Jökull - 01.12.1985, Blaðsíða 8

phere whereas the sulfur yield was negligible. Our
results (Devine et al. 1984) support this observation in
that they indicate that the mass of chlorine, and in some
cases fluorine, may be larger than the sulfur mass
degassed during some eruptions. „Volcanic eruptions
can be divided into two groups in terms of volatile
composition, expressed as H2SO4/HCI ratio. Firstly, the
sulfur-rich eruptions, with ratio greater than 20. These
include all the Icelandic eruptions studied to date and
may be typical of mantle-fed rift-zone volcanism.
Secondly, the halogen-rich eruptions, with ratio less
than 3, characteristic of the volcanic arc systems. No
intermediate types are present in our data set. A third
type are the exceptionally sulfur-rich events, where
decomposition of anhydrite or other sulfur-bearing
minerals dominates the aerosol composition (E1
Chichon-type) and the contribution of degassing from
the magma is relatively minor.
Although present in high concentration in the near-
field eruption plume, the halogens may be fractionated
out early and thus their importance in the stratospheric
volcanic aerosol is unknown. Direct chemical analyses
of the acidity layers in Greenland ice cores indicate,
however, that some chlorine compounds may persist.
The acidity layer from the Eldgja 934 A.D. eruption in
the Crete core contains at least 65% HCl (Hammer
1980) and Herron (1982) has also shown high levels of
both C1 and F in the Dye 3 Greenland ice core layer
from this eruption. Finally, data from this core show an
acidity peak in early 1889, where the concentration of
C1 is at least twice that of the sulfate anion (Herron
1982). This acidity peak, which may have originated
from the 1888 eruptions of Bandai San or Ritter Islands,
is indication that chlorine may be an important compo-
nent in some volcanic aerosols. In contrast, studies of
modern volcanic aerosols have generally shown that the
halogens are absent or present only as trace gases,
compared to sulfur.
As pointed out above, the Cl' peak in the Greenland
ice core acidity layer from the Laki eruption of 1783
does not correspond with a Na peak and thus sea-salts
can be excluded. Other direct evidence for volcanoes as
an important source of HCl in the stratosphere has
recently come from studies of the volcanic aeerosol
cloud resulting from the E1 Chichon eruption. Mankin
and Coffey (1984) have shown a 40 percent increase in
HCl in the stratosphere following the 1982 E1 Chichon
eruption, or equivalent to about 4xl010 g HCl. The
original suggestion of Stolarski and Cicerone (1974) of
direct injection of chlorine into the stratosphere by
volcanoes has now gained credibility through the study
of the E1 Chichon eruption and as suggested by Mankin
and Coffey (1984) should lead to a reassessment of the
role of volcanoes in determining the stratospheric che-
mistry of chlorine.
Accurate petrologic estimates of the mass of sulfur
and halogens degassed from magma during volcanic
eruptions require a variety of data on the petrology and
geologic features of the deposit. Firstly, the mass of
erupted material must be determined by geologic mapp-
ing of the volcanic deposit. In the case of lava eruptions
this is straight-forward and subject to small error,
whereas volume-estimates of tephra from explosive
eruptions can be in error by an order of magnitude.
Secondly, the mass of erupted liquid must be deter-
mined by subtraction of the mass of crystals and lithics
from the total erupted mass. This requires knowledge of
the modal composition of the entire deposit. The com-
bined volume of these components is rarely more than
10%, but tephra from some eruptions may contain up to
40% crystals, e.g. Mount St. Helens, 1980 (Carey and
Sigurdsson 1984) and up to 60% lithics, e.g. in the 1979
Soufriere eruption (Sigurdsson 1982b). Third, the com-
positional variation in erupted magma must be deter-
mined, and the proportions of the two magmas estab-
lished, in the case of eruptions of mixed magmas or
compositionally zoned deposits. Finally, the pre-erup-
tion (glass inclusion) and post-eruption (matrix glass)
concentration of sulfur and halogens can be determined
by electron microprobe analysis. We have found by
analysis of standards, that accuracy of sulfur and chlor-
ine microprobe analyses is about 1 to 6% of the amount
present and precision (lo) about 2 to 7%, with detec-
tion limits of 50 ppm (Devine et al. 1984). The microp-
robe analysis for fluorine is much more difficult and our
detection limit for this element is only 300 ppm, with
accuracy and precision of 3 to 5% for silicic glasses. In
practice, the largest potential errors in petrologic esti-
mates of the mass of sulfur and halogen degassing from
magmas do not therefore relate to the analysis of these
elements in the glass, but rather are due to inadequate
information on the total mass and modal composition of
erupted tephra. Very few deposits have yet been stu-
died in sufficient detail to permit such a rigorous analy-
sis as outlined above, and all published petrologic esti-
mates of volcanic volatile mass must therefore be
regarded as preliminary. We find the correspondence
between our estimates and those determined by other
methods encouraging, however, indicating that the
potential errors should not prevent the application of
this method in further examination of the link between
volcanism and climate changes in the geologic record.
6 JÖKULL 35. ÁR