z2

Z1

200m

SicilyChannel

Lake

Caldera

Feedingwatershed

Mt.Gelfiser

400

300

394

200

200

100

100

1000m

Sicily Channel

SICILY

N

Pantelleria

Malta

200km

12°E

38°N

Aeolianislands

Tyrrhenian Sea

TU

NIS

IAFig. 1 Map of the Lake Specchio di Venere with bathymetry and location of sampling sites (Black circles May 1996; white circles October

1996). Grey circles shows the position of the thermal springs and the grey square that of the mofette. Top right inset shows the location of the

island of Pantelleria, while the bottom right inset shows the extension of the watershed that feeds the lake

904 Environ Geol (2007) 53:903–913

123

Bocchi et al. (1988) measured a maximum depth of

12.5 m. The SW area is characterized by shallow depths

with gently sloping bottom covered by whitish sediments,

while in the NE one the slopes are much steeper (Fig. 1).

Sub aqueous inspection highlighted the presence of a dark

coloured layer of decaying organic matter covering the

bottom of the lake in its deeper part. Its thickness is vari-

able, probably due to organic productivity within the lake’s

water (in May 1996, it exceeded 50 cm of thickness, while

in October of the same year it has been estimated in about

10 cm). This layer is followed by about 2 m of dark grey

uncompacted sediments with water contents higher than

50% by weight. Beneath this level, sediments become

progressively more compacted.

The lake’s surface and the altitude above sea-level are

variable with time. On the map of the ‘‘Istituto Geografico

Militare’’ (scale 1:25,000) the lake has a surface of about

194,000 m2 and an altitude of 2 m above sea-level, while

on the map of the ‘‘Regione Siciliana’’ (scale 1:10,000), a

surface of about 136,000 m2 and an elevation of 0.3 m

above sea-level is reported. Note that the latter map was

made on the basis of aerial photos taken in September 1992

at the end of the dry season. During our study large vari-

ations of the lake’s geometry also occurred. In May 1996,

the surface occupied by the lake and the water volume have

been estimated in about 200,000 m2 and 925,000 m3,

respectively. During the wet season, from October 1995 to

May 1996, rainfall was very close to long time average,

except in March (123 mm against a 30 mm average) and in

May (59 mm against a 10 mm average) when very high

values were recorded. During May intense rainfall

(43 mm) preceded of 1 week our survey. This unusually

rainy period caused the raising of the water level up to the

surrounding road, whose N and NE tracts have been totally

flooded. In October 1996, after a summer with average

climatic conditions, water level dropped about 30 cm,

causing a decrease of the estimated surface and total vol-

ume of about 15 and 7%, respectively. These fluctuations

have notable effects on lake salinity: Literature data on the

lake’s water chemistry (Table 1) show variations, like

those of chloride, higher than 50% (from 5.6 to 12 g l–1)

with maximum values in the dry seasons and minimum

values in the wet season.

In addition to the rain, the lake is fed by the hydro-

thermal manifestations located along its southern shores

(Fig. 1). In the south-eastern sector, right near the road, an

area of diffuse emission extends for tens of meter. This

area, characterized by sluggish but persistent emissions

(with flow rates <0.1 l s–1), temperatures between 34 and

58�C, and sometimes abundant bubbling gases, has a

maximum width of 2 m and is located some meters away

from the shore of the lake. Five of these emissions have

been collected in October. Hot waters issuing from these

manifestations kill the nearby vegetation and deposit car-

bonate crusts where waters stagnate. In some of these

springs, water has a bottle-green colour due to the presence

of algae. At several points, deposits of very porous trav-

ertine, signed by a great number of plant marks, are also

present. Other thermal springs are located on the south-

western sector of lake. Two of these have notable flows and

are always covered by the waters of the lake for about 1 m

of depth. The southernmost spring of this group (Polla 1) is

characterized by a high gas flow, whose pressure sustains a

10-cm water column.

Table 1 Literature data of the chemical composition of the lake (meq/l)

Date Na K Ca Mg Cl Br SO4 Alk Reference

1881 173 5.29 4.03 159 7.1 58 Foerstner (1881)

1963 264.2 9.87 0.5 5.6 197.5 14.4 37.8 Bencini et al. (1966)

1966 210 5.91 0.2 6.91 160 7.1 72.8 Bencini et al. (1966)

Jun-1980 272 8.6 0.3 9.3 227.5 13.8 62.1 Dongarra et al. (1983)

Sep-1980 281 10.2 0.3 10.3 232.9 0.43 14.9 63.2 Dongarra et al. (1983)

Dec-1980 278 14.6 0.2 9.7 229.2 20.3 63.4 Dongarra et al. (1983)

15-03-1987 182.7 3.26 4 12 183.2 45 Bocchi et al. (1988)

10-06-1987 310 10.3 0.1 10.8 255 0.36 17.9 61.5 D’Alessandro et al. (1994)

25-04-1989 357 10.6 0.3 12.9 301 19.9 67.2 D’Alessandro et al. (1994)

03-05-1990 313 13.7 0.5 12.1 260 22.9 58.7 D’Alessandro et al. (1994)

1990 384.8 20.5 0.5 7.98 340 20.6 79.8 Duchi et al. (1994)

10-01-1994 325.5 10.0 0.54 13.1 271.3 0.53 18.4 59.9 D’Alessandro et al. (1996)

17-10-1994 280 5.17 14.2 7.6 238.5 0.4 16.3 48.9 D’Alessandro et al. (1996)

15-09-1995 358 14.2 17.9 9.95 327 20.8 54.2 D’Alessandro et al. (1996)

Environ Geol (2007) 53:903–913 905

123

Sampling methods and analysis

Surveys have been carried out during May and October

1996. The first survey attempted to investigate the spatial

variability of the chemical and physico-chemical parame-

ters within the lake through a three-dimensional net. In

October, measurements were limited only along two sec-

tions of the lake (Fig. 1).

The selection of sampling points on the lake’s surface

has been made using a topographic map on the scale

1:3,000 on which several sections have been traced

(Fig. 1). Tightening some ropes across the lake material-

ized these sections.

Sampling has been carried out from a small boat with a

multiparametric probe (Aquamaster 345) and a water

sampler. The multiparametric probe, fitted with depth,

temperature, pH, electric conductivity and Eh sensors, was

used to collect data every meter along 41 depth profiles,

totalling 258 measurement points in May. In October only

68 sampling points were covered. Forty-six lake water

samples for chemical analysis have been collected, every

2 m of depth, along selected vertical profiles in May and 13

in October. Five samples of the thermal springs were col-

lected in both surveys. Furthermore, one thermal water

sample and one lake sample far from the thermal mani-

festations were collected each month for 1 year. Water

samples were filtered in the field trough 0.45 lm Millipore

filters and then kept into polyethylene bottles. The aliquot

for cations analysis was acidified with HNO3.

Chemical and isotopic analyses have been carried out in

the laboratory. Alkalinity has been determined by means of

titration. Sodium, K, Mg, Ca, Cl, NO3 and SO4 have been

determined by ionic chromatography, Li, Sr, Rb and Cs by

emission flame spectrophotometry, and Si and B trough

specific colorimetric methods using a spectrophotometer.

Water samples for oxygen isotope determination have

been prepared according to Epstein and Mayeda (1953) and

d18O(H2O) values, expressed as & with respect to the V-

SMOW standard, have been determined by mass spec-

trometry.

With the objective of following the temporal evolution

of the temperature of the lake, a series of miniaturized data

loggers with internal temperature sensors (Gemini Tiny-

talk) have been installed above the deepest zone of the

lake, at depths of 2, 6, 9.5, and 11 m. They have been

located along a vertical alignment, obtained using a rope

provided with a float and anchored to a weight at the

bottom of the lake; temperature data have been acquired

each 48 min (30 data each day) in the period May to Au-

gust 1997. The sensor installed at 11 m acquired data until

June 18.

Three cores of lake sediments have been collected

during the month of October at about 0.3, 1 and 12 m of

depth and to a relative distance from the southern coast of

10 (S2), 40 (S3) and 240 (S5) m, respectively (S2 and S3

were close to water sampling points X0 and X5; Fig. 1).

Cores have been collected with PVC tubes, 6 cm in

diameter, forced in the bottom of the lake. The sampling of

sediments confirmed that the bottom of the lake, in the

deeper zone (12.5 m), is not coherent and that the portion

of not consolidated sediments reaches a thickness of more

than 2 m. The choice of the sampling points has been

carried out considering the hot-springs (close to the shore

of the lake) and the centre of the lake (the deepest zone) as

end-members.

Cores S2 and S3 have been cut in fractions of 10–15 cm

length and oven dried at 60�C. Core S5, being unconsoli-

dated, could not be subdivided in sub-fractions, and only

two samples were collected from the top and from the

bottom of the core, respectively. Dried samples have been

pulverized in an agate mortar and their mineralogical

composition has been investigated by XRD, using a CuKaradiation Ni filtered, scanning rate 2� 2h min–1 in the range

2–60�. Semi-quantitative estimation of the abundance of

the identified mineralogical phases was performed using

the method of Schultz (1964).

Results

Hydrological balance

The hydrological balance has been carried out using data

collected in the period 1990–1994 at the Pantelleria Airport

by the Meteorological Service of the Italian Air Force. The

relevant data used in hydrological equations are reported in

Table 2.

Monthly distribution of air temperatures and precipita-

tion show the usual trend of the Mediterranean area

(Fig. 2), characterized by hot and dry summers and mild

and rainy winters. January is the rainiest month charac-

terized by 107 mm of rain and a monthly mean temperature

of about 12�C, while June to August are the driest periods

characterized by less than 10 mm of rain and mean tem-

peratures varying from 22 to 26�C.

From a hydrological point of view, Lake Specchio di

Venere could be divided into two sub-systems: the lake

itself and the catchment area feeding it. The comprehensive

hydrological balance Dtot is expressed by the formula:

Dtot ¼ Dl þ Dc; ð1Þ

where Dl and Dc are respectively the water deficits of the

lake and of the catchment area. Lake deficit is given by the

formula:

Dl ¼ ðR� E1Þ � Al; ð2Þ

906 Environ Geol (2007) 53:903–913

123

where R is the yearly rainfall, El the yearly evaporation,

calculated by the Meyer formula (Natale 1981) and Al the

lake surface.

Due to the local orographic conditions, both run-off and

infiltration feed the lake. As illustrated in Fig. 1, the lake

lies on a coastal plain, very close to the shoreline and

downslope a steep 400 m high hill (Mt. Gelfiser). Under

these conditions, the water table under the surface wa-

tershed is expected having the same, more smoothed,

behaviour of the northern hill’s flank, being very close to

topographic surface at its foot due to the abrupt intersection

of the coastal plain with the relief. The deep cut of the

topographic surface due to the presence of the lake causes

the outcropping of the piezometric surface, as testified by

the presence of several springs located on the SW lake’s

shore. A scheme of the proposed feeding systems of the

Specchio di Venere lake is reported in Fig. 3. The water

contribution from the catchment area is then given by the

classical hydrological balance equation:

Dc ¼ ðR� EcÞ � Ac; ð3Þ

where Ec is the actual yearly evapotranspiration and Ac the

planar surface of the catchment area. Evapotranspiration

was estimated with the Turc formula modified by Santoro

(1970) for Sicilian basins, which has been already suc-

cessfully applied in the study of small endoreic lakes in

Sicily (Madonia et al. 2006),

Results of hydrological calculations on yearly basis are

illustrated in Table 3.

As showed in the table, during a hydrological year the

lake receives a water surplus equal to 565,940 m3. This

surplus is essentially due to the watershed contribution,

being negative the mass balance between direct recharge

and evaporation from lake surface. This surplus, if accu-

mulated, would determine a water level rising and a con-

sequent periodic flooding. Since this fact has never been

observed, except few unusual events like that one previ-

ously discussed, the lake is supposed releasing to the

aquifer the water excess through seepage.

Lake water surveys

Figure 4a is a vertical profile of lake water pH on May

1996. It shows a modest pH variability (from 9.18 to 9.27),

more evident in the shallowest portions of the various

profiles. Distance from the thermal springs is probably the

main factor controlling pH variations. Near the thermal

springs (pH � 6.2), the lake waters exhibit lower pH val-

120

80

40 10

20

30

0

mm

J F M A S O N DJJ

°C

Temp.Rainfall

M A

Fig. 2 Monthly mean rainfall and temperature at Pantelleria island in

the period 1990–1994

WATER TABLE

FRESH WATERBRACKISH WATER

SEEPAGE

DIRECT METEORIC INPUTRUN OFF EVAPORATION

HYDROTHERMAL INPUT

SEA

S N

Fig. 3 Hydrogeological section of the lake’s area highlighting all

water inputs and outputs

Table 3 Yearly hydrological balance for the Specchio di Venere

Lake

Parameter Lake Catchment basin Total

Area (m2) 196,000 7,288,000

Rain (m3) 99,901 3,714,694 3,814,595

Evapo-transpiration (m3) 127,204 3,121,450 3,248,654

Water deficit (m3) –27,303 593,243 565,940

Table 2 Monthly values of parameters used in hydrological calculations

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

R (mm) 107 47.8 32 38.4 15.3 1.8 8.8 2.8 33.8 91.1 54.4 77.1

tA (�C) 12.3 12.0 13.9 14.8 18.8 22.4 24.7 26.7 25.0 21.6 16.8 10.6

Rh (%) 80.5 77.9 79 79 75 75.8 74.3 76.9 82.8 82.6 78.9 83.2

Ws (m/s) 7.81 7.36 7.7 7.31 6.36 5.56 4.58 4.86 5.11 5.92 6.69 7.72

Es (mmHg) 10.5 10.5 12 12.8 16.5 19.8 23.7 26.7 23.7 19.8 14.6 9.8

El (mm) 33 37 41 43 63 71 86 88 59 52 48 27

Environ Geol (2007) 53:903–913 907

123

ues (9.18 for the G profile), while further away from the hot

springs pH increases progressively (with a maximum val-

ues in the B profile). The large scatter of pH values, ob-

served on the lake’s surface waters, decreases

progressively on increasing depth, reaching a maximum

homogenization in all profiles between 4 and 8 m depth

(where a pH between 9.24 and 9.27 is attained). The slight

pH decrease in the deepest layers of the lake (below 9 m) is

probably due to the release of CO2 from decaying organic

matter at the sediment–water interface.

Figure 4b highlights a large variability of conductivities

on the lake’s surface waters on May 1996, from 33.0 to

35.7 mS cm–1 (with the highest values near the springs),

with a narrower range (31.2–31.6 mS cm–1) being reached

at 5 m depth. In the interval from 5 to 9 m depth, con-

ductivity increases slowly and then suddenly below 9 m,

reaching values exceeding 38 mS cm–1. This supports the

probable presence of a chemical stratification below 9 m.

The vertical heterogeneity of the lake in May is also

highlighted by the contrasting oxygen isotope composition

and Cl contents of deep (>8 m) and shallow waters. Both

profiles exhibit a trend of increasing 18O and Cl with depth

(Fig. 5).

The temperature profile (Fig. 4c) also shows a great

variability at the surface. Many factors contribute to

determine these variations: the distance the hot springs

(whose temperature ranges from 35 to 55�C), changes in air

temperature, sun radiation, and wind velocity. On the lake

surface samples, temperature ranges from 20.6�C (far from

the thermal springs) to 25.4�C (closer to the thermal

springs). Temperature decreases with depth, reaching

18.5�C on the bottom of the lake. Data do not support any

clear thermal stratification, although waters below 9 m

seem to be thermally more homogenized (18.5–19.1�C)

than on the surface.

Quite a different scenario was revealed by the October

survey. At that time, all parameters displayed a very low

variability in the 1–12 m depth interval (pH: 8.87–8.91;

conductivity: 28.4–29.1 mS cm–1; temperature: 20.9–

21.6�C) highlighting a homogenized condition of the lake’s

water.

9.309.259.209.15

Profile BProfile CProfile DProfile EProfile FProfile G

Profile BProfile CProfile DProfile EProfile FProfileG

pH

403632

cond (mS/cm)

262218-12

-8

-4

0

dept

h m

dept

h m

dept

h m

-12

-8

-4

0

-12

-8

-4

0

MayOctober

T°C

(a)

(c)

(b)

Fig. 4 pH, electric conductivity and temperature profiles obtained

with the multiparametric probe during the May and October (only

temperature) 1996 surveys. Profiles are identified by the same letter as

in Fig. 1

80007000 9000 10000

Cl mg/l

δ18O ‰

0

4

8

12

dept

h m

1 1.5 2 2.5

Fig. 5 Chlorine (grey squares) and d18O (white circles) vertical

profiles of the lake in May 1996. The two vertical profiles refer to

sampling points C3 (0–12 m) and D3 (0–10 m)

908 Environ Geol (2007) 53:903–913

123

Relationships between thermal springs and the lake

In the scatter diagrams of Fig. 6, the concentrations of

many dissolved species are contrasted against those of

chlorine. The diagrams are drawn in the attempt to evaluate

the extent to which lake waters form by simple evaporation

of thermal waters on the lake’s shore, in the reasonable

hypothesis that, in the lake’s salinity range, chlorine be-

haves as a conservative element. As such, Fig. 6 supports

both (1) the genetic link between lake and springs and (2)

the conservative behaviour of the elements Na, K, Li, Rb,

B and SO4 (which element/chlorine ratios are independent

on salinity and fit the respective ratios in the thermal

springs; black solid line). The lower Mg/Cl and Cs/Cl ra-

tios of the lake with respect to the springs evidences some

mineral sink for both elements, probably hydromagnesite

and dolomite (Mg) and amorphous silica or montmoril-

lonite (Cs). Calcium, Sr and SiO2, finally, are strongly

depleted in the lake (compared with the thermal springs),

evidencing an efficient removal of these elements from the

solution due to carbonate phases (aragonite, calcite) and

amorphous silica precipitation (Azzaro et al. 1983).

Variations with time

During the period from May 1996 to May 1997, one sur-

face lake’s water sample (at the opposite shore with respect

to thermal springs) and one thermal spring (polla 3) sample

have been monthly collected and analysed for their

chemical and isotope composition.

While the thermal spring had stable chemical and isoto-

pic composition during the study period, the lake displayed

important changes, reflecting variable rainfall influx and

evaporation. As an example, the temporal trend of Cl con-

tents in the lake is reported in Fig. 7, combined with rainfall

data from the Pantelleria airport. The lowest Cl contents

were recorded in May 1996, reflecting the abundant (much

higher than long term average) rains of the previous months.

During the summer, salinity increased peaking in Septem-

0

2000

4000

6000

8000N

a m

g/l

Li m

g/l

Mg

mg/

lC

a m

g/l

Sr m

g/l

Cs

mg/

lSi

o 2 mg/

l

(a)

MayLake

Springs MayOctober

October

0

125

250

375

500

K m

g/l

Rb

mg/

lB

mg/

l

0

250

500

750

1000

0 2500 5000 7500 10000

Cl mg/l Cl mg/l Cl mg/l

0 2500 5000 7500 10000

SO4

mg/

l

0

4

8

12

16

0

0.5

1.0

1.5

2.0

0

2

4

6

8

(b)

(c)

(d)

(f)

(e)

00 2500 5000 7500 10000

0 2500 5000 7500 10000

Cl mg/l

0

0

50

100

150

200

0

50

100

150

200

0

0.1

0.2

0.3

0.4

0

30

60

90

120

0.15

0.30

0.45

0.60

(g) (h)

(i) (j)

(k)

Fig. 6 Binary diagrams of chemical composition of waters collected

from the thermal springs and in the lake at different depths and

distance from the springs. All chemical parameters are plotted versus

chloride concentration considered a conservative element. Sodium, K,

SO4, Li, Rb and B (a–f) display in the lake waters the same ratio as in

the thermal springs being affected only by the evaporative process,

while Mg, Cs (g–h) and more strongly Ca, SiO2 and Sr (i–k), being

affected also by precipitation of solid phases, show lower ratios than

in thermal waters. Black line is the element/chlorine ratio in the

thermal waters and the grey line is the same ratio in sea water

4000

5000

8000

100

50

0

25

75

mm

Cl m

g/l

9000

10000

01-05-96 31-08-96 31-12-96 02-05-971.3

1.8

2.3

2.8

3.3

δ18O

‰

Fig. 7 Variations of chlorine content (grey circles) and oxygen

isotopic composition (white triangles) of the lake’s water in the

period from May 1996 to May 1997. Chlorine content of thermal

spring water is also plotted (black squares) together with rainfall

Environ Geol (2007) 53:903–913 909

123

ber. In October, after abundant rains, the lake’s waters be-

came progressively more diluted reaching a relative salinity

minimum in February 1997. The d18O isotope composition

of the lake (Fig. 7) was in good agreement with the above

evaporation-meteoric dilution process.

Temperature values measured from the 13th of May

1997 to the 7th of August 1997 at different depths (Fig. 8)

followed closely the mean daily air temperature variation.

During this period (from late spring to early summer),

characterized by increasing air temperatures, temperature

in shallowest lake’s level were the highest and in the

deeper level the lowest. A daily temperature cycle char-

acterised the shallow lake’s water level, with up to 2�C

temperature excursion. In the deepest lake’s layers, daily

variations were below instrumental sensitivity (about

0.3�C). Temperature differences between the shallowest

and the deepest measuring point varied between 0 and 4�C,

pointing to limited temporarily thermal stratifications of the

lake. Stratifications lasted for maximum 17 days and were

interrupted by mixing episodes, which occurred always in

days with maximum wind intensities exceeding about

15 m s–1 (Fig. 8). No effective rainfall events occurred in

this period.

Sediment mineralogy

A previous mineralogical characterisation (Azzaro et al.

1983) of two sediment cores collected near the western

springs detected the presence of clay minerals probably

belonging to the smectite group. Others crystalline phases

identified were: dolomite, calcite, carbonato-apatite, hyd-

romagnesite, pyrite. Minor amounts of quartz, feldspar and

pyroxene (egirina—NaFeSi2O6), probably deriving from

mechanical weathering of the surrounding rocks, were also

been identified by the authors.

In this work, 13 samples obtained from three cores of

lake sediments have been analysed by XRD on the tout-

venant. Results revealed the presence of carbonate phases

in all cores. Aragonite was detected in all samples, while

calcite, dolomite and hydromagnesite only occasionally.

Clay minerals were always present with minor quantities of

quartz. Other identified phases include pyroxene (egirina),

feldspars, carbonato-apatite and pyrite.

A semi-quantitative analysis (Table 4) showed that

aragonite and clay minerals were the most abundant pha-

ses, representing more than 90% of the total. Carbonato-

apatite and pyrite were accessory minerals and increased in

abundance in the deepest parts of the cores. Feldspar and

Pyroxenes, of probable detrital origin, were identified in

small quantities in the shallower parts of core S2 (the

closest to the shores of the lake) and in the whole column

of core S5 (in the deeper part of the lake). Finally quartz,

which was identified in small quantities in all samples, and

did not show any significant variations with depth.

Discussion

Compositional and thermal stratification

Contrasting compositional profiles in the lake were ob-

served in the two surveys: in May, a striking salinity and

chemical stratification was observed, while major ion

chemistry, pH, conductivity, and temperature of the lake

were almost invariable in October, suggesting that

homogenization of the lake waters had occurred. During

the May survey, clear weather and weak wind prevailed

with an average air temperature of 25�C (during the day),

while strong south-east winds and showers characterised

the October survey (air temperature was about 22�C). The

wind intensity was so notable in October to produce a

strong wave motion on the lake surface: this probably gave

rise to the turnover of the lake waters, and totally obscured

the chemical stratification previously observed in May.

Temperature data acquired in the period May–August 1997

supports further the idea that Specchio di Venere lake can

occasionally undergo stratification events: they also high-

light however, that these events are short lasting and not

seasonal as observed in several other lakes. The two sur-

veys of May and October 1996 are probably too far away to

conclude that the lake has seasonal stratification episodes.

30

26

22

18

30

35

25

20

10

0 0

15

5 1

2

3

4

5

6

7

2 m

6 m9.5 m

11 m

Air temp.(°C)

°C°C

Rain (mm)

mm

Wind (m/s)m/s

10-05-97 10-06-97 11-07-97 11-08-97

Fig. 8 Water temperature acquired each 48 min (30 data each day) in

the period 14/05/97–7/8/97 at 2, 6, 9.5, and 11 m depths (uppergraph), together with mean daily air temperature, rainfall and wind

intensity data measured at the Pantelleria airport (lower graph)

910 Environ Geol (2007) 53:903–913

123

At least two factors causing density variations and thus

vertical instability of water column have to be considered:

(1) temperature variations of the waters due to the input of

the colder winter rains and/or to the decreasing air tem-

perature; (2) the salinity variations resulting from changing

input/output from meteoric waters, hydrothermal waters,

run-off waters, all having contrasting chemical composi-

tions and salinities.

The higher Cl contents of lake’s bottom waters in May

(Fig. 5), combined with the parallel d18O increase, would

paradoxically indicate a greater evaporation taking place at

depth. In the reality, it is the final result of ‘‘sinking’’ of the

dense shallow water masses, after intense evaporation at

the lake’s surface: during the summer months, the lake is

subjected to an intense evaporation, and the lake’s surface

waters becomes enriched in both heavy isotopes and con-

servative dissolved ions (Cl). When the density of the

evaporated surface waters, increased further by the pres-

ence of microcrystals of precipitating solid phases, exceeds

the isotopically lighter and more dilute deeper ones, the

former will sink to the bottom. Accumulation of these

evaporated waters will slowly build up a compositional

stratification of the waters such as that found in the May

1996 survey. Such a stratification will last until some

external cause, such as changing weather conditions in

autumn/winter, will trigger remixing of the waters with a

subsequent homogenization.

Chemical evolution

Compared with other volcanic lakes, ‘‘Specchio di Ve-

nere’’ has a peculiar composition. Based on its high Cl and

SO4 contents, the lake would fall into the field of the

‘‘active crater lakes’’ as defined by Varekamp et al. (2000).

The latter, however, display very low pH values (<3), while

‘‘Specchio di Venere’’ is typically basic, falling at the

upper pH range of volcanic lakes quoted by Marini et al.

(2003).

The composition of volcanic lakes is principally con-

trolled by the flux and composition of incoming volcanic or

geothermal fluids. ‘‘Specchio di Venere’’ lake makes no

exception: its chemical features are inherited from the

thermal springs feeding it (Fig. 6). In turn, the main pro-

cesses governing the composition of the latter are: (a)

mixing between infiltrated meteoric and sea waters; (b)

heating to 230–280�C and water–rock interaction, resulting

in the formation of a brine with typical enrichment in

alkaline elements and depletion in Mg; (c) dissolution of

magma-derived volatiles (mainly CO2) (Dongarra et al.

1983; D’Alessandro et al. 1994; Squarci et al. 1994; Parello

et al. 2000). On spring discharge, evaporation contributes

to increase further lake’s salinity, while precipitation of

carbonates and clay minerals act as sinks for Mg, Sr, Cs

and silica.

Precipitation of solid phases

Speciation calculations, performed with EQ3 code (Wolery

1992), suggest that the lake waters are super-satured with

respect to aragonite and calcite, as long as with respect to

dolomite, hydromagnesite, huntite and magnesite. These

minerals, except the last two, were detected in the sedi-

ments with the abundances shown in paragraph 5.4.

The higher amount of aragonite in the sediments, with

respect to other carbonate phases (calcite, hydromagnesite

and dolomite), is consistent with the high Mg and Sr

contents of the springs waters (Pentecost 2005). This latter

phase is also kinetically favoured, due to the strong super-

Table 4 Semiquantitative mineralogic analysis of the sediment samples collected on the bottom of lake Specchio di Venere

Sample Arag. Calc. Dol. Hydrm. Ca-Ap. Clay Min. Qz Fels. Pyr. Pirox.

S2 0–15 **** * * ** **** * * *

S2 15–25 **** ** * * **** ** * *

S2 25–35 **** * * **** * *

S2 35–45 **** * *** *** * *

S2 45–55 **** * * * **** *

S2 45–55 **** * * ** **** *

S3 0–15 **** * ** *** * *** *

S3 15–25 **** * ** **** *

S3 25–35 **** * * * *** * * *

S3 35–45 **** * * * *** * *

S3 45–55 **** * * * *** *

S5 Top **** * * ** *** * * *

S5 Bottom *** * ** *** * ** **

****Very abundant, ***Abundant, **Less abundant, *Scarce

Environ Geol (2007) 53:903–913 911

123

saturation conditions created by mixing of spring and lake

waters, the formation and growth of aragonite nuclei are

favoured with respect to those of calcite.

The formation of primary inorganic Dolomite has

sometimes been supposed in saline lakes. Indeed, some of

the conditions favouring this process such as high salinity

and high alkalinity (De Deckker and Last 1988) are also

present in the lake Specchio di Venere. The Mg/Ca (2–16),

however, is probably too low to allow direct precipitation

of dolomite. De Deckker and Last (1988), for example,

measured ratios always higher than 25 in some saline lakes

with primary dolomite precipitation. In the lake Specchio

di Venere, dolomite is probably formed through early

diagenetic processes starting from hydromagnesite.

Apatites may have several origins, i.e. magmatic,

hydrothermal, sedimentary, and chemical precipitation.

The latter origin is the most probable for the lake sedi-

ments. These apatites precipitate in similar way as calcium

carbonate. Super-saturation is reached with sudden pH in-

crease (like for the thermal springs flowing into the lake),

as calcium phosphates are soluble in acid solutions and

insoluble in alkaline solutions (Gottardi 1984). The phos-

phorous content in the lake’s waters, can derive both from

weathering of the igneous rocks and from the organic

matter present on the bottom of the lake, fertilized by

permanent and migratory birds’ guano. During the organic

matter decay, the released CO2 is partially adsorbed by the

forming carbonato-apatites (Wedepohl 1969).

Pyrite forms under reducing conditions at the sediment–

water interface or inside the sediment itself, when a certain

amount of Fe is present. Reducing conditions, favouring

sulphide formation, were measured in the lake water close

to the bottom and are probably present also inside the

sediment due to the abundant presence of organic matter.

Gas hazard

Previous studies (Favara et al. 2000) assessed a relatively

high magmatic CO2 output (12.4 kg s–1) at Pantelleria. The

Specchio di Venere lake area is one of the most active

degassing areas, with measured soil gas fluxes of up to

1,300 gCO2 m–2 per day (Favara et al. 2000). Apart from

high soil degassing, gas bubbling in the thermal springs and

a few mofettes on the western part of the lake’s area

contribute to this elevated CO2 release. This high gas flux

raises concern on the potential CO2 accumulation in the

lake up to hazardous levels.

After the lake Nyos disaster many studies were made to

investigate possible gas accumulation in volcanic lakes all

around the world. These studies highlighted that the

development of significant gas accumulations, apart from a

high endogenous gas flux, requires high depths and a

meromictic or oligomictic regime. The latter condition is

most probably encountered in the Cameroonian lakes

Monoun and Nyos and in Lake Kivu in east Africa, where

CO2 accumulations are known to exist. In the temperate

zone, seasonal overturns prevent long-term stratification

and gas built-up also in lakes with endogenous gas fluxes

similar to those characterising the African lakes (e.g., Lake

Mashu in Japan and Eifel maar in Germany; Eby and Evans

2006).

In the case of Lake Specchio di Venere, despite the high

presence of high gas fluxes, the shallow depth (12.5 m), the

peculiar chemical characteristics and frequent overturning

of the lake contribute to prevent dangerous gas accumu-

lations. As a matter of fact, the effective pCO2 of the lake,

calculated with EQ3 code, is higher than the atmospheric

one (1.1 · 10–3 against 3.3 · 10–4), as an evidence of CO2

contribution from the springs on the lake’s shore.

Conclusions

Hydrologic modelling explains the formation of lake

Specchio di Venere through the emerging of the water ta-

ble. Hydrological balance indicates that during the 1990–

1994 mean hydrological year, the yearly water deficit of

the lake was positive and in the order of 570,000 m3. This

surplus has to be considered a lower limit, since the lake

also receives a contribution from the thermal aquifer, as

documented by the presence of numerous hot springs on

the southern shores. The lake, being endorheic, looses this

surplus through seepage to groundwater.

The chemical and isotopic analyses carried out on the

waters of lake ‘‘Specchio di Venere’’ during the period

May 1996–June 1997 showed that both chemical and iso-

topic composition change in time remarkably. The greatest

documented variations in the studied period were recorded

between May 1996 (Cl � 7,800 mg/l), and September

1996 (Cl � 9,050 mg/l), with a relative variation of about

14%. Even greater salinity variations are documented in

literature, where 50% salinity differences are reported.

The chemical composition of the lake is partially

inherited from the feeding waters (mainly thermal springs),

modified by evaporation processes and precipitation of

solid phases. Spring waters (whose chemical composition

is constant in time), degas CO2 on discharge at the surface

on the lake’s shore, which, together with evaporation, leads

to the precipitation of carbonate phases (aragonite, calcite,

dolomite, and hydromagnesite), clay minerals and amor-

phous silica. The resulting salinity increases, and the den-

sity gradients established between the shallow and bottom

waters, induce sinking of the former ones, favouring the

establishment of stratifications. The latter can be more or

less lasting, depending on the delicate equilibrium between

the meteorological factors and the chemical–physical fac-

912 Environ Geol (2007) 53:903–913

123

tors affecting water density at different depths. The fre-

quent overturning of the water, together with morpholog-

ical (small volume and shallow depth) and chemical (high

pH) factors, precludes the build-up of dangerous gas levels

within the lake.

References

Azzaro E, Badalamenti F, Dongarra G, Hauser S (1983) Geochemical

and mineralogical studies of lake Specchio di Venere, Pantelleria

island, Italy. Chem Geol 40:149–165

Bencini A, Duchi V, Tonani F (1966) Rapporto idrogeochimico sulle

acque termali di Pantelleria. Internal Report of the CNR-IIRG,

Pisa

Bocchi G, Gabbianelli G, Lanzafame G, Lucchini F, Rabbi E, Rossi

PL (1988) Relazione sui rilievi eseguiti sul ‘‘Lago di Venere’’,

Pantelleria. Bull Italian Natl Group Volcanol 4:63–73

Chiodini G, Cioni R, Guidi M, Magro G, Marini L, Raco B (2000)

Gas chemistry of the Lake Piccolo of Monticchio Mt. Vulture, in

December 1996. Acta Vulcanol 12:139–142

Cioni R, Guidi M, Raco B, Marini L, Gambardella B (2003) Water

chemistry of Lake Albano (Italy). J Volcanol Geotherm Res

120:179–195

Civetta L, Cornette Y, Crisci G, Gillot PY, Orsi G, Requeiros CS

(1984) Geology, geochronology and chemical evolution of the

island of Pantelleria. Geol Magaz 121:541–562

Civetta L, Cornette Y, Gillot PY, Orsi G (1988) The eruptive history

of Pantelleria (Sicily Channel) in the last 50 ka. Bull Volcanol

50:47–57

D’Alessandro W, Dongarra G, Gurrieri S, Parello F, Valenza M

(1994) Geochemical characterization of naturally occurring

fluids on the Island of Pantelleria (Italy). Mineral Petrogr Acta

37:91–102

D’Alessandro W, Inguaggiato S, Parello F, Pecoraino G, Romeo L,

Sortino F (1996) Pantelleria island: geochemical survey of hot

waters and gases in the period 1993–1995. Acta Vulcanol

8(2):238–240

De Deckker P, Last WM (1988) Modern dolomite deposition in

continental, saline lakes, western Victoria, Australia. Geology

16:29–32

Dongarra G, Hauser S, Alaimo R, Carapezza M, Tonani F (1983) Hot

waters on Pantelleria island. Geochemical features and pre-

liminary geothermal investigation Geothermics 12:49–63

Duchi V, Campana ME, Minissale A, Thompson M (1994)

Geochemistry of thermal fluids on the volcanic island of

Pantelleria, southern Italy. Appl Geochem 9:147–160

Eby GN, Evans WC (2006) Taming the killer lakes of Cameroon.

Geol Today 22:18–22

Epstein S, Mayeda T (1953) Variation of O18 content of waters from

natural sources. Geochim Cosmochim Acta 4:213–224

Favara R, Giammanco S, Inguaggiato S, Pecoraino G (2000)

Preliminary estimate of CO2 output from Pantelleria island

volcano (Sicily, Italy): evidence of active mantle degassing.

Appl Geochem 16:883–894

Foerstner H (1881) Nota preliminare sulla geologia dell’isola di

Pantelleria secondo gli studi fatti negli anni 1974 e 1881. Boll R

Comit Geol It 12:532–556

Gottardi G (1984) I minerali, 3rd edn. Boringhieri, Torino

Le Guern F, Sigvaldason GE (eds) (1990) The Lake Nyos event and

natural CO2 degassing, II. J Volcanol Geotherm Res 42:307–402

Madonia P, Naselli-Flores L, Parello F, Parlato B, Viola A (2006)

Geological development of a gypsum lake formed at the

beginning of the 20th century in central Sicily, Italy: integration

of historical data with modern survey techniques. Chem Ecol

22:S333–S347. DOI:10.1080/02757540600557728

Mahood GA, Hildreth W (1986) Geology of the peralkaline volcano

at Pantelleria, Strait of Sicily. Bull Volcanol 48:143–172

Marini L, Vetuschi Zuccolini M, Saldi G (2003) The bimodal pH

distribution of volcanic lake waters. J Volcanol Geotherm Res

121:83–98

Natale L (1981) Idrologia tecnica. Manuale di ingegneria civile, vol1.

Zanichelli, Bologna, p 339

Parello F, Allard P, D’Alessandro W, Federico C, Jean-Baptiste P,

Catani O (2000) Isotope geochemistry of Pantelleria volcanic

fluids, Sicily Channel rift: a mantle volatile end-member for

volcanism in southern Europe. Earth Planet Sci Lett 180:325–

339

Pentecost A (2005) Mineralogy and elemental composition. In:

Travertine. Springer, Berlin, pp 101–128

Rittmann A (1967) Studio geovulcanologico e magmatologico

dell’isola di Pantelleria. Riv Min Sicil 106–108:147–182

Santoro M (1970) Sulla applicabilita della formula di Turc per il

calcolo della evapotraspitrazione effettiva in Sicilia. In: Pro-

ceedings of the 1st international symposium on groundwaters,

Palermo

Schultz LG (1964) Quantitative interpretation of mineralogical

composition from X-ray and chemical data for the Pierre Shale.

USGS Prof Paper 391-C:C1–C31

Sigurdsson H, Devine JD, Tchoua FM, Presser TS, Pringle MKW,

Evans WC (1987) Origin of the lethal gas burst from Lake

Monoun, Cameroun. J Volcanol Geotherm Res 31:1–16

Squarci P, Giannelli G, Grassi S, Mussi M, D’Amore F (1994)

Preliminary results of geothermal prospecting on the Island of

Pantelleria. Acta Vulcanol 5:117–123

Varekamp JC, Pasternack GB, Rowe GL (2000) Volcanic lake

systematics II. Chemical constraints. In: Varekamp GJ, Rowe

GL (eds) Crater Lakes. J Volcanol Geotherm Res 97:161–179

Washington HS (1913) The volcanoes and rocks of Pantelleria I. J

Geol 21:653–670

Wedepohl KH (1969) Handbook of geochemistry. Springer, Heidel-

berg

Wolery T (1992) EQ3NR, a computer program for geochemical

aqueous speciation-solubility calculations: theoretical manual,

user’s guide and related documentation. Lawrence Livermore

National Laboratory, Livermore

Environ Geol (2007) 53:903–913 913

123