Content uploaded by Chris Harris
Author content
All content in this area was uploaded by Chris Harris on Jan 18, 2018
Content may be subject to copyright.
Pergamon Journal of African Earth Sciences, Vol. 31, No. 3/4, pp. 467-481. 2000
0 2001 Elsavier Science Ltd
Pll:SO899-5382(00)00002-1 All rights reserved. Printed in Great Brdain
0699-5362/01 5. see front matter
Oxygen and hydrogen isotope geochemistry of thermal
springs of the Western Cape, South Africa:
recharge at high altitude?
R.E. DIAMOND and C. HARRIS*
Department of Geological Sciences, University of Cape Town,
Rondebosch 7700, South Africa
ABSTRACT-A number of thermal springs with temperatures up to 64°C are found in the Western
Cape Province of South Africa. The average 613C value of gas (CO,+CH,) released at three
springs is -22%0, which is consistent with an entirely biogenic origin for the C and supports
previous investigations which showed that the springs are not associated with recent or nascent
volcanic activity. Most springs issue from rocks of the Table Mountain Group, where faulted and
highly jointed quartzites and sandstones of the Cape Fold Belt act as the main deep aquifer. The
6D and 6’*0 values of the springs range from -46 to -18% and from -7.3 to -3.9%0, respectively.
Although the thermal springs have isotope compositions that plot close to the local meteoric
water line, their 6D and VO values are significantly lower than ambient meteoric water or
groundwater. It is, therefore, suggested that the recharge of most of the thermal springs is at a
significantly higher altitude than the spring itself. The isotope ratios decrease with increasing
distance from the west coast of South Africa, which is in part related to the continental effect.
However, a negative correlation between the spring water temperature and the 6180 value in the
thermal springs closest to the west coast indicates a progressive increase in the average altitude
of recharge away from the coast. o 2001 Elsevier Science Limited. All rights reserved.
RESUME-La province Ouest du Cap (Afrique du sud) contient plusieurs sources thermales, dont
certaines atteignent des temperatures de 64°C. A I’exutoire de trois de ces sources, la valeur
moyenne du 613C des gaz (CH,+ CO,) est de -22%. Ces mesures isotopiques correspondent a la
decomposition de mat&es organique, ce qui est en accord avec les precedentes etudes. Ces
dernieres indiquerent que ces sources ne sont pas associees a des circulations d’eaux juveniles
ou bien liees a une activite volcanique. La plupat-t des sources emergent dans des roches appartenant
au Groupe de ‘Table Mountain’ et correspondent a des zones de plissement fortement silicifiees
(quartzite). La zone de gres de la ceinture plissee du Cape sousjaccent joue alors le role d’aquifere
profond. Les valeurs 6D et de VO de ces sources sont respectivement comprises entre -46 a
-18960 et -7,3 et -3,9%. Bien que ces sources thermales aient des compositions isotopiques
proches des valeurs de la droite des eaux meteoriques locales. Elles sont significativement plus
basses que les eaux meteoriques et les eaux souterraines. Ces resultats suggerent que la plupart
des zones de recharge des sources se situent a une altitude superieure a la source. De plus, ces
valeurs isotopiques decroissent avec I’eloignement de la c&e ouest de I’Afrique du sud, ce qui
indique que ces variations sont partiellement Ii&es a un effet de continentalite. Cependant, la
correlation negative entre la temperature des sources et la valeur de VO de la source thermale la
plus proche de la c&e indique une augmentation progressive de I’altitude moyenne de la zone de
recharge loin de la zone c&i&e. o 2001 Elsevier Science Limited. All rights reserved.
(Received 13/l 2/99: revised version received 29/5/00: accepted 24/l O/00)
*Corresponding author
charris@geology.uct.ac.za
Journalof African Earth Scimcas 467
R. E. DIAMOND and C. HARRIS
INTRODUCTION
Most hot springs world-wide, are associated with the
waning stages of volcanic activity (e.g. Kent, 1949).
Hot springs, which are not associated with volcanic
activity, are often associated with recent uplift, for
example in the Pakistan Himalayas (e.g. Chamberlain
et al., 19951, where meteoric water is heated by
cooling magmatic rocks. There are over 87 thermal
springs in South Africa ranging in temperature from
25-64°C. None of the springs are associated with
recent volcanic activity, which is unknown in this part
of Africa. The geology and chemical composition of
the springs has been described by Kent (1949) and
Hoffmann (19791, respectively.
The aims of this paper are as follows:
il to establish the degree of variation in 0 and H
isotope data for the 12 thermal springs (Table I) from
the Western Cape Province. Mazor and Verhagen
(I 983) reported stable isotope data from seven of
the Western Cape springs (but both 0 and H isotope
data from only four);
ii) to determine the monthly isotope variability of
the spring waters by analysing samples collected
from four of the springs every month for a period
of eight months. Long term variability can be as-
sessed by comparing data for samples from this
study collected in 1995-7 with those of Mazor and
Verhagen (I 983) whose samples were collected in
1971-2;
ii. to compare the isotope composition of the
springs to meteoric water and cold groundwater in
the area; and
iv) to use the stable isotope data to constrain the
nature of the recharge and the mechanism(s) of the
heating of the thermal springs.
REGIONAL BACKGROUND
Geolosy
The geology of the Western Cape is dominated by
the Palaeozoic Cape Supergroup, of which the resistant
sandstones and quartzites of the Table Mountain
Group are the most prominent. The basement consists
of Late Precambrian low-grade metamorphic rocks
of the Malmesbury and Kango Groups and the - 540
Ma plutons (Armstrong et a/., 1998) of the Cape
Granite Suite. The Table Mountain Group forms the
lower part of the Cape Supergroup, above which lie
the shale and sandstone formations of the Bokkeveld
and Witteberg Groups. The Cape Supergroup is
overlain by the varied sedimentary succession of the
Karoo Supergroup. The great thickness and well-
cemented character of the Table Mountain Group
sandstones and quartzites results in them being the
major component of the high relief areas (up to 2000
468 Journal of African Earth Sciences
m) of the Cape Fold Belt (Fig. I) (Broquet, 1992;
Halbich, 1992).
Multiphase deformation of the basement Kango and
Malmesbury Groups occurred during Pan-African
orogenesis between 600 and 500 Ma (Gresse et al.,
1992). Many of the structures were reactivated during
the Cape Orogeny (250 Ma: HIlbich et al., 1992) and
during the break-up of Gondwana during the Mesozoic
(Gresse et a/., 1992). Metamorphic conditions during
the Cape Orogeny reached greenschist-facies grade,
and much of the sandstone in the Table Mountain
Group recrystallised to quartzite. Movement of water
through rocks of the Cape Supergroup is, therefore,
primarily via these fractures because cementation
destroyed the primary porosity.
Climate
The Western Cape is the small portion of South Africa
(Fig. 1) which experiences a Mediterranean-type
climate. To the north, this climate regime grades into
semi-desert. To the east, the climate becomes less
seasonal and tends towards subtropical on the coast.
The essence of a Mediterranean climate is cold wet
winters and warm dry summers. The generally
mountainous nature of the Cape Fold Belt results in
the entire region having sharp changes in climate.
Rainfall is highly variable and ranges from low
summer (December-March) monthly means of 1 O-
20 mm in the wide inter-montane valleys and on
the coastal plains and + 50 mm in the mountains,
to winter (June-August) monthly means of 40-I 00
mm and over 200 mm, respectively. Temperatures
vary from winter mean minimum daily temperatures
of < 5°C in the inland valleys and * 10°C on the
coastal plains to summer mean maximum daily
temperatures of > 30°C inland and +25’C on the
coastal plains (SAWB, 1996).
THERMAL SPRINGS
All groundwater that sinks to any appreciable depth
will become heated because of the geothermal
gradient. Mazor (1991) suggested a purely arbitrary
temperature divide between cold springs and thermal
springs of 6’C above average annual surface
temperature. The Western Cape valleys and coastal
plains experience annual average temperatures
between 15% and 20°C, so any water discharging
at or above about 26“C can be classified as a thermal
spring.
In the Western Cape, there is a full gradation from
the cold ( < 2O’C) to the hottest spring in the country,
Brandvlei, at 64°C. All of the well-known thermal
springs in the area were sampled during this work
(Table 1). The majority are above 40°C, with two
Oxygen and hydrogen isotope geochemistry of thermalsprings of the Western Cape, South Africa
’
Victoria West
LTulbagh Baden-Baden
Area of study
Warmwaterberg
100 60 20
LI- 80 40 0
100
I I
200km
Figure 1. Sketch map of the Western Cape showing the location of thermal springs sampled. The location of rainfall
monitoring stations at the University of Cape Town (UCTI, Cape Town International Airport IIAEAJ, Citrusdal, Tulbagh and
Oodtshoom are also shown. The thermal spring at Citrusdal is known as ‘The Baths’, but to avoid confusion it is referred to
as Citrusdal in the text. The area of outcrop of the Caoe SuDergroup forming the Cape Fold Belt Mountains is indicated (taken
from Theron et al., 1991a).
springs, Witzenberg (28“C) and Rietfontein (27OC),
just falling within the classification of thermal. Most
of the springs are found at relatively low altitude
(~300 m), with three springs found at 700 m or above
(Toowerwater, Rietfontein and Witzenberg). The
constancy of discharge temperature and volume is a
generally known fact (Kent, 1949) and was confirmed
by discussion with the resort managers at The Baths
(Citrusdal), Calitzdorp Spa (for which measurements
go back to the 19th century), Caledon and Goudini.
The 12 springs sampled have yields that vary from
< 5 I s-l to 126 I s-l. The spring with the highest
yield (Brandvlei) is also the hottest, whereas most of
the springs with low discharge are relatively cool.
This may in part be due to more effective cooling by
heat loss to the surrounding rock in the case of the
springs with low yield.
All but two of the thermal springs described in this
paper occur within, or close to, rocks of the Table
Mountain Group, which, as described above, has very
limited residual primary porosity. Deep groundwater
movement in the Table Mountain Group is via
fractures, which are either horizontal bedding planes
or vertical joints. The joints occur in three roughly
parallel sets throughout the Cape Fold Belt: northwest-
southeast, northeast-southwest and east-west.
These and the bedding planes provide a network of
interconnecting fractures through which water can
flow. The Table Mountain Group contains two main
aquifers separated by the thin, but impermeable, shales
and siltstones of the Cedarberg Formation; the lower
is the Peninsula Formation and the upper is the
Nardouw Subgroup. Faults punctuate the stratigraphy
and are present at nearly all the springs. It seems,
therefore, that faults are critical in providing channel-
ways through the otherwise impermeable Cedarberg
Formation for heated water to percolate upwards.
Geological cross-sections for Brandvlei, Calitzdorp and
Citrusdal are shown in Fig. 2.
The geothermal gradient of the Cape Fold Belt area
is not well established. An estimate can be made
from two boreholes drilled into the Karoo Supergroup,
north of the Cape Fold Belt, about 50 km from the
spring Rietfontein (Fig. 1). The first borehole pene-
trated 850 m of rock with an average geothermal
gradient of about 18°C km-‘, The second borehole
reached to 1760 m below s&ace; and a geothermal
gradient of about 21’C km-’ was observed in the
R. E. DIAMOND and C. HARRIS
Table 1. General information about sampled thermal springs
Spring
Baden-Baden
Temp Flow Altitude Distance Geological environment Fe, Mn and Si mineralisation
(OC) (I s-11 (ml (km)
38 280 150 TMG-Bokkeveld Group contact + near regional fault in
TMG
Brandvlei
Caledon
Calitzdorp
Citrusdal
(‘The Baths’)
Goudini
Malmesbury
Montagu
64
53
52
43
39
34
45
126
10
30
3
i4
220
360
200
250
290
120
280
90 TMG-Bokkeveld Group contact + regional fault in TMG
100 TMG-Bokkeveld Group contact + regional fault in TMG;
Fe, Mn and Si mineralisation
310 TMG-Bokkeveld Group contact + TMG-Uitenhage Group
unconformity
80 Fault in Nardouw Subgroup of TMG
80 Regional fault in TMG (Peninsula Formation/Nardouw
Subgroup faulted together)
40 Fault in Malmesbury Batholith of Cape Granite Suite
155 TMG-Bokkeveld Group contact + near regional fault in
TMG
Rietfontein -27 *2 700 260 Dwyka Group-Prince Albert Formation contact
Toowerwater 49 800 455 Regional fault in TMG (Peninsula Formation/Enon
Formation faulted together); Fe, Mn and Si mineralisation
Warmwaterberg 44 *5 500 225 Near top of Nardouw Subgroup + regional fault in TMG;
Fe, Mn and Si mineralisation
Witzenberg -28 *1 800 105 Peninsula Formation
Distance: distance from the West Coast measured in a straight line with an east-west orientaion; TMG: Table Mountain Group;
f : flow rate was estimated; -: temperatures were measured on one occasion only.
upper 1450 m, which then increased to about 27%
km-’ in the deep section of the hole (Theron et al.,
1991a; Jones, 1992). In the Cango Caves, near
Oudtshoorn, the air temperature is constant at 17%
(Doel, 1995). If this temperature were typical of
shallow groundwater in the area, then Brandvlei
thermal water has been heated to at least 47OC above
that of shallow groundwater. If an average geothermal
gradient of 20% km-’ is assumed, then the thermal
water at Brandvlei must come from an average depth
of 2.35 km. Although this is a minimum estimate,
because the water must have cooled on its way to
the surface, the rate of flow (I 26 I s-l) is large and
the degree of cooling must be slight. The geological
cross-section is consistent with this interpretation
(Fig. 2).
ANALYTICAL
Sampling methods
Water samples were stored in 1 DO ml plastic (‘medical
flats’) bottles and analysed as soon as possible after
collection. Some springs were sampled only once, but
four [Brandvlei, Calitzdorp, Citrusdal (The Baths) and
Malmesburyl were sampled every month. At
Citrusdal, samples were taken at the eye of the spring;
and at Baden-Baden, Caledon, Rietfontein, Goudini,
Montagu, Toowerwater and Warmwaterberg,
samples were taken from pipes which directly tapped
the spring. The springs at Brandvlei, Calitzdorp and
Malmesbury issue directly into pools, and samples
were collected from as close to the source as possible
in order to minimise the influence of evaporation. Gas
bubbling up through the source pools at Brandvlei,
Calitzdorp and Malmesbury was collected in
November 1995. Gas bubbles were caught in a plastic
funnel before being allowed to expand into evacuated
glass vessels.
Isotope analysis
For 0, the CO, equilibration method of Socki er al.
(1992) employing disposable pre-evacuated 7 ml glass
vials was used. For H, 2 mg of water contained in a
microcapilliary tube was dropped into a Pyrex@ tube
containing - 10 grains of Indiana Zn. The tube was
attached to the vacuum line, frozen in liquid N,
evacuated and then sealed using a torch. Once a large
enough batch of samples had been prepared, they
were placed in a furnace at 45O’C to reduce the water
to H,. Isotope ratios of CO, and H, were measured
470 Journal of African Earth Sciences
Oxygen and hydrogen isotope geochemistry of thermal springs of the Western Cape, South Africa
W The Baths E
r 3 km / Kouebokkeveldberge
~
....
/ Warmbadberg .~-. ,/ ?So .......... ~"'~., ! ,-- . ~ ~..~...,.
6 km
I I
1. Citrusdal ("The baths")
N S
Klein Swartberge Calitzdorp spdng
I"
4 km \ BS H.,,,vie org0 \
- 6 km , , 2. Calitzdorp
$W NE
Brandvlei sprtng
"-I.' '--.
Malmesbury
4 km Group
5
km
| J
granite ~ TMG
3. Brandvlei
Ftgure 2. Cross-sections which illustrate the sub-surface geology at Citrusdal, Ca#tzdorp and BrandvleL Sections were drawn from
published survey maps (Diamond, 1997J. MG: Malmesbury Group; hiS." Nardouw Subgroup; CF: Cederberg Formation; BG: Bokkeveld
Group; WG: Witteberg Group; PF: Peninsula Formation; BS: Bidouw Subgroup; CS: Ceres Subgroup; TMG: Table Mountain Group.
using a Finnegan MAT252 mass spectrometer, and
the fractionation factor between CO 2 and water at
25°C was assumed to be 1.0412 (Coplen, 1993).
Data are reported in the familiar 5 notation, relative
to SMOW, where 5 = (R mp,,/RsMow-1)'lOOO, and
R = 180/180 or D/H. The average difference between
duplicates of internal water standard (CTMP) over
the course of this research was 0.48%0 for H (n = 23)
and O. 10%o for O (n = 18). These correspond to values
of 2a of 0.74%0 and O.14%o, respectively. The
standards V-SMOW and SLAP were analysed to
determine the degree of compression of raw data,
and the equations of Coplen (1993) were used to
convert raw data to the SMOW scale. Our internal
water standard (CTMP 5D =-9%; 5180 =-2.85%o),
which had been calibrated against V-SMOW and SLAP
and independently analysed, was run with each batch
of samples and used to correct for drift in the refer-
ence gases.
The gas samples were analysed as follows: the
sample bottle was placed onto the vacuum line and
the condensable gases were collected in a U-trap
immersed in liquid N. The line was then opened to a
furnace containing CuO at 700°C so that any CH 4
present would be converted to CO=. The liquid N was
replaced by frozen isopropyl alcohol, and the dry CO 2
Journal of African Earth Sciences 471
R. E. DIAMOND and C. HARRIS
was collected in a second U-tube. From there, the
CO, was frozen into a break seal tube for analysis.
The Brandvlei and Calitzdorp gas samples appeared
to be dominantly CO, based on the relative proportion
of gas frozen directly into liquid N, The Malmesbury
gas contained about 20% CH,. The standards NBS1 9
(calcite) and NBS21 (graphite) were used to convert
the raw data to the PDB scale. The 613C measured in
this way is that of the total C present (CH, + CO,).
Thermal springs RESULTS
Water 6D and 6180 values are presented in Table 2.
There is no correlation between isotope ratios and
temperature or altitude of the spring (Fig. 3). However,
on the V’O versus temperature plot, there are two
distinct groups of samples, which show a negative
correlation. The springs plotting in the upper group
are Malmesbury, Goudini, Caledon, Citrusdal and
Brandvlei, all of which are found in the belt of
mountains closest to the coast (the’coastal group’).
The springs, which plot on the lower group, are found
in the mountain belts further inland (see Fig. 1 I. This
negative correlation is less strong for 6D versus
temperature. There is no correlation between isotope
ratios and height above sea level. However, the three
highest altitude springs (700-800 m) have a
significantly lower mean 6D and 6180 than the other
springs.
For those springs less than 200 km from the west
coast, there is a good correlation between isotope
ratios and distance from the west coast (Fig. 41, which
is clearly not influenced by the altitude of the spring.
Rietfontein (700 m asl), which is the only spring in
this study from the Karoo region, has an anomalously
high 6180 compared to the other springs > 200 km
from the west coast.
The variation in 6D and 6’*0 values of the Brandvlei,
Calitzdorp, Citrusdal and Malmesbury over eight
0
z
Go
-20
-30
\Ma - --..
\ Cit ‘.\
’ F----~)ci,
ma 1,
\
‘1
‘1 \ !
1% Coastal group
‘\ \
‘\
- ‘\ \ G o’*\ ‘\
. . O-
-i Wi Wi a
!
-!
_I 0 0 Bad
-40 - i
i R Bad 0
- 10 Cali
- ‘\ --_ ww
-.-._._. -.-._-___./
,....l....l....l.... I * I n I I I -
I *--..
..‘\o xb,.\Coastal group
Ma\ \ i \
iiCit 0 .,
i ‘\
‘\
BA,,o o,j
._._.--0. Cale k 0
Wi I
v
Bad 0 0 -0
-._
“Qyw ww- IO J
Il....lly group .-.__ Cali 0
--.-__/
I....I....l....I..~I I , I I I I I ,
30 50
Tern;erature (“C)
60 200 400 600 600
Heiaht a.s.1 (rn)
Figure 3. Plot of 6D and 6”O values of thermal springs versus temperature and height of spring above
sea level. The coastal and inland groups of thermal springs are indicated. Ma: Malmesbury; G: Goudini;
Cale: Caledon; Cit: Citrusdal; Br: Brandvlei; Bad: Baden-Baden; Cali: Calitzdorp; MO: Montagu; R:
Reitfontein; To: Toowerwater; Wi: Witzenberg; WW: Warmwaterberg.
472 Journal of African Earth Sciences
Table 2. Hydrogen and oxygen isotope data for sampled thermal springs
T Goudini
6D S’s0
Malmesbury
6D S’s0
Montagu
6D S’s0
Rietfontein
6D 6’*0
Toowerwater
SD 6°C
I Baden-Baden Brandvlei Caledon Calitzdorp Citrusdal
1995 6D 6’8o 6D s’% 6D 8°C SD S”C 6D 6°C
Feb -44 -6,s
March -33 -5.6 -31 -5.5 -16 -4.8
April -27 -5.5 -22 -4.6
Mav -32 -5.8 -37 -6.5 -19 -5.7
June -30 -5.0 -39 -7.7 -19 -4.1
July -28 -5.5 -40 -9.0 -22 -6.0
Aug -31 -6.1 -36 -5.8 -18 -5.0
Sept -31 -5.6 -42 -6.8 -22 -4.9
Ott -28 -5.9 -41 -8.5 -20 -4.1
1997
March -37 -6.9 -34 -6.4
mean -37 -6.9 -30 -5.6 -31 -5.5 -40 -7.3 -18 -3.9 -33 -6.4 -42 -5.2
M&V -31 -6.1 -6.2 -35 -7.4 -18 -4.2 -7.1
Samples were collected during the month indicated. M&V: Data from Mazor and Verhagen (1983) for samples collected in 1971-1972.
-23 -3.7
-20 -3.7
-15 -2.8
-18 -3.8
-18 -4.3
-17 -3.8
-20 -4.5
-17 -4.3
-33 -6.4 -41 -6.9
-26 -4.4
-42 -5.2
-41 -6.9
Warmwaterberg Witzenberg
SD S’*O 6D S”O
-46 -7.0
-30 -5.6
-46 -7.0 -30 -5.6
R. E. DIAMOND and C. HARRIS
-20 - y 0
\. & = o.aa
-30 -
is a ‘10
o‘\
-40 - 0 0 0
0
-50 - 1. 9. I I I. I I I I .,. I I.,.,
_ 120m
-4 - Q ., 290m
r = 0.95 \O
-5 - 250m
tk!A
700m
,o 360m 0
220m
k -6- 8 \ m 280m
0 500m 8OOm
-7 - 280m 0 200m 0
0
-0 -
0 I. .I. I. I I *I,,.,,,,.,,
100 200 300 400 500
Distance from West Coast (km)
F&we 4. Plot of 6D and 6180 of thermal spring water versus distance from the west coast
of southern Africa measured in an east-west direction. Lines of best fit and correlation
coefficients are given for thermal springs situated at altitudes <2DD m.
months of 1995 is shown in Fig. 5. All four springs
show variations that are larger than the expected
analytical errors of f 1 .O and + 0.1 %O for 6D and
6180, respectively. In the case of Calitzdorp and
Malmesbury, there is a reasonable degree of
correspondence between the 6D and PO values,
which indicates that analytical error alone is not
the cause of the variation, since the methods of
analysis for 0 and H are completely separate. These
springs were sampled from pools fed by the spring,
and the variations in 6D and PO values could have
been caused by varying degrees of evaporation from
the pool. There is no evidence for any systematic
difference in isotope ratios between summer and
winter, which indicates that the spring waters
originate from extensive aquifers which are unaffected
by seasonal changes in the SD and al80 values of
rainwater.
The springs show a good correlation between the
average 6D and 6180 values (Fig. 51, Rietfontein
again being a significant outlying point. The line of
best fit through the data calculated using the re-
duced major axis method (RMA: Rock, 1988) has
the equation 6D = 7.816’*0 + 12.45. If Rietfontein
474 Journal of African Earth Sciences
is excluded from the data, the equation becomes
SD = 7.48PO + 11 74 . .
Gas data
The 613C values obtained for samples of gas
discharged with the spring water are given in Table
3. The gas was collected at all the springs where the
water discharges directly from the ground upward
into a pool above and the collection of gas bubbles
was possible. The quantities of gas bubbling up
appear to be proportional to the water discharge, with
Brandvlei releasing on the order of a litre or so of gas
every second, Calitzdorp significantly less and
Malmesbury releasing streams of bubbles every few
seconds of up to only a few millilitres each. The 613C
values range from -21.5 to -23.2%0 compared to
typical P3C values for volcanic and geothermal gas
CO, of 0 to -11960 (Taylor, 1986).
Dl8ClJ88lON
Carbon isotopes in gas bubbles
The gas from the three springs analysed (Brandvlei,
Calitzdorp and Malmesbury) yielded 613C values
Oxygen and hydrogen isotope geochemistry of thermalsprings of the Western Cape, South Africa
Calitzdorp
0
Ciirusdal
-20 >
/
cl 8’” kJ\~~@+-& 0
Malmesbury
P
erg -so /“\o,o~o~o_olo
0
Figure 5. Variation of 6D and Sr80 values of Malmesbury, Citrusdal, Brandvlei
and Calitzdorp thermal springs with the month each was samDIed.
between -21.5 and -23.2%0, which clearly labels the
C as being of organic origin (Dai et a/. , 1996). Mazor
and Verhagen (I 983) obtained a range of 613C values
from -16.6 to -24.5% for dissolved bicarbonate in
some Western Cape springs. The data for Malmes-
bury (-16.6960) and Brandvlei (-18.9%0) of Mazor and
Verhagen are significantly higher than the data
obtained during the present work. This is probably
due to differences in the material analysed, viz. gas
bubbles (this work) versus dissolved bicarbonate.
The large C isotope fractionation between CO, and
CH, (&J~_cH~ = +70%0 at 20°C: Bottinga, 1969)
means that in a system where the 613C value of the
total C present remains constant, the 613C values of
the dissolved bicarbonate will increase as the CH,/
CO, ratio increases. The 613C value of bicarbonate in
Malmesbury water (Mazor and Verhagen, 1983) is
the least negative, which is consistent with our
observation that the gas sample contained significant
quantities of CH,.
Despite the problems in interpreting the 613C values
of the mixtures of CO, and CH, without knowing the
quantities of each gas present, these data are im-
portant in the context of the present study because
they confirm a non-volcanic origin and support the
conclusions of Mazor and Verhagen (1983) that the
C is of an entirely biogenic origin. Mazor and
Verhagen (1983) concluded that “no significant
exchange with 14C-free aquifer materials has taken
place”. This seems reasonable given that rocks of
the Cape Supergroup contain very little carbonate
material. The Cape Mountains are known for their
nutrient poor, structureless and nearly topsoil-free
soils. There are, however, flat areas that become
waterlogged in winter and have black organic-rich
soils. These soils would tend to be reducing, as well
as having a large supply of C. The fynbos (heath-like)
vegetation that grows on the Cape Mountains is
distinctive in producing fulvic and humic acids, which,
if present in sufficient quantities, stain the water
Table 3. Stable isotope data for gas
Spring 6%
Brandvlei -22.7
Calitzdorp -21.5
Malmesbury -23.2
Carbon isotope ratios were measured on the total C present
in gas bubbles collected from the spring water.
Journalof Afrfcan Earth Sciences 475
R. E. DIAMOND and C. HARRIS
reddish-brown. These organic compounds and possibly
others could allow the water to contain appreciable
dissolved organic C, which is released during heating
of the water at depth.
Long-term changes in 6D and 6180
The data presented here are similar, but not identical,
to the data of Mazor and Verhagen (I 983) obtained
on samples collected in 1971 and 1972. A limited
amount of H isotope data (four springs) are available
for comparison, and on average the 6D values of
Mazor and Verhagen are slightly higher. The 6180
values reported in this paper are generally 0.3-0.5%
lower than those reported by Mazor and Verhagen
(1983). It is possible that these differences reflect
long-term changes in the isotope composition of
recharge due to climate change, but any shift in S’*O
values with time ought to be accompanied by a
similar shift in 6D values, and this is not observed.
Thus, differences in 6D and PO between the 1971/
2 and 1995/7 samples are far more likely to be a
function of analytical procedures employed by the
two laboratories involved.
Comparison with meteoric water
One of the main conclusions of Mazor and Verhagen
(1983) was that the thermal springs have system-
atically lower 6D and ZPO values than rivers sampled
in the same area at the same time and, hence,
ambient meteoric water. However, as acknow-
ledged by Mazor and Verhagen (19831, this con-
clusion is weakened by the probability that seasonal
variations in the 6D and 6’*0 values of the rivers
exist, as well as possible isotope gradients, with
water depth. The 6D and PO values of rivers might
not, therefore, be a good approximation to the
integrated annual rainfall in a particular area.
The isotope data for the springs had been chosen
to be compared with data for ambient meteoric water.
The ideal comparison would be with rainwater
collected at the spring site over a period of several
years, but such data are not available. The
International Atomic Energy Agency database (IAEA,
1997) has a monthly record for Cape Town
International (formerly D.F.Malan) Airport from 1962-
1974, and Diamond [I 997) and Diamond and Harris
(I 997) reported monthly 6D and PO values for the
University of Cape Town (UCT) and elsewhere in the
Western Cape. The rainfall data are compared to the
thermal spring data on Fig. 5, and it can be seen that
the springs have systematically lower 6D and 6180
values compared to the rain. The weighted mean
annual 6D and 6180 values for UCT and the IAEA
data are plotted, and it can be seen that they are
476 Journal of African Earth Sciences
significantly higher than the thermal spring values.
Rain data from inland stations at Oudtshoorn, Citrusdal
and Tulbagh are not complete annual records;
nevertheless, they all include the winter months when
rainfall is highest and temperatures are lowest. Hence,
the weighted mean 6D and PO values from these
rainwater collecting stations ought to be somewhat
lower than the weighted mean annual values.
The Malmesbury spring has 6D and S180 values,
which are only slightly lower than the mean annual
rainfall value for UCT. Malmesbury is 70 km north
of Cape Town and further inland. Hence, the data
are consistent with the spring being recharged by
ambient rainwater. The situation is similar for both
the Citrusdal and Witzenberg Springs. The spring
water has slightly more negative 6D and PO values
than the measured rain data. The average spring
6D and 6’*0 values for Citrusdal are -20 and -4.9%
compared to the weighted mean for rain (Diamond,
1997) of -11 and -4.4%0. The average spring 6D and
6’*0 values for Witzenberg are -30 and -5.5960 com-
pared to the weighted mean for rain (Diamond,
1997) for Tulbagh of -20 and -5.1960. The Calitzdorp
Spring has the lowest 6D and 6180 values of all the
springs analysed and these values (6D and 6180
equal to -40 and -7.3%0, respectively) are consider-
ably lower than rainfall at Oudtshoorn, 40 km east
of Calitzdorp Spa, at the same altitude (weighted
mean 6D and al80 equal to -11.6 and -4.1 %o, respec-
tively). No data for rainfall in the vicinity of Montagu,
Baden-Baden, Warmwaterberg, Toowerwater and
Rietfontein exist, but there is no reason to suppose
that it should be significantly different from the
analysed rainfall samples. It is, therefore, concluded
that most of the thermal springs have isotope ratios
that are significantly lower than ambient rainfall.
Isotope exchange between rock and water
As discussed above, the 6D and 6180 values of the
hot springs are generally lower than ambient rainfall.
In addition, the springs plot slightly below the local
meteoric water line (Fig. 6). One possible explanation
for this is that the 6’*0 values of the springs increased
as a result of the exchange of 0 between the water
and the rocks through which they passed. This is
commonly observed in geothermal waters of volcanic
regions (e.g. Sheppard, 1986). Water-rock interaction
usually affects 6’*0 values but not 6D values because
rocks generally consist of 50 wt% 0 and very little
H. The potential shift in 6180 value of the thermal
water is dependent on the 0 isotope fractionation
factor between the rock and water, temperature and
the 6180 value of the rock.
The fractionation factor between quartz (the do-
minant mineral in the rocks) and water is large at low
Oxygen and h ydrogen isotope geochemistry of thermat springs of the Western Cape, South Africa
X IAEA
+ UCT
0 Citrusdal Rain
A Oudlshoorn
UCT weighted average _
IAEA weighted average
0 Rielfontein
Figure 6. Plot of 6D versus 61B0 for thermal springs and rainwater from various places. All rain
data are integrated monthly samples; the UCT data are for a two year period (Diamond and
Harris, 1997J and the IAEA data for most (but not al// months between 1962 and 1974 (IAEA,
1997); the Citrusdal, Oudtshoorn and Tulbagh data are for March-October 1995. The weighted
annual mean values for the UCT and IAEA collection stations are shown and the line of best fit
through the rain data is from Diamond and Harris f 1997).
temperatures (Aqu,.__ = 3.38.1 06*T2-3.4, where T
is the temperature in K: Clayton et al., 1972). This
translates to a difference between quartz and water
6180 values of 25.3%0 at 70°C. The sandstones and
quattzites of the Cape Supergroup have average PO
values of 10.91960 (n = 28: Diamond, 1997) and the
Malmesbury Basement has an average 6180 of
13.06% (Harris et al., 1997). It therefore follows
that the PO values of water in equilibrium, with
average Malmesbury Group Basement and Cape
Supergroup, would have been -12.3 and -14.4% at
70°C. Any change in PO value of water as a result
of interaction with rocks at this temperature would
have been to lower, not higher, values and such
exchanged waters would plot to the left of the
meteoric water line on Fig. 6. In order to cause shifts
to higher PO values in the water, interaction would
have had to take place above about 1 OO’C because
at this temperature the waters are in approximate
0 isotope equilibrium with the average Table
Mountain Group. This temperature is much hotter
than any of the thermal springs, and it is therefore
concluded that water-rock interaction did not affect
their 6180 values. In any case, 0 isotope exchange
at such low temperatures is likely to have been
sufficiently slow that water-rock interaction has no
effect on isotope ratios.
Comparison with groundwater
In this study, the thermal spring data has been
compared with data (Harris et al,, 1999) from cold
springs issuing from the lower slopes of Table
Mountain (next to UCT; Fig. I) and water sampled
from boreholes in the area around Victoria West
Jwmal of African Earth Sciences 477
R. E. DIAMOND and C. HARRIS
(altitude 1200 m; Fig. 1) in southwest Karoo (C. Harris
and S. Peth, unpub/. data). These data give some idea
of the range of 6D and 6’*0 values of unheated
groundwater as one proceeds from the west coast
inland and are compared to the thermal springs in
Fig. 7. The Victoria West water samples were taken
from various depths (O-250 m) from a number of
boreholes drilled by the Department of Water
Affairs and Forestry in the area. Those samples
from > 1 DO m tend to have lower 6D and 6’*0 values
than samples from < 100 m and this is most likely
to be caused by selective recharge of deep waters
by heavy rainfall events.
The Table Mountain springs plot close to the
Western Cape meteoric water line, whereas the
Victoria West borehole waters form an array which
is approximately parallel to the Western Cape
meteoric water line with much lower 6D values for a
given 6180 value. The equation of best fit through the
Victoria West data has the equation 6D = 6.9@0
-1.8. The negative intercept value is uncharacteristic
of meteoric water data arrays and may reflect signi-
ficant evaporation in the near surface environment
during recharge. Note that the Rietfontein thermal
spring, which is geographically closest to Victoria
West, and which is situated in the southern Karoo
region to the north of the Cape Fold Belt, has 6D and
6180 values which lie within the range shown by the
Victoria West ground-waters. For the most part, the
thermal springs plot between the lines of best fit
through the Table Mountain and Victoria West data
but generally have lower 6D and 6180 values.
A Table Mountain
0 Victoria West c 1OOm
0 Victoria West z 1 OOm
Line ot best-lit through Vic.Wsst data
Fm 7. Comparison of thermal spring SD and 6180 values with those of cold springs on the lower
slopes of Table Mountain (Harris et al., 19991 and groundwater from the area around Victoria West
(Harris and Peth, unpubl. data). Meteoric water line for Western Cape is from Diamond and Harris
11997). The line of best fit through the Victoria West data was calculated using the RMA method
(see text). The Global Meteoric Water Line of Craig /1961/ is shown for reference.
478 Journal of African Earth Scfences
Oxygen and hydrogen isotope geochemistry of thermalsprings of the Western Cape, South Africa
Origin of low 6D and 6’*0 values
The comparison of 6D and PO values between the
thermal springs and meteoric and groundwater water
samples confirms that the thermal springs have
significantly lower 6D and Pa0 values than ambient
rainwater, Various combinations of the following may
be responsible for these low 6D and PO values:
i) the continental effect (e.g. Dansgaard, 1964);
E) selective recharge during periods of abnormally
high rainfall (as suggested by Mazor and Verhagen,
1983);
i@) recharge during an earlier period of colder climate;
and
iv) recharge at higher altitude.
The continental effect cannot account for low 6D and
PO values of the thermal springs because they have
lower 6D and al80 values than the groundwater at
Victoria West, which is further inland. Mazor and
Verhagen (I 983) concluded that the springs were
selectively recharged by direct rain infiltration after
heavy rains without any evaporation or averaging
associated with rivers. Heavy rain events generally
produce rain that has more negative 6D and PO
values than normal rainfall at the same place (the
‘amount effect’ of Dansgaard, 1964). Selective re-
charge by heavy rain events is the likely cause of the
differences in isotope composition between the deep
and shallow groundwaters at Victoria West, but this
effect is too small to account for the observed isotope
differences between thermal springs and ground-
water. The possibility that the springs were recharged
during a colder climate regime was rejected by Mazor
and Verhagen (1983) because of the lack of
correlation between 14C data (as a proxy for time)
and 0 and H isotope ratios.
There remains the possibility that high average
altitude of recharge is the cause of the low isotope
ratios of the thermal springs. It is well known that
the 6D and PO values of rainfall decrease as altitude
increases (Dansgaard, 1964). Midgley and Scott
(1994) reported an altitude effect on PO of -0.32%
per 100 m for the Jonkershoek Mountains, about 70
km east of Cape Town. At Calitzdorp, the possibility
exists that the zone of recharge of the spring could
be in the Klein Swartberg Mountains to the north,
which rise up to 2000 m (Fig. 2). The difference
between the PO value of the spring and Oudtshoorn
rain is 3.2%0, which could be interpreted as the
recharge zone being on average 1000 m higher than
the spring that is at about 1200 m.
Regiial variation
The small number of thermal springs available for
analysis preclude a detailed discussion on the regional
variation of their 6D and PO values. Nevertheless,
the stable isotope data present several interesting
features. The most obvious feature is the apparent
effect of continentality, whereby the 6D and S’80
values decrease with increasing distance from the
west coast. The difference between the Table Moun-
tain Springs data and the Victoria West groundwater
data illustrate a second effect, that is a much lower
‘deuterium excess’ (d), where d=6D-84’*0 for a
given data point (Dansgaard, 1964; Whelan, 1987)
for the inland groundwater. Regardless of whether
the low y-axis intercept value for the line of best
fit through the Victoria West data is indicative of
evaporation prior to recharge, the thermal springs
also show a similar decrease in deuterium excess
as their distance from the west coast increases.
The apparent grouping of thermal springs into
coastal and inland groups (Fig. 31, which both show a
negative correlation between EPO and water tem-
perature, is more difficult to explain in the light of the
observations made above. Within each group, higher
temperatures of spring water can only be explained
by circulation of water to greater depths. As dis-
cussed above, lower 6D and PO values can generally
be explained by recharge at higher altitude, thus the
data are consistent with the higher temperature
springs being recharged at higher altitude. This is to
be expected as a greater depth of circulation would
be expected in aquifers with a greater hydraulic head
of water. The correlation between iY*O values and
distance from the west coast in the coastal group
must, therefore, reflect an increase in the average
altitude of recharge with increasing distance from the
coast and is not simply due to the continental effect.
The inland group of thermal springs shows a negative
correlation between al80 values and water tem-
perature with a similar gradient but with 6180 values
about 2%0 lower for a given temperature. This offset
is presumably due to the greater ‘continentality’ of
these springs. The lack of correlation between
distance from the west coast and isotope ratios in
those springs > 200 km from the west coast (Fig. 4)
may, in part, be due to the change in geometry of the
Cape Fold Belt from east to west. The coastal group
of thermal springs is located in mountain belts which
trend north-south, perpendicular to the movement of
weather systems, whereas the inland group is
situated in mountain belts which trend east-west.
CONCLUSIONS
The authors agree with previous work by Mazor and
Verhagen (1983) that the source of water in the
Western Cape thermal springs is meteoric in origin
and that there is no evidence for water-rock interaction
having any effect on 0 isotope ratios. No systematic
Journal otAtriwn Eah Sciences 479
R. E. DIAMOND and C. HARRIS
changes in 6D and ZY80 values were detected over a
period of eight repeated samplings, suggesting that
the aquifers contain significant volumes of water
which are not affected by seasonal changes in the
6D and ?Y80 values of rainfall. The main feature
distinguishing the thermal springs from ambient
meteoric water is the significantly lower 6D and 6180
values. Although the isotope ratios of the thermal
springs become progressively more negative with
increasing distance from the west coast (for the first
200 km), it appears that high average recharge altitude
is the most important factor responsible for the low
6D and 6180 values .
ACKNOWLEDGEMENTS
The authors are grateful to the FRD for financial
support in the form of a studentship to RED, and a
core grant to CH. The authors are indebted to their
water samplers, Captain D.C. Taljaard of Brandvlei
Prison, Worcester, Mr H. van Huysteen of the
Caliizdorp Spa, Mr M. Gordon of The Baths, Cirusdal,
Mr B. Beylevelde of Citrusdal, Mrs K. of Oudtshoorn
and Mrs V. Humphris of Tulbagh. They are also
grateful to all the personnel at the other thermal
springs for allowing them to take water samples. K.
Faure, P. Dennis, A. Issar, I. Cartwright, B. Verhagen,
S. Talma and J. Weaver are thanked for helpful
discussions and comments at various stages of this
work. This paper was written by CH during periods
of sabbatical leave at Monash University, Australia
and Universite Jean Monnet, St. Etienne, France.
Finally, the authors are again indebted to F. Rawoot
for help with the analytical work.
Editorial handling - I? Bowden
RBZRENCXS
Armstrong, R.A., de Wit, M.J., Reid, D.L., York, D.,
Zartman, R., 1998. Cape Town’s Table Mountain reveals
rapid Pan-African uplift of its basement rocks. Journal
African Earth Sciences 27 (1 A), 10-l 1.
Bottinga, Y ., 1969 Calculated fractionation factors for
carbon and hydrogen exchange in the system calcite-
CO,-graphite-methane-hydrogen-water vapour. Geo-
chimica Cosmochimica Acta 33, 49-64.
Broquet, C.A.M., 1992. The sedimentary record of the
Cape Supergroup: A review. In: De Wit, M.J., Ransome,
I.G.D. (Eds.), Inversion Tectonics of the Cape Fold Belt,
Karoo and Cretaceous Basins of Southern Africa.
Balkema, Rotterdam, pp. 159-183.
Chamberlain, C.P., Zeitler, P.K., Barnett, D.E., Winslow,
D.. Poulson, S.R., Leahy, T., Hammer, J.E., 1995.
Active hydrothermal systems during the recent uplift of
Nanga Parbat, Pakistan Himalaya. Journal Geophysical
Research 100, 439-453.
Clayton, R.N., O’Neil, J.R., Mayeda, T.K., 1972. Oxygen
isotope exchange between quartz and water. Journal
Geophysical Research 77, 3057-3067.
480 Journal of Aftfcan Earth Sciences
Coplen, T.B., 1993. Normalization of oxygen and hydrogen
isotope data. Chemical Geology (Isotope Geoscience)
72, 293-297.
Craig, H., 1961. Isotopic variations in natural waters.
Science 133, 1702-l 703.
Dai, J-X., Song, Y., Dai, C-S., Wang, D-R., 1996.
Geochemistry and accumulation of carbon dioxide gases
in China. American Association Petroleum Geologists
Bulletin 80, 1615-l 626.
Dansgaard, W., 1964. Stable isotopes in precipitation.
Tellus 16, 436-468.
Diamond, R.E., 1997. Stable isotopes of the thermal springs
of the Cape Fold Belt. MSc. thesis funpubl.), University
of Cape Town, 82~.
Diamond, R.E., Harris, C., 1997. Oxygen and hydrogen
isotope composition of Western Cape meteoric water.
South African Journal Science 93, 371-374.
Doel, S.L., 1995. Cango Caves stable isotope stratigraphy.
B.Sc. (Hans) thesis funpubl.), University of Cape Town,
45p.
Gresse, P.G., Theron, J.N., Fitch, F.J., Miller, J.A., 1992.
Tectonic inversion and radiometric resetting of the
basement in the Cape Fold Belt. In: De Wit, M.J.,
Ransome, I.G.D. (Eds.), Inversion Tectonics of the Cape
Fold Belt, Karoo and Cretaceous Basins of Southern
Africa, Balkema, Rotterdam, pp. 217-228.
Hllbich, I.W., 1992. The Cape Fold Belt Orogeny: State
of the art 1970’s_1980’s. In: De Wit, M.J., Ransome,
I.G.D. (Eds.), Inversion Tectonics of the Cape Fold Belt,
Karoo and Cretaceous Basins of Southern Africa.
Balkema, Rotterdam, pp. 141-l 58.
Harris, C., Faure, K., Diamond, R., Scheepers, R., 1997.
Oxygen and hydrogen isotope geochemistry of S- and I-
type granitoids of the Cape Granite Suite, South Africa.
Chemical Geology 143, 95-l 14.
Harris, C., Oom, B.M., Diamond, R.E., 1999. A pre-
liminary investigation of the urban isotope hydrology of
the Cape Town area. Water South Africa 25, 15-24.
Hoffmann, J.R.H., 1979. Die chemiese samestelling van
warmwaterbronne in Suid- en Suid-wes Afrika. CSIR,
Pretoria, Special Report WAT 56A.
IAEA, 1997. Station 6881600 ‘Malan’ (Cape Town) South
Africa. Global Network of Isotopes in Precipitation (GNIP),
http:Ilwww.iaea.or.atlprogramslrilgniplgnipmain.htm
Jones, M., 1992. Heat flow in South Africa. Handbook
Geological Survey 14, 174~.
Kent, L.E., 1949. The thermal waters of the Union of
South Africa and South West Africa. Transactions
Geological Society South Africa 52, 231-264.
Mazor, E., 1991. Applied Chemical and Isotopic
Groundwater Hydrology. Open University Press, 274~.
Mazor, E., Verhagen, B.T., 1983. Dissolved ions, stable
and radioactive isotopes and noble gases in thermal
waters of South Africa. Journal Hydrology 63, 315-
329.
Midgley, J.J., Scott, D.F., 1994. The use of stable iso-
topes of water ID and ‘8O) in hydrological studies in the
Jonkershoek Valley. Water South Africa 20, 151-l 54.
Rock, N.M.S., 1988. Lecture Notes in Earth Sciences,
18: Numerical Geology. Springer Verlag. Berlin, 427~.
SAWB (South African Weather Bureau), 1996. The
weather and climate of the extreme south-western Cape.
Department of Environmental Affairs and Tourism, South
Africa.
Sheppard, S.M.F., 1986. Characterization and isotopic
variations in natural waters. In: Valley, J.W., Taylor
Jr, H.P., O’Neil, J.R. (Eds.), Stable Isotopes in High
Temperature Geological Processes. Reviews in Miner-
alogy 16, pp. 165-183.
Oxygen and h ydrogen isotope geochemistry of thermal springs of the Western Cape, South Africa
Socki, R.A., Karlsson, H.R., Gibson Jr, E.K., 1992.
Extraction technique for the determination of oxygen-
18 in water using pre-evacuated glass vials. Analytical
Chemistry 64, 829-831.
Taylor, 8.E.. 1986. Magmatic volatiles: isotopic variation
of C, H, and S. In: Valley, J.W., Taylor Jr, H.P., O’Neil,
JR. (Eds.), Stable Isotopes in High Temperature Geological
Processes. Reviews in Mineralogy 16, pp. 185-225.
Theron, J.N., Gressa, P.G., Siegfried, H.P., Rogers, J.,
1991a. The geology of the Cape Town area-explanation
of sheet 3318. Geological Survey, Republic of South Africa,
14op.
Wefhan, J.A., 1987. Stable isotope hydrology. In: Kyser,
T.K. (Ed.), Short Course in Stable Isotope Geochemistry
of Low Temperature Fluids. Mineralogical Association of
Canada 13, pp, 129-161.
Joum.slofAf&an Eanh S&cas 481
