The average eruption rate has increased from 0.012 per year in the earlier analysis to 0.016 (Turner et al. 2008a) as a consequence of the additional events from the Rotokare core. The more subtle, although very important, change in the mixture-of-Weibulls model is the result of the additional events filling in some of the gaps of the original single site Umutekai record, `sharpening’ the secondary mode, which now represents only 4% of the recurrence distribution. As a result, the 50 year probability has increased from 0.37 to 0.52 (or almost doubling the present annual eruption probability (AEP) from 0.9% to 1.6%).In general, filling in the record with additional events increases the AEP, and also appears to smooth out the peak and trough in probability following the last eruption (cf., Fig. 3.11). If the remaining long intervals corresponding to the second mode were to be filled in, then the parameter (p) would tend to 1, and the best-fit model would then be the simple unimodal Weibull distribution shown as the dashed line in Figure 3.11. However, within the 6250 B.P. – 10 150 years B.P. Umutekai record there are 2 distinct periods where there are apparently no events: 6500-7000 years B.P. and 8200-8700 years B.P. Moreover, these periods of decreased activity, and the long inter- event times in the combined record post 6250 years B.P. occur fairly regularly on 1500- 2000 year cycles (Fig. 3.9). This regularity may correspond to long term cycling in magma system replenishment relating to either (1) the recurrence of regional tectonic events affecting the magma system (e.g., triggering magma batch rise), or (2) where large magma batches reach minimum buoyancy to rise and erupt (cf., Turner et al. 2008b).
The combination of all three datasets provides a more robust eruption record for this volcano (i.e., Figs. 3.10 and 3.11). However, the combined dataset is still incomplete and therefore likely under-represents the true eruption recurrence from Mt. Taranaki. There are 31 matches based on the age-correlated technique (Tables 3.3 and 3.4), of which 15 were able to be compared by titanomagnetite compositions. The titanomagnetite compositions of the tephras near the base of the Rotokare record indicate that the correlation of these tephras (R40, 41, and 42) to those within the Umutekai core (U60-U65) by their radiocarbon ages is slightly incorrect. Each correlation was out by a consistent factor (i.e., a distance of 1.2-2; Table 3.3), possibly
suggesting a systematic bias in this part of the record in the depth-age profile of one or both of the cores. In addition, some of the age-matched tephras had distinctively different geochemistry to one another, suggesting that they were derived from compositionally different magmas. This indicates that some of the 16 matches that could not be checked by the titanomagnetite geochemically may also represent distinct events, and therefore there are additional eruptions which need to be accounted for in the eruption hazard model. However, it is also possible that a single event from Mt. Taranaki involved the eruption of more than one distinct magma batch (for example a basaltic andesite and dacitic magma) within the defined time period of an event (cf., Simkin and Siebert 1994).
The eruption of more than one magma batch during a relatively short period may be a common process at andesitic volcanoes. This is especially true at Mt. Taranaki where, since ~3000 years B.P. there have been eruptions of different compositionally distinct magma batches from the summit vent and the more basaltic satellite cone of Fanthams Peak (Neall et al. 1986). For example, U10 is a basaltic andesite tephra likely to have been sourced from Fanthams Peak. It partially overlaps with the similarly dated R15 on the geochemical PC diagram. Under binocular microscope, two distinct grain types can be identified within the tephra deposit of R15; finer grained (<500 µm) scoria clasts, likely to be a correlative to U10; and relatively larger (>500 µm) lighter-coloured pumice clasts not observed within the U10 tephra. This suggests that Tephra R15 likely records two closely spaced events from Mt. Taranaki, which given the sedimentation rate for Lake Rotokare, must have been within a year or two of each other because any longer period would result in >1mm of lacustrine sediment being deposited. Therefore, at approximately 2540 years B.P., ash from an eruption of Fanthams Peak was distributed to the NE and deposited in Lake Umutekai whereas shortly after (or before) an eruption producing andesitic pumiceous material distributed ash towards Lake Rotokare to the SE of the volcano.
More significantly, this combined record is almost certainly still missing some events altogether. For example, of 64 events in the period 2250 – 6250 years B.P., 33 (52%) appear only in the Umutekai record, 10 (16%) in only the Rotokare record, and the remainder in both. This indicates that the 138 events in the record could well be further
augmented by additional sites. If 84% from the 2250 – 6250 years B.P. record is taken as the upper bound for the Umutekai “detection probability”, it can be expected that there should be at least 8 missing events within the Umutekai record 6250 – 10 150 years B.P.
During the studied period there are also a number of variations in the eruption frequency curve (cf., changes in the slope of Fig 3.8). Between 96 and ~500 years B.P. the frequency of identified eruptions increases dramatically, in contrast to the relatively lower eruption frequency just prior to it (500-2200 years B.P.). The <500 year record of events is mainly from many smaller, well dated, dominantly block-and-ash-flow-related events (cf., Turner et al. 2008a). Later events (>500 years B.P.) are either unidentified (i.e., buried by the younger events) or un-datable within the older edifice record. In addition to this, block-and-ash-flow related events have comparatively small co-eruptive ash plumes and are therefore likely to be underrepresented within the medial-distal lake records. In support of this, Turner et al. (2008b) identified periods when block-and-ash- flow events identified within Lake Umutekai are more frequent. These increased periods of activity correlate directly to a period when block-and-ash-flows were being directed down the NE oriented channels towards Umutekai (Turner et al. 2008b). The young record between 500 and 1500 years B.P. is from the edifice dataset and Rotokare datasets only. The relatively rapid sedimentation rate of Lake Rotokare provides a less- ideal environment compared to the very slow sedimentation of Lake Umutekai, to preserve small (sub-mm) tephras within a single identifiable layer. In addition, the topographic high near the township of Eltham (Fig. 3.1) means that any pyroclastic flows would have been diverted away 8 km to the west of the lake. Therefore, block-and ash-flows associated thin and fine ash deposits are unlikely to have been recorded as a discrete layers within Rotokare sediments. In comparison, Lake Umutekai is situated just over a kilometre away from a dominant channel (Fig. 3.1). The highly variable nature of block-and-ash-flow events recorded within our dataset also gives evidence that although this is the most robust record yet from Mt. Taranaki, it appears that it substantially under estimates the frequency and reoccurrence of block-and ash-flow events. Similarly, dominantly effusive or lava-flow producing eruptions may also have occurred during this period (cf., Neall et al. 1986) with no tephra produced at all.
Further work to improve the homogeneity in the data is very desirable in order to aid in regime identification in the behaviour of the volcano (Bebbington 2007). In the interim, future eruption reoccurrence modelling, where a complete record is unavailable (due to the lack of well-dated records) could use the identified explosive event record (i.e., excluding the highly variable block-and ash-flow record) and then scaling up the intervening block-and-ash-flow record from parts of the record that are well dated.