Abstract:
This thesis explores whether speleoseismology, the study of earthquake records in caves, is a
viable approach to paleoseismology in New Zealand, one that would add value to the
conventional paleoseismic methods currently in use. Speleoseismic investigations were
undertaken in caves in three contrasting seismic settings using a variety of existing (physical)
and novel (geochemical) methods. Settings included a region of low seismicity (Waitomo
Caves, western North Island), high seismicity (Te Reinga Cave, Wairoa, eastern North
Island), and very high seismicity (Guillotine Cave, Springs Junction, northwestern South
Island).
Macro-scale damage features in five Waitomo caves, such as rockfall and broken
speleothems (i.e., stalagmites and stalactites), were dated by U-series methods on pre- and
post-damage speleothem calcite, and yielded ages of ~0.5 to 225 ka. These ages were
examined to assess clustering of damage ages that could be interpreted as regional
paleoseismic events. To address the possibility of spontaneous damage (i.e., not being
triggered by an environmental event) in the data set, a scheme was developed for weighting
damage features based on their quality as evidence for paleoseismicity. Application of this
scheme to the Waitomo Cave damage ages resulted in an amplified “signal” of cave damage
events relative to the “noise” of spontaneous damage, making it possible for the first time to
estimate the recurrence interval (RI) of inferred damaging earthquakes in the Waitomo area
since ~16 ka (RI = 3.2–16 kyr).
In the high to very high seismicity settings of Te Reinga and Guillotine caves, a record of
physical and geochemical growth changes was constructed for two stalagmites (samples TRA
and GT1, respectively) using chronologies based on closely spaced U-series ages. For the
Wairoa stalagmite, annual layer counts were used to further refine the chronology. Trace
element analyses by laser ablation inductively coupled plasma mass spectrometry conducted
along both stalagmite growth axes, revealed anomalous Mg enrichments that are consistent
with the hypothesis that coseismic fracturing of the host rock induces transient changes in the
chemistry of cave dripwaters through incongruent carbonate dissolution.
In stalagmite TRA from Wairoa, Mg enrichments are commonly followed by cessations in
growth. A comparison with the record of margin-correlated subduction earthquakes on the Hikurangi megathrust shows good correspondence with the ages of onset and cessation of
stalagmite growth. Seven out of eight previously documented subduction events that overlap
with the 4.6-kyr record of TRA broadly coincide with an age of cessation or onset of
stalagmite growth given the estimated uncertainties. In stalagmite GT1, Mg enrichments are
always associated with brown staining of the speleothem calcite, which together are
interpreted as the near-surface signal of host rock fracturing and enhanced transport of
organics in pore water shaken from the overlying soil during a seismic event. Magnesium
enrichments are also commonly associated with other physical growth changes (e.g., growth
onset and cessation, growth axis shifts, and changes in tip morphology). The record of growth
changes in GT1 in the last ~1.6 kyr shows good correspondence with the record of surface
rupturing events on the northern Alpine Fault. Moreover, by combining the geochemical
anomalies with physical growth changes, events of higher and lower shaking intensity can be
distinguished, based on the assemblage of growth changes comprising each growth change
event. In this regard, the GT1 record is consistent with the two intensity levels attributed to
events in the seismically triggered turbidite record of Lake Christabel (~10 km south of
Guillotine Cave).
The Wairoa stalagmite revealed evidence for temporal clustering of damaging earthquakes.
Six shaking events that were strong enough to fracture the Te Reinga Cave host rock are
inferred to have taken place in the space of ~150 yr (~0.816–0.665 ka), and the first event in
the sequence coincides (within error) with the Hikurangi subduction event at 0.815–0.870 ka.
This is in contrast to a later ~170-yr period of continuous stalagmite growth during which no
rock fracturing events were recorded (~0.412–0.240 ka). The first sequence is interpreted as a
series of upper plate faulting events in eastern North Island that followed a partial to full-margin rupture on the Hikurangi megathrust.
The fact that these geochemical signals can be recorded during periods of continuous growth
implies that with a high density U-series dating programme, high-fidelity paleoearthquake
chronologies can be constructed from speleothems in New Zealand and worldwide, provided
that the speleothem has high U, low detrital Th, and that the host rock composition is such
that the geochemical signal of incongruent dissolution can be distinguished from that of prior
calcite precipitation, as demonstrated in the Wairoa stalagmite. This thesis research
demonstrates that speleoseismology can be applied across the diverse seismic settings in New
Zealand (i.e., low to high seismicity), and that speleoseismology can provide new insights
into New Zealand paleoseismology.