Jökull - 01.12.1999, Qupperneq 97

culated for this model fits the measurements reason-
ably well but the fit is unsatisfactory at the northern
end. The second model (Fig. 5b) is based on Kjartans-
son (1966), with the island having a large core of pil-
low lava. In order to match the observed field, the
density of the tuff cone had to be less than that used in
the first model. However, the overall density appears
to be too high and the model fails to account for the
rise in gravity at the far ends of the profile. The third
model (Fig. 5c) uses a cross-section made by Einars-
son (1968, p. 61), with a small pile of pillow lava in
the center and a conduit of relatively high-density
volcanics forming a plug under the crater. The fit is
better than in the second model but again it fails to ac-
count for the rise of the field at the ends. The fourth
model (Fig. 5d) is our modification of the first model.
The northern peninsula has been modeled as being
made up of relatively high-density sediments. The in-
clusion of this body is realistic since the peninsula is
known to be made mainly of rock fragments eroded
from the lava cliffs on the southem side of the island.
Importantly, the model fits the section drawn by
Jakobsson and Moore (1982) on the basis of the re-
sults from the 1979 drill hole. The model of Fig. 5d is
the most plausible of the four.
DISCUSSION
The results of the gravity modeling agree well with
the model proposed by Jakobsson and Moore (1982).
This suggests that anomalous bodies are not present,
i.e. that there is no pillow lava under the island. This
implies that the initial submarine phase of the Surtsey
eruption was most likely explosive. Thus, the depth of
130 m, a confining pressure of 13 bars, was not suffi-
cient to prevent fragmentation of the magma. This is
not surprising since indications from elsewhere sug-
gest that considerably greater water pressure is re-
quired to prevent magma fragmentation. In the case of
the 1996 Gjálp eruption in Vatnajökull, fragmentation
into pyroclastic glass was the dominant style of activi-
ty despite the initial ice thickness of 500-750 m (Gud-
mundsson et al., 1997). Moreover, the magnetization
and density of subglacial ridges in the Grímsvötn area,
which is covered by the Vatnajökull ice cap, is such
that pillow lava cannot make up a significant propor-
tion of their volume. This suggests that explosive sub-
glacial volcanism has been the dominant form of ac-
tivity despite initial confining pressures of some 40
bars (Gudmundsson and Milsom, 1997).
Our models of Surtsey are very different from
those presented by Cameron et al. (1992) despite
being based on the same gravity data. We suggest that
our models more closely match the real geologic
structure of the island. They further demonstrate the
importance of considering variations in density of the
edifice itself when interpreting gravity data from vol-
canoes. The non-uniqueness of the sources of poten-
tial fields means that any number of models can be
fitted to the gravity data. However, constraints pro-
vided by geology or other geophysical data allow im-
plausible models to be rejected.
ACKNOWLEDGMENTS
We thank Gillian R. Foulger for giving us access to
the data collected by B.R. Cameron and F.S. Cranmer
for their B.Sc. dissertation projects at the University of
Durham. S. Jakobsson is thanked for useful discussions
and information. Reviews by Gillian Foulger and Emi-
lie Hooft improved the quality of the paper. Þórdís
Högnadóttir helped us preparing the figures.
REFERENCES
Axelsson, G, V. Stefánsson, G. Guðmundsson, and B. Stein-
grímsson. 1982. Thermal condition of Surtsey. Surtsey Re-
search Progress Report, IX, 102-110.
Bemmelen, R.W., and M. G. Rutten. 1955. The Tablemountains
ofNorthern Iceland. Brill, Leiden. 255 pp.
Cameron, B.R., F. S. Cranmer, and G. R. Foulger. 1992. Shal-
low structures beneath Heimaey and Surtsey from local
gravity data. Surtsey Research Progress Report, X, 79-93.
Einarsson, P., F. Sigmundsson, and I. T. Magnússon. 1994. A
kinematic GPS-survey in Surtsey. RH-23-94,11 pp.
Einarsson, Þ. 1965. Gosið í Surtsey í máli og myndum. Heim-
skringla, Reykjavrk. 53 pp.
Einarsson, Þ. 1968. Jarðfræði. Mál og Menning, Reykjavík,
335 pp.
Gudmundsson, M.T., and J. Milsom. 1997. Gravity and magnet-
ic studies of the subglacial Grímsvötn volcano, Iceland. Im-
plications for cmstal and thermal stmcture. Joumal ofGeo-
JÖKULL, No. 47, 1999
95