In the generation area of the 18th March 2007 lahar, seismic signals aid in the inter-
pretation of the Crater Lake dam-failure mechanisms and timing. Subtle changes in the signal intensity recorded on vertical-component, 10 Hz geophones, on and near the dam, show a sudden and rapid period of headward erosion of rills/small gullies during the rainstorm-initiated dam collapse. Over only one minute, the geophones picked up a strong signal pulse (Stage I, Figure 6.2), which probably resulted from sudden slips and rockfalls developing in the existing erosion gullies, causing them to rapidly propagate inward to the dam crest. This led to a short respite period (Stage II) where gradual draw-down of the lake occurred through a small and stable channel that was not visible on photographic records (Massey et al. (2009)), but could be seen as an eight-minute long medium-amplitude peak in the geophone signal. Sudden, rapid slumping and failure of the dam face occurred again to generate a strong energy peak on the geophone, heralding the catastrophic release of Crater Lake water (Stage III). This initial, violent release tailed off after c. 20 minutes, and was followed by a further peak in seismic energy (Stage IV), associated with additional tephra slumping, bank erosion and collapse that led to widening of the breakout breach.
It appears that throughout the dam-failure event, the geophone signals reacted most strongly to rockfall, slumping and debris collapsing events, with apparent low signal response relating to differing outflow rates from the Crater Lake. As seen at the Round-the-Mountain-Track, OnTrack Flood Gauge, and Colliers Bridge sites (Fig- ures 6.8, 6.11, and 5.4, respectively), vertical-component seismic signals in the 1-10 Hz range from broadband seismographs are the least responsive to lahar flow, with dom- inant energy magnification in cross- and parallel-channel signals. Vertical component signals appear to reflect the saltation or collision of large particles in the flow with the channel base. As described in Chapter 3, the dominant momentum of flows and vectors of particle motion are often parallel with flow; therefore, flow-parallel signals are exacerbated. Due to the U-shaped geometry of the channel, cross-channel energy is also often amplified, with collisions against any channel wall augmenting this. At the breach area, the initial flow would have been primarily water, relatively shallow
and with low sediment concentration. This water-dominated flow likely generated little vertical-component signal, except when bank-erosion, or other tephra-bank collapses, occurred.
From the initiation point to the next recording sites, 2.2 and 4.6 km downstream, evidence of the initiation events and initial spill-over flow (Stages I and II) had largely disappeared. Instead, these sites show that two major peaks in flow had formed, with a third peak developing at the more distal site (Figure 6.4). The two major peaks show a strong correlation to Stage III and IV peaks in the dam-geophone records. The relatively subtle transition to Stage IV implied by the dam-geophone record appears to have marked a major bank-erosion and widening episode, leading to a strong surge-like increase in discharge from the lake. The third apparent peak in flow, recorded weakly at 2.2 km and more strongly at 4.6 km from source, seems also to relate to a spike in a geophone signal (Geophone 3 on Figure 6.4) recorded at the source. This may reflect a landslide of up to 1 x 106 m3 of debris, with its toe entering the Whangaehu channel
at c. 2 km from source (immediately upstream of the 2.2 km geophone site). Entry of this debris appears to be caused by erosion of the toe of a pre-existing landslide by the initial part of the lahar.
Downstream, on the multi-component record at the Round-the-Mountain-Track (RTMT) site 7.4 km from source (Figure 6.6), there appears to be no record of the initial breach sequence, but a strong representation of three pulses of flow in the river-level record. The first two pulses in the river appear to reflect the geophone-signal defined Stages III and IV described above, with the third pulse representing an expansion of the seismic peak observed at the 4.6 km geophone site (possibly landslide related). This evidence confirms that the geophone signal’s Stages III and IV were related to peaks in flow discharge from the Crater Lake, which propagated through to at least 7.4 km from source.
The analysis of frequency response in relation to the passage of the lahar at the RTMT shows that after the arrival of the first flow peak (Figure 6.8d) and during the second peak flow (Figure 6.8e), the strongest amplification of signal was in the channel-parallel component. As discussed in Chapters 3 and 4, strong channel-parallel signals in the low-frequency range indicate friction between particles and the river bed, associated with sliding or rolling motion downstream, or also saltation signals with
a dominant down-channel vector, rather than random and cross-channel motion. In Chapter 4, the difference between the relatively turbulent 18th March 2007 lahar and
the plug-flow like 25th September 2007 snow-slurry lahar were emphasised. Here, by
analysing the relative signal strengths inside the 18th March lahar, it appears that
the periods indicated by Figure 6.8d and e show the strongest evidence for sliding, rolling or down-flow saltation of large particles, consistent with periods when particle concentrations were highest. This implies that the initial peak of flow may have been more dilute than these later portions, in accordance with the observations of watery peak-flow and clear-water peak tide-line erosion as reported by Procter et al.(2010).