Macro-characterisation of cave damage for palaeoseismological investigations in regions of low strain: A case study from central–western North Island (Waitomo caves), New Zealand

Abstract

Speleoseismology reconstructs palaeo-earthquake records from damage preserved in caves, such as rockfall and broken or deformed speleothems (i.e., stalagmites and stalactites), which are uranium-series dated by analysis of pre- and post-damage speleothem calcite. However, in regions of low strain and infrequent damaging earthquakes, there is potential for cave features to fail spontaneously without a triggering event, which could lead to misinterpretation of the damage record. Here we present a scheme for characterising and weighting diverse types of cave damage based on their quality of evidence for palaeoseismicity. Three levels of criteria are proposed for assessing the site-, cave- and region-specific processes likely to have influenced the stability of each feature prior to damage. We apply this scheme to a suite of 69 damage features (0.5–225 ka) sampled from caves in the seismically quiet Waitomo district, western North Island, New Zealand, which exhibit varying degrees of age clustering. Application of the criteria to each damage feature (i.e., potential earthquake evidence) resulted in an amplified signal relative to the noise, which enabled us to define a threshold above which the degree of clustering of damage ages is significant, and may be interpreted as a damage ‘event’ with more confidence. Based on the clustering of damage ages, 13–16 damage events are identified over the 225-kyr period, with three defined levels of confidence (Tier 1 damage event = low confidence to Tier 3 = moderate confidence). Of these, we identify 4–5 damage events of Tier ≥2 since 16 ka, that exceed the noise–signal threshold established for this dataset, yielding recurrence intervals for this period of 3.2–4.0 kyr (Tiers 2 and 3) or 5.3–16.0 kyr (Tier 3 only). Two additional Tier ≥2 damage events are identified at 33.9 ± 0.8 and 134.0 ± 1.3 ka. We estimate that Modified Mercalli intensities in excess of VII–VIII are required to generate the damage features documented in this study. Results of a probabilistic seismic hazard assessment indicate the likely sources of such shaking at Waitomo are the distal Hikurangi subduction margin or a shallow, proximal fault. The former is supported by the ages of the only two Tier ≥2 damage events in Waitomo in this period, which coincide with subduction earthquakes of the Hikurangi margin record at 5020–4450 and 870–815 yr BP, while the latter warrants further searching for evidence of active faulting in the Waitomo area. This study demonstrates the potential for speleoseismological investigations in New Zealand and regions of low earthquake recurrence.