Geography 490 Thesis, 2005
The Effect of Tsunami Hazard on the Avon-Heathcote Estuary Area
By Nick Griffiths
1
Abstract
This thesis assesses the tsunami hazard for Christchurch’s Avon-Heathcote Estuary
and surrounding area. An understanding of the physical tsunami wave dynamics is
developed, with review of literature and consideration of historic tsunami events.
Based on this knowledge, four realistic tsunami flooding scenarios of 2, 3, 4 and 5 m
above mean sea level are adopted, and the areas each will affect quantified through
utilisation of accurate data and up-to-date techniques. The total land area that would
be flooded is calculated and mapped for each scenario, and consideration is given to
the type of tsunami event which could generate each depth and extent of inundation.
2
Contents
Chapter 1: Introduction
1
1.1 Thesis Statement
1
1.2 Field Area
1
1.3 Research Rationale
6
1.4 Thesis Outline
7
Chapter 2: Review of Pertinent Tsunami Literature
8
2.1 Introduction
8
2.2 Tsunami Waves
8
2.3 Tsunami Effects in Enclosed Areas
11
2.4 Historical Tsunami Record for the Avon-Heathcote Estuary
15
2.5 Methods Used to Predict Effects of Tsunami
16
Chapter 3: Methodology
18
3.1 Data
18
3.2 Inundation Study
19
3.3 Field Observations
22
Chapter 4: Results
23
Chapter 5: Discussion
30
5.1 Tsunami Events which could generate the Flooding Scenarios
30
5.2 Flooding Implications
31
5.3 Review of Findings and Techniques
32
5.4 Potential Future Research
32
5.5 Conclusion
33
References
34
3
Chapter 1: Introduction
1.1 Thesis Statement
Tsunamis are potentially destructive waves generated by the displacement of water
resulting from earthquakes, underwater landslides, volcanic eruptions and meteorite
impacts. As was highlighted recently by the catastrophic Indian Ocean events of
Boxing Day 2004, tsunamis are an important and significant hazard. For economic
and social reasons the world’s ocean coastlines have historically attracted people to
live near them, with approximately two thirds of the world’s population living directly
landward of the ocean’s edge (Komar 1998). This trend is especially evident in New
Zealand, with the coastal environment continuing to attract large numbers of residents
and visitors. It is known that enclosed coastal areas such as bays, estuaries and
harbours can have an amplifying effect on tsunami waves, although limited detailed
knowledge exists on this aspect of the hazard.
The primary aim of this study is to develop a better understanding of tsunami
characteristics and the effects a tsunami event would have on the Avon-Heathcote
Estuary and its surrounding infrastructure. The ability to protect life and property
from any hazard is primarily dependant on (a) understanding the physical processes
involved and being able to predict the damage they will cause and (b) assessing the
options for modifying the event and/or the human system. This study will focus on (a)
and build on current knowledge and previous research, utilising up to date techniques
to quantify the areas that would likely be flooded in realistic tsunami scenarios.
The next section of this chapter describes the Holocene and recent
development of the field area, which has shaped the characteristics of the tsunami
hazard examined in this study. Section 1.3 will discuss the research rationale, and go
on to provide a content outline of subsequent chapters.
1.2 Field Area
The Avon-Heathcote Estuary is located in South Pegasus Bay on the east coast of
New Zealand’s South Island, and is surrounded on all sides by the city of
Christchurch (Figure 1.1). The estuary is located approximately 12 km from the city
4
centre and covers an area of around 880 ha (Owen 1992). It is fed by the Avon River
entering from the north, and the Heathcote River from the south-west. New Brighton
Spit all but separates the Estuary from the Pacific Ocean.
Figure 1.1: Location of the Avon-Heathcote Estuary, Christchurch, New Zealand
(Topomap).
At the height of the last major ice age, sea levels were around 135-150 m lower than
they are today on the Canterbury coast, and the shoreline was 50-60 km further east.
This major period of glaciation ended around 14,000 years ago and sea levels rose
dramatically, with the shoreline reaching its peak inland transgression 6000-7000
years ago where the suburbs of Fendalton and Riccarton are today (Owen 1992).
From this time, the Pegasus Bay coast has prograded at an average rate of around 2 m
yr-1 to reach its current position, which is thought to be a function of an increased
supply of sediment from the Waimakariri River. However, based on field evidence
and comparisons made with the first surveys conducted in the mid 1800s, it is thought
that the beaches of Pegasus Bay are currently in a state of long term equilibrium;
5
despite the recent history of rapid coastal advance (Kirk 1979). These periods of
significant shoreline advance and retreat mean that on a geological timescale, coastal
landforms in Pegasus Bay are relatively recent in origin (Allan et al. 1999).
1000 years ago Travis Swamp was a shallow protected bay with the Avon
River flowing into it from the north, the western margin of the current estuary was the
coastline, and the Heathcote River emptied directly into the sea. Sand brought to the
coast by the Ashley and Waimakariri Rivers was transported southward by longshore
currents and began the formation of Brighton Spit. As the spit grew it gradually
forced the Avon River mouth further and further southward until the waters from both
the Avon and Heathcote Rivers were enclosed by the spit around 450 years ago. The
earlier Avon Estuary then became Travis Swamp, and the discharge from the two
rivers pooled to form the Avon-Heathcote Estuary (Owen 1992). Figure 1.2 shows the
approximate locations of the past coastlines in Pegasus Bay.
Figure 1.2: The changing coastlines of Southern Pegasus Bay (Owen 1992).
6
People have been living beside the Avon-Heathcote Estuary for hundreds of years,
with numerous findings of Maori tools, burial sites, spears, bird and animal remains,
and stone artefacts providing evidence for this. The first Europeans arrived in the area
in 1843, and the Christchurch City settlement formally began in 1850 with the arrival
of the first four ships into Lyttelton Harbour. By 1875 over 10,600 people lived in
Christchurch. These early settlers were extremely dependant on the estuary for trade,
food and social contact, with £700,000 worth of goods entering the estuary in 1860
via a range of vessels up to 109 ton in size. The development of metalled roads and
the opening of the Lyttelton rail tunnel in 1867 destroyed the basis of the river trade,
and by the early 1900s the rivers had become very choked with silt, putting an end to
ship passage through the estuary. Various schemes were put forward between 1900
and the 1940s for dredging parts of the estuary and developing it into a port, and
connecting it to the city via a canal. However, these ideas became impractical,
especially with the success of the Lyttelton road tunnel (Owen 1992).
Today the Estuary is used for a variety of recreational activities including
fishing, swimming, windsurfing, kayaking, yachting and bird watching. As can be
seen from the air-photo in Figure 1.3 and the detail map in Figure 1.1, a significant
portion of the Estuary’s banks are occupied by houses and roads. The photos shown in
Figure 1.4 give an indication of the close proximity of some of the infrastructure to
the banks of the estuary. The map in Figure 1.1 also shows the Bromley sewage
treatment plant and oxidation ponds which border on, and discharge into, the estuary.
Figure 1.3: The Southern portion of the Estuary (Topomap).
7
Figure 1.4: Infrastructure surrounding the banks of the Estuary.
8
The Avon-Heathcote Estuary and surrounding low-lying land area is vulnerable to
tsunami due to its location on the east coast of New Zealand. The islands of New
Zealand form part of the seismically active ‘Ring of Fire’, an arc of subduction related
volcanoes which ring the Pacific Ocean. Seventy-five percent of the Earth’s active
volcanoes and most of the planet’s earthquakes and tsunamis occur as a result of the
Ring of Fire (Prager 2000), meaning that regions bordering the Pacific are especially
vulnerable to tsunami attack. In addition to its global location, the geological
development of the study site and its subsequent human inhabitancy, have combined
to generate the potential for a tsunami hazard event to occur.
1.3 Research Rationale
Tsunamis have struck the Canterbury coast on several occasions in recent history, and
have impacted at various levels on the Avon-Heathcote Estuary. As was outlined in
Section 1.2, the Estuary is surrounded by significant infrastructure, and it is therefore
important to try and quantify as accurately as possible, the effect that a potential
tsunami event would have on the area.
The Risks and Realities Report (1997) based on the work of the Christchurch
Engineering Lifelines Group assessed the vulnerability of Christchurch to a variety of
natural hazards. As part of this thesis an investigation was conducted into the effect a
tsunami event could have on the city. Some specific attention was paid to the Estuary,
but the thesis only provides an estimation of what areas could be affected, and the
inundation map is also much generalised. The section on the Estuary is concluded by
highlighting that ‘considerably more detailed investigations and modelling is required
on tsunami effects in estuary and river systems’.
This study aims to consider a range of realistic tsunami induced flooding
scenarios, and provide a more detailed and accurate account of the type of event that
could generate varying levels of inundation. The implementation of a Digital
Elevation Model and modern Geographic Information System (GIS) techniques will
allow a more precise picture of what could occur to be developed, and identification
of infrastructure that is at risk.
9
1.4 Thesis Outline
The following chapter (2) of this thesis will provide a review of tsunami literature
pertinent to this research. Chapter 3 will describe how the data was obtained and
utilised, and also describe techniques implemented for conducting the flood
inundation study. Chapter 4 will present the results of this work and Chapter 5 will
discuss the implications of these findings and their limitations. Finally a concluding
statement will provide a summary of all aspects of the study.
10
Chapter 2: Review of Pertinent Tsunami Literature
2.1 Introduction
This chapter will examine several aspects of tsunami waves relevant to the study
being conducted. Section 2.2 will provide a brief overview of how tsunami waves are
generated, how they propagate across the ocean, and how they behave as they reach
the shore. Section 2.3 considers the effects that tsunami waves can have in enclosed
spaces such as harbours and estuaries. The historic tsunami record for the Canterbury
Coast and some of the effects that have previously been observed in the AvonHeathcote Estuary and Lyttelton Harbour are outlined in Section 2.4, and Section 2.5
completes the chapter by comparing some common methods used to predict tsunami
effects.
2.2 Tsunami Waves
Tsunami or seismic sea waves can be generated by underwater landslides, volcanic
eruptions and meteorite impacts, but are most commonly triggered by submarine
earthquakes. However, not all earthquakes occurring under the sea generate a tsunami.
The vertical displacement of the sea-floor plays a vital role in determining the size of
a tsunami that is triggered, or whether one is triggered at all. Thrust and Normal faults
are good at generating tsunami waves as they push the overlying water up (thrust
fault) or pull it down (normal fault) which in turn activates wave motion as the water
surface attempts to restore itself. In contrast, strike-slip faults involve only horizontal
motion of the Earth’s crust, so there is typically little response in the overlying water
(Prager 2000). The magnitude and depth of an earthquake also influences the amount
of crustal deformation, and therefore wave generation, that occurs in a given event.
The diagram in Figure 2.1 shows the key processes involved in the typical generation
of a tsunami.
Both submarine landslides and landslides which collapse into the ocean are
also important tsunami generators. The largest tsunami in recorded history (Alaska,
1958) was triggered by a 40,000,000 m3 earthquake-induced rock-fall into Lituya Bay
(http://tsunamifury.org), and in 1998, 15 m high waves struck the coast of Papua New
11
Guinea after a relatively moderate earthquake triggered a massive underwater
landslide (Prager 2000). However in both of these cases, and in general, the energy of
tsunami waves generated from landslides or rock falls tends to be released over a
small area, and dissipates rapidly as they travel away from their generation area.
Therefore, although landslides have the potential to generate phenomenal wave
heights, a single event generally poses a lesser widespread threat than tsunami
generated directly from sea-floor displacement.
Figure 2.1. The typical processes of tsunami generation (Atwater et al., 1999).
The various generation events trigger a series of low, fast-moving waves which
radiate outwards in all directions from their generation area. In the open ocean these
waves typically have a period ranging between ten minutes to over one hour, a height
of 1-2 m, a wavelength hundreds of kilometres long, and move at speeds up to 700
kph (Prager, 2000, Bryant 2001). These characteristics, alongside masking by wind
generated sea and swell, make them virtually imperceptible to the human eye, and it is
often not until they reach shallower water that the true magnitude of the waves is
revealed.
Tsunamis are gravity waves; that is the restoring force is gravity, not elasticity
as in ordinary seismic waves. Their velocity V, is dependant on the depth of water h,
12
and the acceleration of gravity g, for wavelengths much greater than water depths
according to the equation
V = (gh) 1/2
As a tsunami approaches a coastline and the water depth begins to decrease, so does
the velocity of the waves. Some of the wave’s energy is lost to friction as they begin
to interact with the sea-floor, but most of the energy is preserved by an increase in
wave height (Doyle 1995).
The way in which tsunami waves are formed is therefore markedly different to
the way in which wind waves are formed. It is important to understand these
differences, as they have a direct consequence on the wave’s physical properties and
characteristics, which are different to the extent that they are incomparable in terms of
modelling their behaviour and effect.
As a tsunami wave propagates up the continental slope and across the
continental shelf it is greatly distorted, with processes of refraction, diffraction,
shoaling and reflection causing it to evolve and change shape (Zelt 1986). On entering
shallow water, the wave interacts with the sea-floor, flattening the orbital motion of
the water particles near the bottom as it begins to shoal. The wavelength and speed of
the wave decrease, and the height increases as the wave bunches up. Due to a
tsunamis long wavelength it can begin to shoal and steepen far offshore, growing to a
significant height by the time it reaches land (Prager 2000). A tsunami wave will not
always break at the shore like a typical wind wave, but in many cases will be
observed as a rapid rise in water level and inundation of the shore.
The near-shore bathymetry and coastal geometry are generally a crucial factor
with respect to the size, form and destructive power of a tsunami (Zelt 1986). In 1992
a 6 m tsunami hit the Pacific Coast of Nicaragua with a very high degree of
irregularity. In some regions little damage was done, whereas other sections of the
shoreline were completely decimated. A subsequent survey of the offshore
bathymetry revealed that most of the Nicaraguan coast is fronted by a submarine coral
reef, and that openings in the reef coincided with the greatest amount of damage along
the shore (Prager 2000). This provided strong evidence that the underlying structure
of the sea floor can greatly influence how tsunamis strike the coast.
13
2.3 Tsunami Effects in Enclosed Areas
‘Tsunami’ is a Japanese word which translates to ‘great harbour wave’. Japanese
historical documentation of tsunamis extends back over 1000 years, and it is likely
that in early Japan, harbours were where most people witnessed and recognised the
phenomena, and hence they were termed great harbour waves or tsunami (Prager
2000). It is not surprising this association was made, as enclosed areas can act to
amplify and trap long wavelength waves, resulting in bigger and more damaging
waves than are observed along more open stretches of coast (Zelt 1986). Murty (1977)
describes how waves generated by the 1964 Alaska Earthquake reached 5.2 m in
Alberni Inlet (Vancouver Island, British Columbia, Canada) but never exceeded 2.5 m
just 65 km up the coast at the exposed town of Tofino. This is one of many examples
where tsunami waves have been observed to become amplified in an embayment or
harbour relative to more open sections of coast. Eiby (1989) states that the speed of a
tsunami depends upon the depth of the water, and when shallow water is reached the
wave is abruptly slowed down, so that the front piles up and breaks with tremendous
force, particularly if it is confined within an estuary or a narrow bay. Doyle (1995)
also makes the assertion that the greatest tsunami wave heights occur in funnelled
bays and estuaries.
When tsunami waves enter a harbour they can be amplified by bouncing off of
the harbour’s embankments and combining with reflected waves, allowing them to
reach many times their open-coast height (Prager 2000). Also, if a tsunami enters an
inlet or harbour with a natural period of oscillation near the dominant period of the
tsunami wave, its amplitude may be amplified greatly by the process of resonance.
This occurs because a wave which reflects from the head of a basin propagates back
offshore toward the sea, but then gets reflected from the harbour entrance back into
the basin, trapping some of the wave’s energy within the harbour. If these trapped
waves are in phase with incident waves entering from the open sea they can combine
and become even larger and more destructive (Zelt 1986). A study conducted by
Raichlen et al. (1983) into the excitation of harbours by tsunamis, demonstrates the
influence of local features, incident waves and harbour dynamics on the nature of the
waves that reach the shore.
In May 1960, a subduction related earthquake occurred off the coast of Chile
with a magnitude of 9.5 on the Richter scale, the largest earthquake ever recorded.
14
The release of energy triggered a large tsunami which was felt throughout the Pacific
causing significant damage and loss of life (Atwater et al. 1999). In some areas of the
globe such as Japan, water levels increased as much as 20 m as the wave propagated
across the Pacific. In New Zealand, Lyttelton Harbour responded the strongest of all
the countries harbours, with a 3 m rise in water level being recorded by the gauge
located within the basin (Heath 1974). Heath concludes that the high response of the
harbours in and around Banks Peninsula to the Chilean tsunami can be attributed
significantly to the openness of the harbours, and the channelling of energy from the
north toward Lyttelton by the Hikurangi Trench and trapping by the Chatham Rise
and New Zealand land mass. These features are shown in Figure 2.2.
Figure 2.2: Campbell Plateau to the South-east and Chatham Rise directly Eastward
of Banks Peninsula (www.niwascience.co.nz.).
15
Although the offshore bathymetry is attributed with helping to deflect waves toward
Banks Peninsula, it can also be accredited with helping to dissipate a significant
amount of wave energy. Eiby (1989) describes how a tsunami wave is forced to slow
down as it reaches the outer edge of the large and comparatively shallow Campbell
Plateau (Figure 2.2) and a large part of its energy is dissipated before it reaches the
New Zealand Coast. This is thought to account for the significant difference in wave
heights recorded on the New Zealand and Japanese coasts, especially noticeable
during the 1960 Chilean tsunami.
There is very little literature which gives specific consideration to the
observed or predicted behaviour of tsunami waves in estuaries such as the AvonHeathcote. Papers such as McFadgen and Goff (2005), and Nichol et al. (2001)
discuss the stratigraphic record of the Estuary and show there to be evidence of
tsunamis occurring previously in the area. But aside from providing further evidence
that tsunamis have struck the Canterbury coast in the past, these give little assistance
to predicting the area a tsunami would inundate in today’s setting. However, based on
the knowledge of general tsunami behaviour and tsunami behaviour in enclosed areas,
some predictions about what could occur in the Avon-Heathcote Estuary can be made,
and are listed below.
•
Amplification of the waves may occur as they are confined by the dimensions
of the Estuary inlet.
•
Tsunami waves will likely be diffracted as they pass from the open ocean,
through the inlet and into the Estuary (shown graphically in Figure 2.3).
•
If wave is big enough, inundation and water entering the estuary via
overtopping of the New Brighton Spit are likely to occur.
•
It is likely that the process of resonance (outlined earlier in this Section) may
trap wave energy within the Estuary, causing inundation of the Spit from the
Estuary side, and further amplification of incoming waves.
•
Scouring of the Estuary inlet will occur, allowing greater volumes of water
from following tsunami waves to enter the estuary.
16
Figure 2.3: The diffraction of water waves as they pass through a gap
(www.gscescience.com).
The most significant difference between the Estuary and a harbour such as Lyttelton
is that the Estuary is fronted by a shallow-sloping beach, whereas a harbour generally
has a direct connection to the open ocean. The photo in Figure 2.4 looking south from
the end of New Brighton Spit shows waves breaking before they pass through the
inlet and into the Estuary. This indicates a dissipation of wave energy not generally
seen in harbour entrances. A similar contrast could potentially be evident in the event
of a tsunami.
Figure 2.4: Breaking waves at the entrance to the Avon-Heathcote Estuary.
17
2.4 Historical Tsunami Record for the Avon-Heathcote Estuary
Since written records began around 1840, at least 32 tsunami events have been
recorded in New Zealand, most of which have affected the east coast. These tsunamis
have generally been earthquake triggered, but some events have also been attributed
to large rotational slumps, submarine slumping along the Chatham Rise, and
submarine volcanism off the east coast of the North Island. Nine tsunamis from
distant sources have been identified, and those generated in South America tend to
produce higher waves in New Zealand, although the highest recorded waves have
been derived from locally generated tsunamis (De Lange and Healy 1986). Table 2.1
provides a summary of the past tsunami events which have had a notable impact on
Lyttelton Harbour and/or the Avon Heathcote Estuary.
Table 2.1: Significant historic tsunami events observed along the Canterbury coast.
Derived from De Lange and Healy (1986) and Findlay and Kirk (1988).
Date
Trigger
Description
January 23
Magnitude 8 earthquake,
Small bore travelled up Avon River, seaweed left up to
1855
West Wairarapa.
0.3 m above normal water level
August 13
Unknown Magnitude
Major fluctuations in Lyttelton Harbour. 0.9 m above
1868
earthquake, North Chile
highest spring tide. Okains Bay, water level 3 m above
normal high water, and reached over 3 km inland
May 10
Unknown Magnitude
Lyttelton Harbour: Water level reached above high water
1877
earthquake, North Chile
mark at half flood tide and then fell exceedingly low
August 27
Volcanic Eruption,
Unusually high tides in Lyttelton, and then sudden drops
1883
Krakatoa, Indonesia
in water level up to 1.8 m below normal low water level
November 11
Magnitude 8.3 earthquake
Minor tidal fluctuations observed at New Brighton
1922
off Chilean coast
May 22
Magnitude 9.5 earthquake
Lyttelton Harbour: Highest level above normal high tide
1960
off Chilean coast
was 3.3 – 3.5 m. Many damage claims from Banks
Peninsula
March 28
Magnitude 8.4
Lyttelton Harbour: Fluctuations of 0.9 m every 20 min,
1964
earthquake, Alaska
with a maximum rise of 1.25 m
18
2.5 Methods Used to Predict Effects of Tsunami
The importance of offshore bathymetry and form of the land to the distortion and
development of tsunami waves is extensively discussed in most relevant literature.
However another common underlying theme is that although the importance of these
topographic features is known in general, a dependable way of predicting the effect
they will have on a particular tsunami wave at a particular site is not. Tsunamis strike
coasts relatively infrequently and usually with little warning, meaning that (aside from
obvious safety reasons) observations, measurements, and documentation of their
behaviour are severely restricted. It is also unrealistic to assume that if a tsunami
behaved in a certain way on a given piece of coast that another tsunami generated at a
different location in a different way will behave in a similar fashion.
Attempting to accurately simulate a tsunami wave in a laboratory is also
highly problematic. Consider trying to model a 1 m high tsunami generated at an
ocean depth of 2000 m, with a wavelength of 100 km. To create a 1:100 scale model
would require a tank 20 m deep and at least 1 km long. If a larger scaling ratio such as
1:1000 was used, the dimensions of the tank would be manageable but the wave
would be only 1 mm high, with surface tension making interpretation of any results
nearly impossible or highly inaccurate (Prager 2000).
A large number of mathematical models and computer simulations have been
developed to try and gain a more comprehensive understanding of how tsunamis
form, travel and strike the shore (Bryant 2001, Li 2000, Goring 1978), and how they
behave in harbours (Zelt 1986, Raichlen et al. 1983) with the aim of providing more
accurate warnings and better risk assessments. However, a general lack of field
evidence and the extreme difficulty in conducting representative laboratory
experiments makes judging the reliability of any models extremely difficult. The
magnitude and nature of a trigger event has a significant influence on the magnitude
and nature of the tsunami that it creates. Therefore to accurately predict a tsunami’s
propagation, it is also necessary to accurately predict the characteristics and location
of the trigger event, which is a near impossible task in itself.
It can now be clearly seen that even a moderately accurate prediction of
tsunami events is an extremely difficult task, as consequence of a variety of impeding
factors. It is therefore essential that all assumptions made in calculating tsunami wave
run-up or land inundation are clearly acknowledged, something which this study takes
19
into careful consideration. If assumptions and estimations are not made clear it is
possible that studies such as this can have a negative impact from a hazard
management perspective, especially if they are taken as gospel and prove to be
incorrect.
The available mathematical models formulated to calculate tsunami behaviour
are either inapplicable for estuary environments, have significant associated
inaccuracies, or are too complex for the purposes of this study. As a result, estuary
environments are often not given the special consideration of which they are worthy.
Flood inundation analysis is a common approach for modelling potential fluvial
hazards such as rivers, dams, and lakes. Many different methods have been developed
with the aim of trying to predict or simulate the extent of flooding in a specified
scenario. These include simple buffering of rivers, the use of 2-dimensional
mathematical models, event-based probabilistic modelling, and the use of 3dimensional Digital Elevation Models (Kleinschroth 2005).
These types of modelling techniques are generally not used in coastal settings
such as the Avon-Heathcote Estuary, but for the purposes of this research, a flood
modelling approach provides a more useful means for assessing the likely tsunami
inundation than the more traditional wave run-up focused approach.
20
Chapter 3: Methodology
This section describes how the Digital Elevation Model for the estuary and
surrounding area was constructed, how the tsunami inundation modelling was carried
out, and how the relevant statistical data was calculated
3.1 Data
The primary data set used for this project was a series of 17 LIDAR (Light Detection
and Ranging) tiles which covered the estuary and some surrounding land area. These
tiles were obtained from, and are copyright to the Christchurch City Council. A
LIDAR system works in a similar fashion to the more well known RADAR, but
instead of radio waves it emits rapid pulses of laser light to precisely measure
distances from a sensor to targets on the ground. The sensor is typically mounted on
an aircraft, and the time taken for the pulse of laser light to be returned to the sensor is
measured and converted into distance using the known velocity of light. The
combination of a Global Positioning System (GPS) and Inertial Navigations System
(INS) is used to determine the precise location of the sensor and attitude of the
aircraft, so that each laser ‘hit’ can be georeferenced into x, y, z data-points. These are
then corrected using GPS ground reference stations (www.emporia.edu). Figure 3.1
shows a graphic representation of an operational LIDAR system.
Figure 3.1: The major components of an operational LIDAR system
(www.emporia.edu).
21
The seventeen tiles of LIDAR data (Figure 3.2) covering the Avon-Heathcote Estuary
and surrounding area were obtained from the Christchurch City Council as Golden
Software Surfer 8 grid files. The data has a vertical accuracy of approximately 10 cm,
a spatial resolution of 2 m2 and is georeferenced to New Zealand Map Grid (NZMG)
coordinates, based on the New Zealand Geodetic Datum 1949.
Figure 3.2: The 17 LIDAR tiles used to define the study area. Note that each tile
represents an area of 400 ha (Christchurch City Council).
3.2 Inundation Study
Having obtained the required LIDAR data, each tile was made into a Raster layer by
saving the original Surfer Grid file as a Notepad text file, and then using the ‘ASCII
to Raster’ conversion tool in ESRI ArcMap 9.1. These 17 raster layers were combined
into a single 32 bit float raster layer covering the whole field area using the ‘Mosaic to
New Raster’ function. The height values in the LIDAR data are relative to 1937 mean
sea level (MSL) which, for its use, is assumed to be the same as current MSL (i.e. a
pixel value of 5.2 in the LIDAR raster is equal to 5.2 m above MSL a value of -0.4 is
equal to 0.4 m below MSL etc.).
To determine the areas which could be flooded during varying tsunami
scenarios the LIDAR data was reclassified into a number of binary images which
defined all areas above and below a certain elevation or water level. This method is
22
one of the most common flood analysis techniques, and is in widespread use around
the globe, mostly for representing the effects of predicted sea level rise (Lomax,
2005).
Values of 1 were assigned to the ‘flooded’ areas and also to the ‘No Data’
values in the LIDAR so that areas of the estuary for which there was no elevation
information would also be included in the areas of flooding. This reclassification was
carried out using the ‘Reclassify’ function on the Spatial Analyst toolbar. An example
of this method is shown below in Figure 3.3.
Figure 3.3: Original classification of the LIDAR data values are shown on the left and
the reclassification for a 2 m water level is shown on the right.
A layer defining the Canterbury coastline was added into ArcMap and referenced to
NZMG to match the LIDAR data. This line feature was modified to enclose all of the
land area covered by the LIDAR data, and a new polygon was created from these
lines using the ‘trace’ function. The ‘extract by mask’ spatial analyst tool was used to
extract the land areas of the binary flood images. The new polygon feature was used
as the mask, and enabled everything landward of the coastline to be extracted, and
new binary images to be created with the unwanted data removed.
The total land area that would be flooded in each scenario could then be
calculated, as the area that is always flooded (i.e. the Estuary) had been removed. The
total pixel count for the ‘flooded’ class displayed in the attribute table could be
multiplied by 4 to convert to m2 (as pixels were 2 x 2 m) and divided by 10 000 to
convert the total flooded land area to ha.
23
Three NZMG referenced, 1:50 000 scale Topomap images were added to ArcMap
from the University Geography Network, and their display extent set to match that
defined by the bounds of the LIDAR data. 2.5 m resolution aerial photographs from
Topomap covering the study area were saved as six JPEG files and also added into
ArcMap. These were assigned the NZMG projection and were also set to display to
the same extent as the LIDAR data. The maps and photos were added to provide a
graphical view of the areas and infrastructure that would be submerged during each
flooding scenario. The 0 values of each binary flood image were set to display as ‘no
colour’ and the flooded areas (values of 1) were displayed at 30% transparency over
the maps and 50% transparency over the photos, so the features in the underlying
images could be distinguished. Different colours were used so that more than one
scenario could be displayed on the same map layout.
A cross sectional view of the dunes running the length of Brighton Spit was
also created in ArcMap using the interpolate line feature in the 3D analyst toolbar. A
line was drawn along the length of the dunes, and a graph was generated based on the
LIDAR elevation data, allowing the varying heights of the dunes to be easily
identified.
ArcScene was used to generate 3-dimensional views of what the flooded
estuary would look like. The LIDAR data was added into ArcScene, and set up to
highlight the topographical features of the study area. The LIDAR data was used to
define the base heights for the layer, and the vertical exaggeration was set to 5 to help
display the topographic detail in the relatively flat areas surrounding the Estuary. An
appropriate colour ramp was chosen, and a histogram stretch applied to it. This was
another way of providing added contrast between the lower features, as the high
elevations of the port hills would otherwise use up a large proportion of the 255
display colours available. The various flood scenarios could easily be added to this
view, and assigned the relevant base height (i.e. a 2 m flood scenario assigned a base
height of 2 m) to cover the topography that would be inundated.
Water level scenarios of 2, 3, 4 and 5 m were chosen based on the levels
observed during historic tsunamis along the Canterbury coast, potential future tsunami
events, and the estimates made by the Risks and Realities Report (1997). The type of
tsunami and the processes and characteristics which could lead to these water depths
within the Avon-Heathcote Estuary will be given more detailed consideration in
Chapter 5 of this thesis.
24
Section 2.3 of this thesis outlined some of the potential effects of tsunami waves in
enclosed areas, but also illustrated the lack of specific consideration given to the
prediction and observation of tsunami behaviour in estuaries such as the AvonHeathcote. The many problems and limitations associated with accurate modelling of
tsunamis were discussed in Section 2.5. It should be made clear that the methods used
in this study do not attempt to model the effect of a particular tsunami event. Instead,
a range of potential inundation scenarios are quantified, and consideration then given
to the type of event and processes which could cause these flood conditions.
As with any tsunami inundation modelling, the methods used have some
limitations, and require that some assumptions are made. In this case the flood
modelling approach is unable to take into account the complex wave dynamics and
flow mechanisms discussed in Section 2.3, and assumes the water is able to spread
evenly across the study area. However in the relatively flat Avon-Heathcote area,
physical boundaries which could prevent water flow are not a serious issue.
Considering these factors, the methods used, in conjunction with knowledge of
tsunami behaviour, provide a good practical technique for generating tsunami hazard
maps within the confines of this project.
3.3 Field Observations
An attempt was also made to create a digital elevation model of the Estuary
bathymetry which could be combined with the LIDAR data to give a complete
representation of the topography above and below mean sea level. Two days field
work was carried out to collect depth and location data using a depth sounder and
hand held GPS unit. This data was geometrically corrected and adjusted so that the
depths shared the same reference level as the LIDAR data. However the amount of
data was insufficient to interpolate a surface which represented the floor of the estuary
with any real accuracy, and it was decided using this data would detract from, rather
than add anything to LIDAR data set and the findings of the study. The time spent in
the field allowed for useful observations of the study site and the proximity of
infrastructure to the bounds of the Estuary, and to the water level. It also allowed for
the accuracy of the LIDAR data to be verified, and the scale of the potential areas that
could be affected to be gauged and better understood.
25
Chapter 4: Results
This chapter will present the results of the four relevant flood inundation scenarios,
generated based on the techniques and assumptions outlined in Chapter 3. The type of
tsunami events that could lead to these flooding scenarios and their implications from
a hazard assessment perspective will be discussed in Chapter 5 of this thesis. It should
be noted that all flooding scenarios are relative to mean sea level (i.e. a 2 m flood is
equal to 2 m above mean sea level), and therefore a tsunami that would generate a 3 m
flood at mean sea level could generate a 4 m flood if it coincided with high tide.
The total area (in hectares) flooded under each of the four inundation scenarios
is displayed in Figure 4.1, and Figure 4.2 shows the extent of flooding for each
scenario. Note that the total area considered for the inundation analysis is 4497.3 ha.
3500
Flooded Area (Hectares)
3000
2500
2000
1500
1000
500
0
2
3
4
5
Scenario Water Level (m)
Figure 4.1: Total land area flooded under each scenario
26
Figure 4.2: Areas flooded under each scenario.
Considering that the total area of the estuary is around 880 ha (Owen 1992), a 2 m rise
in water level would lead to a greater than 100% increase in the total proportion of the
study site which is covered by water, and a 3 m rise would result in a more than 200%
increase. The four flood scenarios are shown laid over 2.5 m resolution aerial
photographs of the area in Figure 4.3, and over a 1:50 000 topographic map in Figure
4.4, to highlight the features which would be inundated given the occurrence of each
event.
27
Figure 4.3.1: 2 and 3 m flooding scenarios.
Figure 4.3.2: 4 and 5 m flooding scenarios.
28
Figure 4.4.1: 2 and 3 m flooding scenarios.
Figure 4.4.2: 4 and 5 m flooding scenarios
29
Figure 4.4 is especially useful for highlighting the considerable number of roads that
would be inundated by the flooding scenarios. As can be seen in Figure 4.4.1, there is
a considerable variation between the amount of infrastructure flooded by a 2 m and 3
m rise in water level.
It is worth giving special mention to the Bromley sewage oxidation ponds
which would not likely be inundated by anything less than a 5 m rise in water level.
Wave run-up effects on top of a high water level could lead to some overtopping, but
it is expected these effects should be minimal, especially in this area of the Estuary.
Figure 4.5 provides a 3-dimentional view of a 4 m rise in water level looking
southward across the Estuary. This view highlights the pond stop banks protruding
above the 4 m water level with no gaps between, indicating that this important piece
of infrastructure may not actually be flooded, although the results of the modelling
indicate that it would.
Figure 4.5: 4 m rise in water level. Oxidation ponds in the mid-right of the image.
30
Figures 4.6.1 and 4.6.2 display how the landscape would change given a 3 m rise in
water level. These two images also show that the oxidation ponds are the only
significant discrepancy caused by the flood modelling technique used to obtain these
results. No other areas which are predicted to become flooded have topographical
boundaries separating them from the water source, and it is feasible that the identified
areas could be submerged in a tsunami event.
Figure 4.6.1: The Estuary at MSL.
Figure 4.6.2: The Estuary with a 3 m inundation.
31
The Risks and Realities Report (1997) used a wave run-up formula to calculate the
dune or beach structure heights which would be required to prevent inundation from a
tsunami water level. For a water level of 5 m above MSL it was calculated that a dune
or structure height of 8 m would be required to prevent overtopping. Water levels of 4
and 3 m were calculated to require a dune/structure height of 7 and 5 m respectively
to prevent inundation. Figure 4.7 displays a cross sectional profile of the dunes which
run the length of Brighton Spit, and are the predominant topographical feature
separating the Estuary from the Pacific Ocean.
Figure 4.7: Cross sectional profile of the Brighton Spit sand dunes.
Figure 4.7 shows that a tsunami induced water height of 3 m is likely to be prevented
from entering the estuary via overtopping of the dunes. However there are a number
of areas where the dunes are less than 7 or 8 m in height, which means it is likely that
overtopping would occur in event of a 4 or 5 m rise in water level. Through field
measurements it was also found that the floor of the Estuary is extremely shallow, and
was impassable by boat in many places, even at high tide. In general the floor of the
estuary has a depth of approximately 1 m or less relative to MSL which has important
implications for wave propagation.
32
Chapter 5: Discussion
5.1 Tsunami Events which could generate the Flooding Scenarios
The four inundation scenarios assessed in this study were chosen based on the
tsunami history for the Canterbury coast, the estimations made by the Risks and
Realities Report (1997), and general knowledge of tsunami wave dynamics. This
section will outline the type of tsunami event and associated processes which could
result in the considered flooding magnitudes of 2, 3, 4 and 5 m. Due to a lack of
historical data specific to the Avon-Heathcote Estuary and the complex behaviour of
tsunami waves, the likely response of the Estuary to a given event is difficult to
ascertain and, as a result, these descriptions are somewhat speculative.
Based on historical records (Section 2.4, Table 2.1.), the most likely
generating source for a significant tsunami affecting Christchurch is from a large
seismic event centred off coastal South America. The Risks and Realities Report
(1997) considered an assumed 150 year minimum return period tsunami hazard
scenario in which the water level was raised by 5 m above MSL along the Canterbury
coast. It was predicted that this water level would be reduced to 3 m at the entrance to
the Estuary due to dissipation of the wave energy in the limited water depths on the
ebb tide deltas. Considering these estimates, and the water levels recorded in Lyttelton
during the 1868, 1877 and 1960 events (Section 2.4, Table 2.1), it is reasonable to
assume that a moderate to large Pacific-wide tsunami could generate water levels 2 m
above MSL within the Avon Heathcote Estuary. If these waves were to coincide with
a high tide a water level 3 m above MSL could be realistically achieved.
As was described in Section 2.3, the Chatham Rise and Campbell Plateau
(Section 2.3, Figure 2.2.) to the east and south-east of New Zealand are often
attributed with dissipating a large amount of tsunami wave energy (Eiby 1989, De
Lange and Healy, 1986). De Lange and Healy (1986) also states that New Zealand has
experienced roughly equal numbers of tsunamis from local and distant sources, and
that wave dissipation is more pronounced for distantly generated tsunamis than for
those generated close to shore. This source also indicates that the largest tsunamis
experienced in New Zealand have been locally generated, and that tsunami source
mechanisms have been attributed to submarine slumping along the Chatham Rise.
33
Therefore, based on this basic knowledge, the potential for a locally generated or
near-field tsunami to have an impact on the Avon-Heathcote Estuary should not be
discounted. Therefore, it is feasible that a locally generated tsunami, or a tsunami
generated distantly at higher latitudes (which may retain more of its energy by
travelling through deeper water), could have the potential to generate water levels of 4
or 5 m above MSL in the Estuary. When attempting to carry out a specific hazard
assessment for an area it is often useful to consider a worst case scenario, and as a
result these water levels have been included in the study.
As was discussed in Chapter 4 of this thesis, and shown in Figure 18, there are
many places in which a wave with a height of 4 or 5 m at the coast could easily
overtop the dunes dividing the Estuary from the open ocean. Scouring of the Estuary
inlet, and erosion associated with overtopping of the dunes would have an important
effect on how much water is able to get past these barriers and into the Estuary. This
would be primarily dependant on the number, frequency, and energy of the waves in
the given tsunami event.
5.2 Flooding implications
The maps in Chapter 4 of this thesis (Figures 4.3 and 4.4) show the areas which would
be inundated with each flooding scenario. Based on the work that has been done, it is
extremely difficult to estimate the damage that could result from a tsunami event. Due
to the vast range of variables involved, it would be a near impossible task for anyone
to accurately predict the type and extent of damage that a given event could cause. It
is obvious that the majority of serious damage which is sustained to buildings and
other infrastructure when a tsunami strikes the shore is a function of the wave energy;
something which this study is unable to quantify or account for. However, in the areas
surrounding the estuary, serious structural damage to buildings is expected to be
minimal. Water which reaches these areas will have first had to pass through the
Estuary inlet, or over the New Brighton Spit and dunes, with either route significantly
reducing its destructive potential along the way. The shallow depth of the Estuary
floor (in most places less than 1 m) will also aid in significantly reducing wave speed
and force. Therefore most property and infrastructure within the study area is likely to
sustain inundation related damage as opposed to being smashed up or washed away
by the sudden release of wave energy.
34
5.3 Review of Findings and Techniques
The methods used in this research have allowed the aims of this thesis to be achieved.
A good understanding of the physical properties and characteristics of tsunamis has
been developed, and the likely effects that various tsunami events would have on the
Avon-Heathcote Estuary and surrounding area has been assessed. Accurate data, an
understanding of the physical processes involved, and the use of appropriate analysis
techniques have been utilised to build on previous work done in the area. As was
outlined in Section 2.5, there are a range of difficulties associated with attempting to
accurately determine the tsunami hazard, and predict the effects of a realistic tsunami
event for a specific area. While the flood analysis methodology adopted for this study
has some important limitations (described in Section 3.2), when coupled with high
quality data, it provides a very useful means for quantifying the effects of a tsunami.
The nature of the study area meant that this method had fewer limitations than it could
when applied in other situations. The relatively flat terrain surrounding the Estuary
meant that water was not predicted to inundate areas that it could not possibly reach,
which is a common problem in more complex terrain. This method would be of little
use for estimating tsunami inundation on an exposed piece of coastline due to the
effect of wave run-up. However, in this case, the majority of incident wave energy
will be released along the coast before the water enters the Estuary via the inlet or
gaps in the dune. Therefore, a simple ‘drowning’ of the study area is a reasonable
representation of the way in which the inundation would actually occur. GIS
techniques were also very useful in this scenario for quantifying the total land area
which would be flooded, and also for mapping and displaying the effected regions.
5.4 Potential Future Research
While this study has built on previous knowledge and produced some useful results, it
has also provided a good base for future research to be conducted in the area. The
LIDAR digital elevation model would strongly assist any other attempts made to
model tsunami flood inundation or wave behaviour. The utilisation of a more
comprehensive modelling technique would be useful, to try and predict with more
certainty the areas that would become flooded with the given scenarios. An
investigation into the likelihood of a locally triggered tsunami occurring off the
35
Canterbury coast, and the potential magnitude of such an event would also be a useful
step in attempting to best estimate the hazard posed by tsunami in the AvonHeathcote Estuary. A return period for the four inundation scenarios considered in
this thesis would also add to the value of the overall hazard assessment.
5.5 Conclusion
This study found that significantly large areas of land are likely to be inundated in the
area surrounding the Avon-Heathcote Estuary in event of a tsunami striking the
Canterbury coast. Due to the physical characteristics of the area, it is likely that wave
energy within the Estuary will be significantly lower than that observed along the
open coast. Therefore the infrastructure covering the at risk areas (mostly roads and
houses) is likely to suffer damage more comparable with a large river flood than the
destruction commonly associated with a tsunami wave breaking on the shore.
However, based on this research and historical tsunami records, this hazard still poses
a serious threat to the Avon-Heathcote Estuary.
36
References
Allan J C, Kirk R M, Hemmingson M, and Hart D, 1999, Coastal processes in
Southern Pegasus Bay: a review. Unpublished Report to Environment Canterbury.
Land and Water Studies (International) Ltd, Christchurch.
Atwater B, Cisternas M, Bourgeois J, Dudley W, Hendley J, and Stauffer P, 1999,
Surviving a Tsunami – Lessons from Chile, Hawaii, and Japan, U.S. Geological
Survey Circular 1187.
Bryant E, 2001, Tsunami, the Underrated Hazard, Cambridge, Cambridge University
Press.
Centre for Advanced Engineering, Christchurch Engineering Lifelines Group, 1997,
Risks and Realities: A Multi-Disciplinary Approach to the Vulnerability of Lifelines
to Natural Hazards, Christchurch, New Zealand, Centre for Advanced Engineering,
University of Canterbury.
De Lange, W P and Healy T R, 1986, New Zealand Tsunamis 1840-1982, New
Zealand Journal of Geology and Geophysics, Vol. 29.
Doyle H A, 1995, Seismology, Chichester, England, John Wiley and Sons Ltd.
Eiby G A, 1989, Earthquakes, Auckland, Heinemann Reed
Findlay R H, and Kirk R M, 1988, Post 1847 changes in the Avon-Heathcote Estuary
Christchurch: a study of the effect of urban development around a tidal estuary, New
Zealand Journal of Marine and Freshwater Research, Vol. 22.
Goring D G, 1978, Tsunamis – the propagation of long waves onto a shelf, California,
Keck Laboratory of Hydraulics and Water Resources, California.
37
Heath R A, 1974, The Response of Several New Zealand Harbours to the 1960
Chilean Tsunami, New Zealand Oceanographic Institute Department of Scientific and
Industrial Research, Wellington, New Zealand.
Kirk R M, 1979, Dynamics and Management of Sand Beaches in Southern Pegasus
Bay. Morris & Wilson Consulting Engineers Ltd.
Kleinschroth S, 2005, Flood Modelling: A challenge for natural catastrophe
management, Exposure Property and Engineering
Komar P D, 1998, Beach Processes and Sedimentation, Upper Saddle River, N. J.,
Prentice Hall.
Li Y, 2000, Tsunamis: Non-breaking and breaking solitary wave run-up, California,
Keck Laboratory of Hydraulics and Water Resources, California.
Lomax A, 2005, A coupled shoreline change and flood modelling approach for
coastal impact assessment Tagaqe, Viti Levu, Fiji. Unpublished MSc Thesis,
Auckland University.
McFadgen B G, and Goff J R, 2005, An earth systems approach to understanding the
tectonic and cultural landscapes of linked marine embayments: Avon-Heathcote
Estuary (Ihutai) and Lake Ellesmere (Waihora), New Zealand. J. Quaternary Sci.,
Vol. 20 pp. 227–237.
Murty T S, 1977, Seismic Sea Waves Tsunamis, Ottawa, Department of Fisheries and
the Environment, Fisheries and Marine Service.
Nichol A, Howard M, Campbell J, and Pettinga J, 2001, Palaeoearthquakes on the
Porters Pass Fault. EQC research report 99/386, New Zealand Earthquake
Commission, Wellington, New Zealand.
Prager E J, 2000, Furious Earth: The Science and Nature of Earthquakes, Volcanoes
and Tsunamis, New York, McGraw Hill.
38
Raichlen F, Lepelletier T G, and Tam C K, 1983, The Excitation of Harbours by
Tsunamis. In Iida and Iwasaki, 1983, Tsunamis, Their Science and Engineering, Terra
Scientific Publishing Company, Tokyo, pp 359-385
Zelt J A, 1986, Tsunamis: The Response of Harbours with Sloping Boundaries to
Long Wave Excitation, California, W M Keck Laboratory of Hydraulics and Water
Resources Division of Engineering and Applied Science, California Institute of
Technology.
GSCE Science, http://www.gcsescience.com/a/Diffraction-Water-Waves.gif, (12
October 2005).
Use of LIDAR in Creating Accurate Terrain Elevation Models for Floodplain
Mapping, http://www.emporia.edu/earthsci/student/serr1/project.htm, (28 September
2005).
New Zealand Topographical Maps and Aerial Photography v 3.00.154 (Topomap)
39