Geology of Paleo and Modern Rockfall in Purau Valley with a Special Emphasis on Understanding
Size Distribution
12 June, 2015
Henry Lanman1, Josh Borella2
1Department
of Geological Sciences, Whitman College
2Department
of Geological Sciences, University of Canterbury
Abstract
Banks Peninsula is a volcanic complex that is composed primarily of basaltic rocks and has had
rockfall occur along its flanks since emplacement. In the 2010/2011 Christchurch earthquakes,
boulders were dislodged from various cliffs and caused damage to settlements in the Port Hills of
Banks Peninsula. This study looks at the size distribution of both modern boulders from the
2010/2011 Christchurch earthquakes and “paleo” rockfall from previous seismic/rockfall triggering
events in Purau Valley near Lyttleton Harbor. It was found that the volcanic conglomerate, VB, of the
Lyttleton Volcanic Group produced the largest boulders by volume for both modern and paleo
rockfall. Cliff faces composed primarily of this volcanic conglomerate are more likely to produce larger
volume boulders than a cliff of a similar surface area composed of trachyte/hawaiite, FB, from the
Lyttleton Volcanic Group.
Introduction
As seen in the small town of Rapaki in Lyttleton Harbor on the South Island of New Zealand,
rockfall caused by seismic shaking in the 2010-2011 Christchurch earthquakes destroyed homes,
property, and led to evacuations. Across the harbor above Purau, another small establishment, there
was rockfall during the same earthquakes that also dislodged rocks from the cliffs above. The modern
rockfall caused by these earthquakes was not the first in the area, as the slopes are covered in
boulders ranging from under a half meter in diameter to boulders eighteen meters in diameter.
Source rock for rock fall must be fractured and weathered in order to be dislodged, and
dislodging rocks can be due to freeze-thaw processes, dirt-cracking, or seismic activity. The extent to
which the source rock has weathered both on faces and joints, along with alteration, shows a positive
correlation with the susceptibility to rockfall according to Wieczorek (2002). The longer rocks are
exposed allows for more time to break down and weather minerals along planes of weakness, which
in turn leads to rockfall when triggering events occur. Ronald Dorn (2014), Wieczorek, and Luuk
Dorren (2003) propose multiple triggers for rockfall. The first and most important for the Christchurch
area is seismic activity, where extensive shaking can weaken the source rock, create and widen
fractures, and dislodge boulders from a face or slope. Dorn and Wiezorek propose another method,
where large storms with large precipitation rates and high winds can dislodge and weaken rocks. Dorn
finally proposes two more methods, where the back-wasting of a valley due to colluvial and alluvial
processes will eventually undercut cliffs to produce rockfall. Also, Dorn discusses a type of weathering
called dirt cracking, where caliche and dirt placed in fractures can slowly pry a rock apart. It is known
on Banks Peninsula that seismic activity has triggered rockfall, but due to the precipitation on Banks,
along with the erosion happening within the peninsula, there is reason to believe all these processes
could be working together to produce rockfall.
As seen in the 2010/2011 earthquakes, seismic activity was the main source of the rockfall in
the area, and is thus important to understanding most of the rockfall in the past. A study done by
Benjamin H. Mackey and Mark C. Quigley (2014) showed that in un-fractured rock, a peak
ground velocity (PGV) of ≥ 25–30 cm/s from a seismic event is needed to break and dislodge
rocks from a cliff face in the Port Hills. However, if the rock was previously fractured, only a
PGV of 12 ± 1 cm/s was needed to cause rockfall. Combining this with Dorren's and Dorn's
conclusions, it can be assumed that a small amount of shaking may not cause rockfall, but
may allow for rockfall in the near future due to other processes.
After assessing susceptibility of the host rock for rockfall, and understanding what causes
rockfall, it is important to look at the size distribution of rockfall. To do so, a look into rockfall
mechanics is needed. A study by Tanarro and Muñoz (2011) which looked at rockfall in folded
limestone identified three ways in which rocks move when dislodged. The rocks can either fall,
bounce, or roll. Elaborating on these, Dorren (2003) shows that rock fall occurs on slopes steeper than
76 degrees. Falling rocks undergo both translation and rotation while in the air due to air resistance
and collisions, which both change trajectory. When the slope decreases in steepness, movement at or
near the slope surface, bouncing occurs. After the first bounce of a rock, 75-86% of the rocks initial
energy has been lost. As the slope decreases even more, rolling starts. Due to friction, the rock will
stop movement soon after in the slope does not change. This is important in the Purau Valley, as the
rocks in the valley show a high susceptibility to fracture upon impact and when they start to roll.
When looking specifically at the Rapaki area of the Port Hills, according to Borella (2015), larger
boulders that predated the 2010/2011 earthquake did not travel as far as modern boulders from the
2010/2011 earthquakes. Boulder size increased further away from the cliff in modern boulders, and
large paleo-boulders were found closer to the cliff. When looking at a study done in the Port Hills by
Massey et. al. (2013), similar conclusions can be reached. The larger modern-boulders made it the
furthest. Dorren (2003) agrees with this conclusion, and suggests that this is because larger rocks have
a higher kinetic energy and are not stopped by small topographic changes, trees, or other rocks.
Similarly, a study done by Corona et. al. (2013) showed that interaction between boulders and trees
changed the trajectory of rocks significantly, showing shorter run-out distances for rocks in densely
forested areas. In addition, Massey et. al. (2013) showed that these boulders usually accumulate in
topographic lows and valleys.
Combining the previous work done on rockfall not only in the area but around the world, a size
distribution of rockfall analysis has been performed in Purau Valley. The conclusion of this study can
be used to predict future rockfall, and possibly be used to develop a classification system for different
types of volcanic host rock that could produce rockfall on Banks Peninsula.
Setting
Banks Peninsula is connected to the South Island of New Zealand by the edge of the
Canterbury plains, and the source for the volcanism is relatively unknown. The volcanic complex is
composed primarily of two separate composite cones, with various dykes and scoria cones
throughout (Sewell 2011). The volcano was active from approximately 11.7 Ma, to 7.3 Ma when the
last part of the peninsula was formed. The volcanic rocks vary from trachyte to hawaiite, and local
variations can be seen throughout the peninsula. Purau valley is composed of rocks from the Lyttleton
Volcanic Group, the oldest of the Banks Peninsula rocks (Sewell 2011). In the study area, the host
rocks is composed primarily of “weakly bedded, yellow-brown, poorly sorted matrix to clastsupported polymict volcanic conglomerate” (Sewell 2011) with interbedded trachyte hawaiite flows
and dykes. The flanks below the cliffs are composed of rockfall and talus, with large quantities of loess
from the last glacial maximum. These wind-blown sediments have tunnel gully erosion in them
overlying bedrock. Many mass-wasting scarps can be seen along the interfluvs, with large debris flow
deposits in gullies and alluvial fans near the shore line.
Banks peninsula is located on the coast, with many bays and inlets that cut into the complex.
The weather consists of high rainfall, peaking in the winter, with moderate temperatures throughout
the summer and winter months (Ryan 1987). Precipitation can hover over the peninsula, staying
above and away from the clear plains below. Much like the rest of New Zealand, Banks is effected by
mid-latitude westerlies that vary between depressions, low pressure-rainy systems, and subtropical
anticyclones, high pressure systems with good weather (Ryan 1987). Occasional cyclone storms occur
on the peninsula, affecting much of New Zealand as well. Before European settlement in New
Zealand, much of Banks Peninsula was covered in dense forest, and now, tussock grasses and shrubs
dominate the hill slopes. Cattle and sheep actively graze Purau valley as well, reducing the size of the
foliage significantly.
Methods
Parameters were set at Purau valley, including two main valleys with various interfluv ridges
spaced throughout as the valleys branched towards the cliffs. To understand the control of the valleys
and ridges, a large study area with multiple ridges and valleys was chosen to evaluate the distribution
of rockfall. Boulders over 1 m3 were analyzed by taking length, width, and height measurements. The
rock type was evaluated, classifying boulders as either finely crystalline basalt (FB), or brecciated
basalt (VB). Surface roughness was estimated on a scale of 1-6, 6 being very high surface roughness
and 1 being little to no surface roughness. The quantity of sediment/burial of the boulders was noted,
and the overall lichen cover of the boulder was estimated. Taking into account the sediment wedges
behind boulders, their lichen cover, and their roughness, the boulders were classified as either
modern boulders or paleo boulders. Modern boulders refer to boulders from the 2011 Christchurch
earthquakes and are classified as boulders with little to no lichen cover, no sediment wedges, and low
roughness. Paleo boulders refer to boulders prior to these earthquakes and are classified by large
amounts of lichen cover, lots of surface roughness, and sediment wedges. Dating methods for
rockfall, such as lichonemetry by Mackey and Quigley (2014) and radiocarbon dating by Dorn (2014),
Borella (2015), and Curry (2003) have been used elsewhere, yet due to time limitations and monetary
constraints, a qualitative dating system was used at Purau.
GPS coordinated were taken at each location. This data was imported into Microsoft Excel,
which was used to analyze and produce results. The boulders were sorted into two main categories,
modern or paleo, and then subdivided into either VB or FB. The source rock for the boulders was
analyzed, looking at fracture spacing, failure scarps, and lithology.
Results
Modern vs. Paleo Boulder Histogram:
Figure 1 Plot showing the distribution of modern and paleo boulders with respect to volume.
Modern and paleo boulders, omitting lithology, were plotted in a histogram and fitted with a
trend line. The paleo trend-line has a steeper slope, while the modern trend-line is slightly less steep.
The overall sample size is not the same, as there are 789 paleo boulders and 125 modern boulders.
The bin size for volume was set at 0.5 m3, increasing throughout the entire data set. The largest paleo
boulder was 616m3, and the largest modern boulder was 79.97 m3.
Paleo Boulder Histogram:
Figure 2 Plot showing the distribution of VB and FB Paleo boulders with respect to volume. Large volumes omitted to show
trend.
When looking at the lithologies in the paleo boulders, 34% of the boulder were FB boulders.
There were 235 FB boulders and 554 VB boulders. The bin size was set at 0.25 m3. Size distribution for
FB and VB boulders is similar. However, since there is about half as many FB boulders as VB, the
frequency of each FB boulder size is lower than the VB. In addition, the FB boulder frequency tapers
off quickly around 4 m3, and the VB boulder frequency decreases at a slower rate. VB rocks also
appear in much large volumes. The largest FB boulder is 6.885 m3, while the largest VB boulder 616
m 3.
Paleo Boulder Frequency Distribution:
Figure 3 Distribution plot showing the cumulative frequency of both paleo volcanic breccia and paleo fine-grained basalt
boulders. Large volumes omitted to show trend.
Boulder
Lithology
1st Quartile
2nd Quartile
3rd Quartile
Mean
Median
FB
1.13 m3
1.68 m3
2.14 m3
1.74 m3
1.74 m3
VB
1.4 m3
2.04 m3
4.9 m3
2.06 m3
5.74 m3
Table 1 Quartiles, means, and medians for paleo VB and FB boulders.
The frequency distribution shows that first quartile of boulders for both VB and FB are close to
the same, being under 1.5 m3. However, the frequency distribution of FB and VB boulders only
diverges as volume increases. The second quartile for FB is 1.68 m3 for FB and 2.04 m3 for VB. The
third quartile of FB is 2.18m3 and 7.8m3 for VB. The right skewed distribution mean gives a nonrepresentative idea of boulder size, so the median is more accurate.
Modern Boulder Histogram:
Figure 4 Plot showing VB/FB modern boulder distribution. Large volumes omitted to show trend.
When looking at modern boulders, there were only 8 FB boulders over 1 m 3 and 117 VB
boulders over 1 m3 within our mapping area. The FB boulders showed no real trend in distribution.
The VB boulders show a rough right skewed normal distribution. The right skewed distribution mean
gives a non-representative idea of boulder size, so the median is more accurate. The bin size is set at
0.25 m3.
Modern Boulder Frequency Distribution:
Figure 5 Distribution plot showing the cumulative frequency of both modern volcanic breccia and modern fine-grained
basalt boulders. Large volumes omitted to show trend.
Boulder
Lithology
1st Quartile
2nd Quartile
3rd Quartile
Mean
Median
FB
1.2 m3
1.56 m3
2.16 m3
1.93 m3
1.56 m3
VB
1.69 m3
3.17 m3
7.65 m3
11.11 m3
3.17 m3
Table 2 Quartiles, means, and medians for modern VB and FB boulders.
The frequency distribution plot shows that there are a higher frequency of relatively larger VB
boulders compared to the FB boulders. The first quartile is relatively close between the two
lithologies, VB being 1.4 m3 and the FB being 1.13 m3. The quartiles start to diverge, with the second
being 1.68 for FB and 2.04 for VB, and the third quartile for VB being 4.90 and FB at 2.14 m 3.
Host Rock Analysis:
The lavas in Purau valley primarily range in dips from the east to south, with contacts varying
locally in orientation. In the valley, there is significantly more volcanic breccia than fine-grained
interbedded basalt. Cliffs of VB vary from 2-15m in thickness, while the FB layers vary between 0.5-5
m in thickness. Fracture spacing in the VB is very random, usually leaving concave surfaces where
rocks have fallen off. Fractures spacing in the FB ranges from cracks spaced about 1 cm apart from
each other and larger 1m fractures. The smaller fractures are very regular, and their orientation varies
greatly throughout the FB units. Modern FB rockfall is heavily fractures, usually with lots of smaller
pieces of rock all around larger pieces.
Volcanic Breccia (VB)
Fine-grained basalt (FB)
Figure 6 Photographs comparing the two lithologies studied at Purau Valley. Note the fracture spacing in the FB and the
random fracture faces in the VB.
Interpretation
Comparing the size distribution between modern and paleo boulders (Figure 1) shows that
there is a very similar trend, with paleo boulders simply having a higher number of boulders. There
were not any modern boulders that matched the largest paleo boulder in volume, yet there was one
that was around 80 m3. For one rockfall event, this is quite a large boulder and is definitely an outlier.
Because paleo boulders are defined by rockfall events prior to the 2010/2011 earthquakes, it makes
sense that there is a wide range in volume of paleo boulders. There has simply been more time for
the boulders to fall, thus allowing for the less frequent larger volume boulders to fall. However, while
seeing the quantity of rockfall created by the last seismic event in Purau Valley, I do not believe that is
the sole source of rockfall. According to Dorn (2014) and Wiezorek (2002), large storms can also
dislodge rocks. While not a significant source of rockfall, large quantities of rain with occasional large
3
storms could create rockfall in the valley. Conversely, He dating of rockfall in the Port Hills by
Mackey and Quigley (2014) showed that boulders grouped around 7,000 years old 13,000 years old.
This suggests that large seismic events do create a significant amount of rockfall. There were outlier
boulders that were not dated in this range, possibly meaning total rockfall cannot be attributed solely
to seismic shaking.
When looking at the distribution of paleo boulder ilithologies (Figure 2), both the VB and FB
boulders show similar size distributions. There is a a little over three times as many VB boulders than
FB, so the shear number of boulders under each bin in the VB/FB paleo histogram is lower. Also, the
FB boulder size tapers off more quickly in the histogram, while the VB boulders show a higher
numbers of boulders in higher volume. A similar trend is apparent in the frequency distribution of FB
and VB boulders (Figure 3), where 95% of the FB boulders fall under 3 m3. The VB boulders curve is
more shallow, showing that more boulders occur in higher volumes. About 95% of boulders fall under
35 m3.. When looking at the properties of the source rock, it makes sense that the FB boulders are
lower in volume and frequency. The cliffs had an estimated 4:1 ratio of VB to FB, and the FB appeared
to have a high fracture density compared to the VB. FB rocks, as seen in modern rockfall, were more
prone to break upon impact and create talus slopes. The VB boulders however, remained fairly intact
as they traveled down slope.
A similar trend was found in the modern boulders. While we sampled 125 modern boulders,
only eight FB boulders and 117 VB boulders were over 1 m3.(Figure 5). Within the study area, usually
occurring near measured modern boulders, there were over 200 modern boulders that were under 1
m3 that did not get included in this study. The lack of large numbers of FB boulders made a trend
within the histogram for the FB and VB modern boulders (Figure 4) quite ambiguous, yet the
frequency distribution still showed a similar trend to that of the paleo boulders for both FB and VB. All
the modern FB boulders were under 2 m3, suggesting that the host rock simply was not capable of
producing large volumes of FB rock. This was backed up by the field observations discussed earlier,
where FB modern rockfall was broken up in talus slopes that matched the fracture patterns seen in
the cliff faces above. The VB modern rockfall, however, produced numerous quite large boulders from
obvious scarps in the cliff. The frequency distribution shows that 95% of the boulders are under 20
m3, and the median boulder volume is 3.17. This is significantly higher than the modern FB which has
all boulders under 2 m3, as well as having more boulders than the modern FB.
After seeing the fragile characteristics of the modern FB boulders, it is interesting how there
are so many large FB paleo boulders. There could be a few reasons for this. When a FB boulder falls, it
falls close to its source. With relatively low impact, and with an almost immediate stop in motion, the
rock will not have time to break apart as it careens down slope. This boulder is then transported,
either by debris flows or creep further down slope to a new location. Additionally, the existence of a
paleo forest could not only prevent these boulders from traveling further down slope, but also shelter
the boulder from sunlight, water, and other agents that could promote weathering and breakdown. If
there still was forest cover in Purau Valley, there may be a similar distribution of FB and VB rocks.
Conclusion
The large numbers of VB boulders relative to the number of FB boulders means that, under the
specific circumstances at Purau Valley, cliff faces composed primarily of volcanic breccia will produce
more rockfall in higher volumes than an area with massive trachyte and hawaiite lava flows in future
rockfall events. The hawaiite/trachyte volcanic conglomerate is part of the Lyttleton Volcanic Group,
the oldest of the Banks Peninsula volcanics, and Purau Valley serves as a perfect proxy to understand
how this lithology will react under rockfall triggering events.
Similarly, Purau Valley also serves as an excellent proxy to understand how finely bedded
trachyte and hawaiite lavas, FB, respond to rockfall triggering events. The FB lithology, due to its
fragile nature, will break apart upon falling and requires a large quantity of rock from the cliff face to
produce a boulder over 1 m3.
Combining these two conclusions, it can be assumed that the lithology of the host rock plays a
very important roll in size distribution of rockfall. Different rock characteristics, such as composition
and hardness, will lead the rock to break in bigger or smaller pieces and will also affect the distance a
rock gets before breaking into smaller pieces.
Future Work
Purau Valley serves as an excellent proxy for two lithologies in the Lyttleton Volcanics Group in
Lyttleton Harbor. As there are a number of communities within the harbor, understanding the
susceptibility of the cliffs above these establishments to rockfall would be essential to protect lives
and capital in the next seismic event in the Christchurch area. To continue this study, identifying areas
that could further serve as proxies for the wide variety of Lyttleton Volcanic Group lithologies in
regards to rockfall could be done. With this in mind, a statistical analysis of the lithology and the
rockfall below would be done at each location. With most lithologies assessed, the table could be
used to analyze risk in populated areas where paleo and modern rockfall have been removed or
relocated. Understanding the size relationship for each lithology, and applying geophysics to estimate
run-out distance, proper reinforcements like rock bunkers or screens could be installed near
settlements to protect them from future events.
References
Azzoni, A., La Barbera, G., & Zaninetti, A. (1995). Analysis and prediction of rockfalls using a
mathematical model. International Journal of Rock Mechanics and Mining Sciences &
Geomechanics Abstracts, 32(7), 709–724. http://doi.org/10.1016/0148-9062(95)00018-C
Borella, Josh. (2015). Verbal Communication.
Bull, W. B. (2014). Using earthquakes to assess lichen growth rates. Geografiska Annaler: Series A,
Physical Geography, 96(2), 117–133. Retrieved from
http://onlinelibrary.wiley.com.ezproxy.canterbury.ac.nz/doi/10.1111/geoa.12035/full
Bull, W. B., & Brandon, B. T. (1998). Lichen dating of earthquake-generated regional rockfall events,
Southern Alps, New Zealand. Geological Society of America Bulletin, 60–84. Retrieved from
http://bulletin.geoscienceworld.org.ezproxy.canterbury.ac.nz/content/110/1/60.full.pdf+html?si
d=e3c0b7ad-c945-4703-8565-90abf0fc1b08
Corona, C., Trappmann, D., & Stoffel, M. (2013). Parameterization of rockfall source areas and
magnitudes with ecological recorders: When disturbances in trees serve the calibration and
validation of simulation runs. Geomorphology, 202, 33–42.
http://doi.org/10.1016/j.geomorph.2013.02.001
Curry, A. M., & Black, R. (2003). Structure, sedimentology and evolution of rockfall talus, Mynydd Du,
south Wales. Proceedings of the Geologists’ Association, 114(1), 49–64.
http://doi.org/10.1016/S0016-7878(03)80027-5
Dorn, R. I. (2014). Chronology of rock falls and slides in a desert mountain range: Case study from the
Sonoran Desert in south-central Arizona. Geomorphology, 223, 81–89.
http://doi.org/10.1016/j.geomorph.2014.07.005
Dorren, L. K. A. (2003). A review of rockfall mechanics and modelling approaches. Progress in Physical
Geography.
Mackey, B. H., & Quigley, M. (2013). Cosmogenic Exposure Dating of Paleo-Rockfall Deposits,
Christchurch, New Zealand. AGU Fall Meeting Abstracts, -1, 0731. Retrieved from
http://adsabs.harvard.edu/abs/2013AGUFMEP53A0731M
Massey, C. I., McSaveney, M. J., Taig, T., Richards, L., Litchfield, N. J., Rhoades, D. A., … Van Dissen, R.
J. (2014). Determining Rockfall Risk in Christchurch Using Rockfalls Triggered by the 2010–2011
Canterbury Earthquake Sequence. Earthquake Spectra, 30(1), 155–181.
http://doi.org/10.1193/021413EQS026M
Pfeiffer, T., & Bowen, T. (1989). Computer simulation of rockfalls. Bulletin of the Association of
Engineering Geologists, 26(1), 135–146. Retrieved from
http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Computer+simulation+of+rock
falls#0
Ryan, A. P. (1987). The climate and weather of Canterbury, including Aorangi. New Zealand
Meteorological Service. Retrieved from https://library.niwa.co.nz/cgi-bin/koha/opacdetail.pl?biblionumber=140288&query_desc=se%3AMiscellaneous publication %2F New Zealand
Meteorological Service
Sewell, R. J. (1988). Late Miocene volcanic stratigraphy of central Banks Peninsula, Canterbury, New
Zealand. New Zealand Journal of Geology and Geophysics, 31(1), 41–64.
http://doi.org/10.1080/00288306.1988.10417809
Tanarro, L. M., & Muñoz, J. (2012). Rockfalls in the Duratón canyon, central Spain: Inventory and
statistical analysis. Geomorphology, 169-170, 17–29.
http://doi.org/10.1016/j.geomorph.2012.01.003
Wieczorek, G. F. (2002). Catastrophic rockfalls and rockslides in the Sierra Nevada, USA. In S. G. Evans
& J. V DeGraff (Eds.), Catastrophic landslides: Effects, occurrence, and mechanisms (Vol. XV, pp.
1–26). Geological Society of America Reviews in Engineering Geology.