Planetary Volcanism 847
within it chills and comes to rest as its viscosity becomes
extremely high, or the pressure within the reservoir falls to
the point where there is no longer a great enough stress at
the dike tip for rock fracturing to continue.
Under certain circumstances, an unusually large volume
of magma may be removed from a shallow reservoir, re-
ducing the internal pressure beyond the point where the
reservoir walls behave elastically. Collapse of the overlying
rocks may then occur to fill the potential void left by the
magma, and a caldera (or, on a smaller scale, a pit crater)
will form. The circumstances causing large-volume erup-
tions on Earth include the rapid eruption to the surface
immediately above the reservoir of large volumes of low-
density, gas-rich silicic (rhyolitic) magma, and the drainage
of magma through extensive lateral dike systems extending
along rift zones to distant flank eruption sites on basaltic
volcanoes. This latter process appears to have been associ-
ated with caldera formation on Kilauea volcano in Hawaii,
and it is tempting to speculate that the very large calderas
on some of the martian basaltic shield volcanoes (especially
Pavonis Mons and Arsia Mons) are directly associated with
the large-volume eruptions seen on the distal parts of their
rift zones. In contrast, we saw earlier that, at the martian
volcano Hecates Tholus, a large explosive summit erup-
tion is implicated in the formation of at least one of its
calderas.
The size of a caldera must be related to the volume of the
underlying magma reservoir, or more exactly to the volume
of magma removed from it in the caldera-forming event. If
the reservoir is shallow enough, the diameter of the caldera
is probably similar to that of the reservoir. Diameters from
1 to 3 km are common on basaltic volcanoes on Earth and
on Venus, with depths up to a few hundred meters implying
magma volumes less than about 10 km^3. In contrast, caldera
diameters up to at least 30 km occur on several volcanoes
on Mars and, coupled with caldera depths up to 3 km, imply
volumes ranging up to as much as 10,000 km^3. The stresses
implied by the patterns of fractures on the floors and near
the edges of some of these martian calderas suggest that the
reservoirs beneath them are centered on depths on the or-
der of 10–15 km, about three to four times greater than the
known depths to the centers of shallow basaltic reservoirs
on Earth. The simplest models of the internal structures
of volcanoes suggest that, due to the progressive closing of
gas cavities in rocks as the pressure increases, the density of
the rocks forming a volcanic edifice should increase, at first
quickly and then more slowly, with depth. Rising magma
from deep partial melt zones may stall when its density is
similar to that of the rocks around it so that it is neither pos-
itively nor negatively buoyant, and a reservoir may develop
in this way. Because the pressure at a given depth inside a
volcano is proportional to the acceleration due to gravity,
and because martian gravity is about three times less than
that on Earth or Venus, the finding that martian magma
reservoirs are centered three to four times deeper than on
Earth is not surprising. However, these simple models do
not address the reason for the martian calderas being much
more than three times wider than any of those on Earth or
many of those on Venus. On Io, we see some caldera-like
structures, not necessarily associated with obvious volcanic
edifices, that are even wider (but not deeper) than those on
Mars, though we have too little information about the in-
ternal structure of Io’s crust to interpret this observation
unambiguously. Much is still not understood about the for-
mation and stability of shallow magma bodies.
Evidence for significant shallow magma storage is con-
spicuously absent from the Moon. The large volumes ob-
served for the great majority of eruptions in the later part of
lunar volcanic history, and the high effusion rates inferred
for them, imply that almost all of the eruptions took place
directly from large bodies of magma stored at very great
depth—at least at the base of the crust and possibly in par-
tial melting zones in the lunar mantle. Not all the dikes
propagating up from these depths will have reached the
surface, however, and some shallow dike intrusions almost
certainly exist. Recent work suggests that many of the lin-
ear rilles on the Moon represent the surface deformation
resulting from the emplacement of such dikes, having thick-
nesses of at least 100 m, horizontal and vertical extents of
∼100 km, and tops extending to within 1 or 2 km of the sur-
face. Minor volcanic activity associated with some of these
features would then be the result of gas loss and small-scale
magma redistribution as the main body of the dike cooled.
The emplacement of very large dike systems extending
most or all of the way from mantle magma source zones
to the surface is not confined to the Moon. It has long
been assumed that such structures must have existed to
feed the high-volume basaltic lava flow sequences called
flood basalts that occur on Earth every few tens of mil-
lions of years. These kinds of feature are probably closely
related to the systems of giant dikes, tens to hundreds of
meters wide and traceable laterally for many hundreds to
more than 1000 km, that are found exposed in very ancient
rocks on the Earth. The radial patterns of these ancient dike
swarms suggest that they are associated with major areas of
mantle upwelling and partial melting, with magma migrat-
ing vertically above the mantle plume to depths of a few
tens of kilometers and then traveling laterally to form the
longest dikes. Some of the radial surface fracture patterns
associated with the novae and coronae on Venus are al-
most certainly similar features that have been formed more
recently in that planet’s geologic history, and on Mars the
systems of lineargraben, some of which show evidence
of localized eruptive vents, extending radially from large
shield volcanoes, also bear witness to the presence of long-
lived mantle plumes generating giant dike swarms. It seems
that there may be a great deal of similarity between the
processes taking place in the mantles of all the Earth-like
planets; it is the near-surface conditions, probably strongly
influenced by the current presence of the oceans, that drive