University of Southern California Dissertations and Theses

Modeling potential impacts of tsunamis on Hilo, Hawaii: comparison of the Joint Research Centre's SCHEMA and FEMA’s HAZUS inundation scenarios

(USC Thesis Other)

Modeling potential impacts of tsunamis on Hilo, Hawaii: comparison of the Joint Research Centre's SCHEMA and FEMA’s HAZUS inundation scenarios

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

MODELING POTENTIAL IMPACTS OF TSUNAMIS ON HILO, HAWAII:
COMPARISON OF THE JOINT RESEARCH CENTRE'S SCHEMA AND
FEMA’S HAZUS INUNDATION SCENARIOS

by

Matthew Kline

A Thesis Presented to the
Faculty of the USC Graduate School
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
Master of Science
(Geographic Information Science and Technology)

August 2016

ii

Copyright ® 2016 by Matthew Kline

iii

Acknowledgements
I would like to sincerely thank Dr. Jennifer Swift for her continual commitment to my
thesis process. I would also like to thank Dr. Karen Kemp, Dr. Daniel Warshawsky, Dr. Laura
Loyola, and Dr. Steven Fleming for their support and guidance.

iv

Table of Contents
Acknowledgements ........................................................................................................................ iii
List of Figures ................................................................................................................................ vi
List of Tables ............................................................................................................................... viii
List of Abbreviations ..................................................................................................................... ix
Abstract ........................................................................................................................................... x
Chapter 1 Introduction .................................................................................................................... 1
1.1 Motivation ............................................................................................................................4
1.2 Background ..........................................................................................................................9
1.2.1. Local Topography and Bathymetry ...........................................................................9
1.2.2. Historical Tsunamis .................................................................................................12
1.2.3. NOAA Tsunami Forecast ........................................................................................17
Chapter 2 Related Works .............................................................................................................. 20
2.1 JRC SCHEMA Methodology ............................................................................................20
2.2 Tsunami Modeling For Seaside, Oregon ...........................................................................23
2.3 HAZUS Flood Technical Manual ......................................................................................26
Chapter 3 Methodology ................................................................................................................ 33
3.1 Data ....................................................................................................................................33
3.2 Methodology ......................................................................................................................38
3.2.1. SCHEMA Model in ArcGIS ....................................................................................38
3.2.2. HAZUS Flood Model ..............................................................................................44
Chapter 4 Results .......................................................................................................................... 50
4.1 SCHEMA Model ...............................................................................................................50
4.1.1. 1946 tsunami event ..................................................................................................52
4.1.2. 1960 tsunami event ..................................................................................................55
4.2 HAZUS Model ...................................................................................................................57
4.2.1. 1946 HAZUS Inundation Scenario ..........................................................................58
4.2.2. 1960 HAZUS Inundation Scenario ..........................................................................64
4.2.3. 2016 HAZUS Inundation Scenario ..........................................................................71
v

4.3 Historical Comparison .......................................................................................................74
4.3.1. 1946 Tsunami...........................................................................................................75
4.3.2. 1960 Tsunami...........................................................................................................76
Chapter 5 Conclusions and Future Work ...................................................................................... 78
5.1 Recommendations ..............................................................................................................79
5.2 Future Work .......................................................................................................................84
References ..................................................................................................................................... 87

vi

List of Figures
Figure 1. 1946 Hilo Tsunami .......................................................................................................... 1
Figure 2. Hilo after 1946 Tsunami ................................................................................................ 2
Figure 3. Ring of Fire...................................................................................................................... 6
Figure 4. Reflecting of Waves in Hilo Bay................................................................................... 10
Figure 5. Hilo Topography ........................................................................................................... 12
Figure 6. Tsunami Heights in 1960............................................................................................... 15
Figure 7. HAZUS Flood Model Schematic .................................................................................. 27
Figure 8. FIMA Building Flood Depth-Damage Curve................................................................ 30
Figure 9. Hilo 1946 Tsunami Inundation ...................................................................................... 35
Figure 10. Hilo 1960 Tsunami Inundation .................................................................................... 36
Figure 11. Overview of the SCHEMA Tsunami Flood Model ..................................................... 39
Figure 12. Hilo Tsunami Evacuation Zones. ................................................................................ 40
Figure 13. Standard Deviation Bell Curve .................................................................................... 47
Figure 14. Overview of the HAZUS Tsunami Flood Model ........................................................ 48
Figure 15. Hilo Buildings at Risk. ................................................................................................ 51
Figure 16. 1946 Tsunami Inundation ............................................................................................ 52
Figure 17. 1946 Hilo Dollar Exposure .......................................................................................... 53
Figure 18. 1960 Tsunami Inundation ............................................................................................ 55
Figure 19. 1960 Hilo Dollar Exposure .......................................................................................... 56
Figure 20. HAZUS Transects ....................................................................................................... 58
Figure 21. 1946 Minimum Inundation Depth ............................................................................... 60
Figure 22. 1946 Minimum Dollar Exposure ................................................................................. 61
Figure 23. 1946 Maximum Inundation Depth .............................................................................. 62
Figure 24. 1946 Maximum Dollar Exposure ................................................................................ 63
vii

Figure 25. 1960 Minimum Inundation Depth ............................................................................... 66
Figure 26. 1960 Minimum Dollar Exposure ................................................................................. 67
Figure 27. 1960 Maximum Inundation Depth .............................................................................. 68
Figure 28. 1960 Maximum Dollar Exposure ................................................................................ 69
Figure 29. 2016 Maximum Inundation Depth .............................................................................. 72
Figure 30. 2016 Maximum Dollar Exposure ................................................................................ 73
Figure 31. HAZUS Maximum Dollar Exposures ......................................................................... 74
Figure 32. 1946 Tsunami Dollar Exposures ................................................................................. 75
Figure 33. 1960 Tsunami Dollar Exposures ................................................................................. 76
Figure 34. Fudai Seawall .............................................................................................................. 80

viii

List of Tables
Table 1. Building Classifications .................................................................................................. 22
Table 2. Tsunami Loss by Event................................................................................................... 24
Table 3. Number of Residences Damaged by Event .................................................................... 25
Table 4. Model Building Types .................................................................................................... 29
Table 5. Debris Weight by Occupancy Class ............................................................................... 31
Table 6. Building Damage Scale .................................................................................................. 42
Table 7. Building Damage Matrix ................................................................................................ 42
Table 8. 1946 Dollar Exposures.................................................................................................... 54
Table 9. 1960 Dollar Exposures.................................................................................................... 57
Table 10. 1946 HAZUS Inundation Values .................................................................................. 59
Table 11. 1946 HAZUS Dollar Exposures. .................................................................................. 64
Table 12. 1960 HAZUS Inundation Values .................................................................................. 65
Table 13. 1960 HAZUS Dollar Exposures. .................................................................................. 70
Table 14. 2016 HAZUS Inundation Values .................................................................................. 71

ix

List of Abbreviations
AEBM Advanced Engineering Building Module
BLS Bureau of Labor Statistics
DART Deep-ocean Assessment and Reporting of Tsunami
DEM Digital Elevation Model
DHS Department of Homeland Security
FEMA Federal Emergency Management Agency
FIMA Federal Insurance and Mitigation Administration
GBS General Building Stock
GIS Geographic Information System
HAZUS Hazards United States
IPSC Institute for the Protection and Security of the Citizen
JISAO Joint Institute for the Study of the Atmosphere and Ocean
JRC Joint Research Centre
MOST Method of Splitting Tsunami Model
NCTR NOAA Center for Tsunami Research
NED National Elevation Dataset
NOAA National Oceanic and Atmospheric Administration
OAR Office of Oceanic and Atmospheric Research
SCHEMA Scenarios for Hazard-induced Emergencies Management
SWEL Stillwater Flood Elevation
USACE The United States Army Corps of Engineers
USGS United States Geological Survey
USC University of Southern California
x

Abstract
The city of Hilo, Hawaii is more vulnerable to tsunamis than any other location in the United
States. Due to the unique bathymetry, topography, and location relative to the Cascadia
Subduction Zone, in the future, Hilo could be struck by a large tsunami similar to the historic
1946 and 1960 events. The Cascadia Subduction Zone can produce a 9.5 M earthquake with the
potential of generating a tsunami with maximum wave heights of over 29 feet. Before
devastating economic loss occurs, it is imperative that such potential flood inundation and
consequent dollar exposure are understood. This study compares the Joint Research Centre’s
(JRC) Scenarios for Hazard-induced Emergencies Management (SCHEMA) flood model
implemented using ArcGIS with the Federal Emergency Management Agency’s (FEMA)
Hazards-United States (HAZUS) flood model to simulate the potential impact of a large-scale
tsunami on the city of Hilo. The SCHEMA and HAZUS models, the National Oceanic and
Atmospheric Administration (NOAA), and the State of Hawaii provided the spatial data required
to build the financial and structural inventory database for these analyses. Field measurements
recorded during the 1946 and 1960 tsunamis and corresponding historical inundation maps
provided input into the models. The results of this research suggest that although the SCHEMA
model has the benefit of being more customizable, the HAZUS inundation scenario can be
implemented with fewer input data and produce results comparable to historical damages. Future
work will involve refining the inundation scenarios to include more detailed input data such as
historical terrain (digital elevation models), field-verified updates to the structural inventory
database, and an increased number of predicted events based on wave height.

1

Chapter 1 Introduction
This research evaluated the economic impact that a large-scale tsunami inundation event caused
by an earthquake from the Pacific Ocean would have on the town of Hilo, Hawaii. Hilo has been
hit by several large tsunamis in the past 75 years, and the local government and has been taking
steps to mitigate the impact of future events (Pararas-Carayannis 1977). These tsunamis ranged
in height from 26 to 49 feet and were associated with the 1946 and 1960 tsunami events
respectively (Eaton, Richter, and Ault 1961). The devastating effects of one of these tsunamis are
clearly illustrated in Figures 1 and 2.

Figure 1. 1946 Hilo Tsunami (Source: University of Idaho). Hilo residents flee from the 1946
tsunami in Hilo.
2

Figure 2. Hilo after 1946 Tsunami (Source: NPR). This is the downtown area of Hilo after the
1946 tsunami impact.
In the context of this study, a tsunami is defined as a giant wave caused by earthquakes or
volcanic eruptions under the sea (National Ocean Service 2015). This project utilized data
obtained from historical tsunami events in 1946 and 1960 to estimate tsunami height as input in
modeling the potential economic damage to the current structural inventory of the city of Hilo.
The economic impact of a large-scale tsunami on Hilo was estimated by comparing two
commonly accepted methodologies to accomplish this goal. First, the European Commission’s
Joint Research Centre’s (JRC) Scenarios for Hazard-induced Emergencies Management
(SCHEMA) (Tinti et al. 2011) flood model was implemented in ArcGIS as part of this study.
Second, the Federal Emergency Management Agency’s (FEMA) Hazards-United States
3

(HAZUS) flood model was also used to simulate the potential impacts (FEMA 2015). Third, the
results obtained utilizing these two methodologies were then compared.
The background research for this project focused on the following topics: the bathymetry
of Hilo Bay, the impact that the bathymetry and topography of Hilo Bay has on the size of
tsunamis as they enter the bay, the size and origin of Hilo’s historical tsunamis, the town of
Hilo’s mitigation strategies for tsunamis, and the SCHEMA and HAZUS Flood Models to help
understand the potential physical damage to the built environment that may be destroyed by a
large-scale tsunami. The town of Hilo has learned first-hand what tsunamis can do and has taken
steps to prevent future financial loss to the structural inventory and loss of life (Pararas-
Carayannis 1977). Researching land use changes over the last 75 years and newly created
mitigation measures, such as stricter building codes, helped determine what Hilo has been doing
to mitigate tsunami impacts (AIA 2009). Hilo Bay is particularly vulnerable to tsunamis because
it reverberates tsunami waves leading to much larger events than any other location in Hawaii
(Palmer, Mulvihill, and Funasaki 1965). The U.S. Army Corps of Engineers (USACE) has
completed studies that examine the impact the bathymetry and topography of the bay have on
approaching tsunamis. In 1946, an Alaskan tsunami impacted Hilo with wave heights as high
three story buildings (Davidson 2004). Then again in 1960, a Chilean tsunami hit Hawaii with an
average height of 6 to 16 feet, but Hilo Bay was impacted by a tsunami height of approximately
33 to 49 feet (Pararas-Carayannis 1977). At the time, the local bathymetry was blamed for the
increased localized tsunami heights in Hilo, as the submarine ridge formations outside of the bay
refract tsunami waves into the bay (Palmer, Mulvihill, and Funasaki 1965). The data from these
previous tsunamis were utilized to see if either SCHEMA or HAZUS can duplicate the historical
4

published cost of damages to the built environment. Thus, the goal of this study was to address
the following research questions:
1) Can the SCHEMA and HAZUS models be used to estimate damage to the built
environment of historical tsunami events in the city of Hilo, Hawaii?
2) Are the JRC SCHEMA methodology results in calculating potential tsunami inundation
damage to the built environment similar or different compared to results obtained using
the HAZUS coastal flood model for a given event?
It was anticipated that the damages predicted using the SCHEMA Flood Model
methodology will be similar to or fall within a maximum and minimum range of damages
predicted using HAZUS Flood Model program. This assumption was based on a preliminary
review of the models and input data required.
To qualitatively address these research questions, historical tsunami events were modeled
to determine the economic loss and dollar exposure that could occur today under similar
circumstances. In the context of this thesis, dollar exposure is defined as the total economic loss
in US dollars from a flooding model (FEMA 2015). By comparing the results of these models to
real-world economic losses caused by historical tsunami events, future improvements to these
models can be suggested after learning some of the strengths and weaknesses of each model.
1.1 Motivation
This research is important to not only evaluate the potential damage to the structural
inventory that the town of Hilo may experience due to a large tsunami event but also to test two
commonly accepted methodologies that can be used to evaluate economic impacts for coastal
communities anywhere that the required spatial information is available. Hilo is an excellent
study area for this project due to its unique location and terrain, as well as the availability of
5

historical references that have been carefully cataloged that allow for a comparison, or
validation, of the two methodologies being tested with recorded damage data of past tsunami
events.
Due to Hawaii’s location in the middle of the Pacific Ocean, all of the islands are
extremely vulnerable to tsunamis (O’Sullivan 2015). Hilo, on the Big Island of Hawaii, is
nicknamed “The Tsunami Capital of the World” due to the shape of the bay that, as mentioned
previously, tends to magnify the height of tsunamis making the town more susceptible to
damage. In Hawaii, tsunamis have killed more people in the last hundred years than earthquakes,
volcanoes, hurricanes, brushfires, and floods combined (Natural Hazards Big Island 2015). This
research is also intended to encourage the study of historical tsunami events to mitigate future
tsunami impacts. Tsunamis, like inland flooding, tend to repeat themselves, but at times may not
occur for hundreds of years. Pre-disaster mitigation, including urban planning for tsunamis, are
the best tools for coastal communities to protect themselves from future economic damage and
loss of life.
The most common cause of tsunamis in the world is earthquakes generated in a
subduction zone, where the oceanic plate is being forced into the mantle (King 2016). The most
active faults in the world are commonly found in the Ring of Fire (Figure 3). The Ring of Fire is
the name given to the Pacific Rim due to the number of major earthquake faults surrounding the
region. The coastlines in this region are the most active seismic and volcanic areas on earth. The
faults have enormous potential to generate underwater earthquakes and potentially destructive
tsunamis (Xie, Nistor, and Murty 2011). Figure 3 displays the major volcanoes which line this
important tectonic subduction zone in the Pacific Ocean. Hawaii is centrally located in this
6

region, which makes it vulnerable to potential earthquake tsunamis from anywhere within the
Ring of Fire (Palmer, Mulvihill, and Funasaki 1965).

Figure 3. Ring of Fire (Source: World Atlas 2015). The Hawaiian Islands are located just below
the label “Ring of Fire” in the middle of this map.
The Cascadia subduction zone, located off the coast of Oregon, is estimated to be capable
of producing an earthquake of up to 9.5 M, which would be similar to the Chilean earthquake
mentioned previously (Satake et al. 1996, Atwater, Yamagushi, and Satoko 2005). The Cascadia
Subduction Zone has not experienced a major earthquake in the last few centuries, so researchers
suggest that another major event is due sometime in the next 300 years (Peters, Jaffe, and
Gelfenbaum 2007). Some scientists even predict that an event may strike within the next fifty
years (Mazzotti and Adams 2004). Tsunami waves measuring 3 to 16 feet in height hit Japan as a
7

result of the 1700 AD Cascadia earthquake (Satake, Wang, and Atwater 2003). The coast of
Hawaii is several thousand miles closer to the Cascadia subduction zone than Japan, so the
impact of the same or similar sized tsunami would potentially be greater. Thus, a Cascadia
earthquake is likely the next big event to trigger a tsunami that would hit the town of Hilo and is
a reason to focus on historical measurements of large tsunami events in order to forecast the
economic impact on the town.
Detecting and tracking the size of tsunamis is important in order to be able to alert coastal
communities of the incoming threat in time for evacuation. The most common way to detect
tsunamis is by using ocean buoys to measure the sea level height, but these do not always
accurately forecast the potential approaching wave height to individual communities in a timely
manner. Since tsunami wave heights can be altered by nearshore bathymetry and coastal
topography, the buoy measurements will probably not provide a direct measurement of tsunami
energy for (Bernard 2005).
Mathematical formulas are a common way to forecast or explain tsunami wave heights,
but there are currently no globally significant formulas that can translate to other areas or events.
A mathematical formula that estimates tsunami size that could hit Japan related to a given
earthquake occurrence may not be appropriate for predicting a tsunami occurrence in Chile. It
has been common for modelers to use ad-hoc wave height amplification factors in their formulas
for run-up predictions when compared to actual observations (Synolakis et al. 2008). In the
context of this thesis, the run-up is defined as the maximum wave height of a tsunami.
Natural disasters are unpredictable and constantly surprise and change the paradigm that
scientists believe to be possible. In 1995, the National Oceanic and Atmospheric Administration
(NOAA) deployed six buoys as part of the Deep-Ocean Assessment and Reporting of Tsunamis
8

(DART) system (Greenslade and Titov 2008). NOAA realized immediately after the 2004 Indian
Ocean tsunami that six buoys gave insufficient coverage to provide adequate tsunami warnings
for all regions in the Pacific (Edward 2008). The DART II buoys were then created, and 31
additional buoys were installed to create a total of 37 Pacific Ocean buoys. The new buoys
include sensors that can detect a tsunami half an inch high in water four miles deep. These
upgraded buoys were developed to improve the timing and detection of tsunamis.
As evidenced, to evaluate if future disasters have been adequately mitigated and prepared
for, city planners and governments must learn from the past. Hilo has a well-documented history
of tsunami activity, and it has been determined by researchers that these events will continue to
happen into the future. The city of Hilo has taken steps in regards to building locations and types
of new construction allowed proximal to the coastline to limit the impacts of future tsunamis, but
no detailed study exists to determine if enough has been done (Bernard 2005). The goal of this
research is to determine if Hilo has taken all of the necessary steps to mitigate the economic
impact of tsunamis and the building related dollar exposure associated with a devastating
tsunami event. This research is also aimed at determining which inundation methodology is most
appropriate for accurately estimating the physical damage to the built environment incurred
should a significant tsunami inundation event occur in the future.

9

1.2 Background
The previous section discussed the dangers that Hilo could face from future tsunamis and
the likelihood that tsunamis may impact Hilo. This section contains background information
related to the studies of local topography and bathymetry, historical tsunamis, and NOAA’s
model of tsunamis in Hilo.
1.2.1. Local Topography and Bathymetry
Studying local topography and bathymetry explains why Hilo has often been impacted by
tsunamis with a higher average wave height than any other location in Hawaii. USACE
conducted a study in 1965 to ascertain why the tsunamis that had struck Hilo in 1946 and 1960
were more destructive due to higher wave heights than those affecting other regions of Hawaii. It
was well known that Hilo’s location on Hawaii is shaped like a funnel, causing incoming
tsunami waves to pile on top of each other to intensify the impact on the city regarding wave run-
up height. However, it was not previously explained why parts of Hilo experienced waves ten to
fifteen feet higher than wave height measured out in the bay. Figure 4 displays the reflecting of
waves off of the cliffs north of Hilo (Palmer, Mulvihill, and Funasaki 1965).
10

Figure 4. Reflecting of Waves in Hilo Bay (Source: Palmer, Mulvihill, Funasaki 1965).
After the 1960 tsunami, which caused wave heights of over 30 feet, the USACE built an
analog, physical model of Hilo Bay composed of concrete, approximately 63 feet by 96 feet
(Palmer, Mulvihill, and Funasaki 1965). The test was designed to investigate the role that
topography and bathymetry have on tsunami heights in Hilo Bay. The model used a pneumatic
11

type wave generator that created a solitary wave in order to reproduce a tsunami in Hilo harbor.
The test was aimed at determining the high water marks, limits of inundation, and the time
history of the wave locations, or marigrams.
After completion of the physical inundation test, the USACE and other participating
scientists found that the physical features of the surrounding topography, specifically the steep
cliffs that were found north of Hilo, reflected waves back towards Hilo harbor, which then
impacted the original incident wave and combined to create a larger, more dangerous wave.
Figure 5 illustrates the steep elevation found north of town and the comparatively lower
elevation Hilo harbor. In the USACE test, the wave solitary wave generated was amplified when
in bounced of the steep cliffs resulting in combined wave heights 55 to 150 percent higher than
the original test wave funneled into the harbor model. Additionally, two other factors found to
increase wave heights included the triangular configuration of the bay and submarine bathymetry
which reflects waves towards the bay.
12

Figure 5. Hilo Topography
1.2.2. Historical Tsunamis
As previously stated, the 1960 tsunami that originated in Chile, and the 1946 tsunami that
originated in Alaska, were the two most devastating tsunamis to hit Hilo in the last one hundred
years. Tsunamis that originate from earthquakes are commonly seen in Hilo, but the heights of
these two tsunamis were more substantial than those that have previously occurred.
An Alaskan earthquake would not be expected to cause a large tsunami that would strike
Hilo, but the 7.5 M Alaska Earthquake that resulted in the 1946 tsunami was said to have
reached over 45 feet on landfall in Hilo Harbor (Palmer, Mulvihill, Funasaki 1965). The third of
seven waves that comprised the tsunami event was the highest and extended half a mile inland to
inundate an area of 0.4 square miles
13

The financial impact on Hilo was $26 million in damage, while the personal loss that day
was 173. This historical dollar exposure is approximately equivalent to $318 million today,
based on the Bureau of Labor Statistics (BLS) inflation calculator (BLS 2016). Most of the
downtown area was demolished by the tsunami with 488 buildings destroyed and 936 additional
buildings damaged. The waterfront was washed out, and the breakwater was badly damaged, and
the main pier was completed destroyed. This tsunami prompted the establishment of NOAA’s
first tsunami warning center in Ewa Beach (Pararas-Carayannis 1977).
In the context of this study, a breakwater is defined as a structure built offshore that helps
to reduce the intensity of waves and prevent coastal erosion (Miller 2011). Hilo had a breakwater
constructed in 1929 that was believed to have helped to mitigate the 1946 tsunami’s effects by
absorbing some of its energy. It was partly destroyed in 1946, then rebuilt. The breakwater was
tested a second time in the 1960 event and again survived with only minor repairs required.
In the context of this study, a seawall is defined as an onshore structure that is designed to
prevent tides, waves, and extraordinary oceanic events from impacting coastal communities
(Miller 2011). After the 1960 tsunami, the USACE suggested building a 22-foot high seawall
that would run from the Wailuku River, just west of downtown Hilo, to the Waiakea Peninsula.
This seawall was never built. It is assumed that it would have been overtopped by a tsunami of
similar in height to the 1946 or 1960 events.
Instead of building a seawall, an elevated highway was constructed along Hilo Bay which
protects the downtown area from storm surges and moderate tsunamis (Miller 2011). This
construction was chosen over a large seawall due to residents not wanting their ocean views
obstructed and engineers being unable to guarantee that a seawall would protect the city in the
event of another large tsunami.
14

Southern Chile was impacted on May 22, 1960, by a 9.5 M earthquake that created
substantial damage in Chile and resulted in a tsunami that even impacted distant Pacific coastal
areas, like Hilo. The tsunami crested at nearly 35 feet on landfall in some places (Britannica
Academic 2015). Outside of Hilo Bay, wave heights on the island of Hawaii were lower, with
heights ranging between 3 and 17 feet. Figure 6 displays the difference in tsunami heights
throughout the island of Hawaii (Eaton, Richter, and Ault 1961). The tsunami generated $23
million in damage, equivalent to $281 million today. The personal loss for the town was 61 lives
lost. This tsunami devastated structures in downtown Hilo, where a total of 537 buildings were
destroyed (Pararas-Carayannis 1977). Nearly 0.94 square miles were inundated, and half of this
impacted area saw near complete destruction of all man-made structures (Eaton, Richter, and
Ault 1961).
15

Figure 6. Tsunami Heights in 1960. (Source: Eaton, Richter, and Ault 1961). Tsunami heights
are in feet.
These two tsunamis, 1946 and 1960, have been thoroughly studied and examined by
engineers and scientists (Shepherd, Macdonald, and Cox 1949; USGS Hawaiian Volcano
Observatory 2015; Davidson 2004). Thus, these observations were compiled to compare with the
financial impact that similar events might have on Hilo today. There were several other tsunamis
over the last one hundred years that impacted Hilo, but not to the extent of the 1946 and 1960
tsunamis. For example, in 1952, a strong earthquake originated near the southeastern coast of
Kamchatka with a moment magnitude of 9.0. Despite the high moment magnitude, the waves
16

that entered Hilo Bay only reached approximately 12 feet (Macdonald and Wentworth 1954).
There was only minor damage reported to the main pier with some cargo floating in the bay.
During the early morning hours of March 9, 1957, a 9.3 M earthquake struck the Aleutian
trench in the Aleutian Islands (USC Tsunami Research Group 2002). The 1957 earthquake did
not generate massive wave heights in Hilo like the 1946 earthquake from Alaska. The 1957 event
was several times stronger than the 1946 earthquake, which proves that earthquake magnitude
does not always correspond to higher tsunami wave heights, as there are many other factors that
can influence wave energy and height. These factors include distance from the epicenter and
direction of earthquake wave propagation resulting from earthquake fault orientation and
hypocenter depth and location (point within the earth where the rupture starts). The highest
tsunami wave heights measured in Hilo Bay were approximately 10 feet, which resulted in only
several $100 thousand worth of damage.
Another example is the 1964 Good Friday earthquake that occurred in Alaska, a 9.2 M
earthquake that struck southeast of Anchorage (USC Tsunami Research Group 2011). Despite
the large earthquake magnitude, due to the orientation of the generating fault, the resulting
tsunami wave heights were below 13 feet and caused minor damage.
Locally generated tsunamis are rare in Hawaii, but in 1975, a 7.2 M earthquake
originating just south of the Kilauea volcano, struck the Big Island of Hawaii. This earthquake
caused a tsunami that struck Hawaii within minutes, and approximately 20 minutes later for Hilo.
The measured waves in Hilo Bay were around 8 feet and caused damage to boats and several
docks. The most dangerous aspect of this tsunami was the reduced warning time due to the
earthquake occurring nearby. Typically lives can be saved with just a few hours of warning and
17

subsequent evacuations, but this relatively mild tsunami killed two people on the Big Island due
to brief advance warning (USGS Hawaiian Volcano Observatory 1998).
In 2011, Japan was struck by a 9.0 M earthquake near Tohoku (Cheung, Bai, and
Yamazaki 2013). Many photographs and videos have shown the devastation caused by tsunamis
throughout Japan, and after this particular earthquake, the waves that entered Hilo Bay measured
approximately 6.9 feet near the harbor, yet did not create any noticeable damage in Hilo.
1.2.3. NOAA Tsunami Forecast
The National Oceanic and Atmospheric Administration (NOAA) produces tsunami
forecast reports from their NOAA Center for Tsunami Research (NCTR) for areas of high risk
within the United States. Hilo’s history of tsunami inundation, population density, transportation
infrastructure, and year-round tourism makes it crucial to develop a forecasting report for
tsunami inundation. A tsunami forecast report for Hilo was generated in 2010 in cooperation
with the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) and the Office of
Oceanic and Atmospheric Research (OAR) (Tang, Titov, and Chamberlin 2010).
The methodology for these tsunami forecast reports involves compiling deep-ocean
tsunami measurements from the Deep-ocean Assessment and Report of Tsunami (DART) buoy
system combined with the Method of Splitting Tsunami (MOST) prediction model. This forecast
is a two-step process which starts with collecting pre-computed tsunami measurements at DART
buoy locations combined with coastal predictions by running the high-resolution MOST model.
It was found that tsunami wave amplitudes increase dramatically due to shoaling in near-shore
areas shallower than 65 feet. Thus the need for accurate near-shore modeling emphasizes the
importance of high-resolution flood models (Tang, Titov, and Chamberlin 2010).
18

The 2010 NOAA tsunami forecast model runs many different scenarios to determine the
areas most at risk depending on the tsunami’s origin and magnitude. Maximum error from the
model was determined to be within 35 percent when the observations are greater than 1.6 feet. A
total of 1435 scenarios were simulated based on earthquake magnitudes of 7.5, 8.2, 8.7, and 9.3
M. The results indicated a nonlinear relationship between offshore and nearshore wave
amplitudes. This means that there is not always a clear correlation between an earthquake’s
magnitude and buoy wave measurements. Buoy measurements varied in relation to the shoreline
measurements, and the seismic magnitude cannot be used to accurately predict maximum wave
amplitudes. This result proved that a high magnitude earthquake does not always create
destructive tsunamis that travel across an ocean. For example, as previously stated, the 9.0 M
Tohoku earthquake in Japan caused only localized damage and mild inundation across the state
of Hawaii. This was a big earthquake but as mentioned above, a big earthquake magnitude does
not always create a large tsunami (Cheung, Bai, and Yamazaki 2013). Results from this model
determined that local bathymetry and topography play an important role in the formation of
localized tsunami waves within Hilo Bay (Tang, Titov, and Chamberlin 2010).
Verification of this 2010 NOAA tsunami forecast model was completed by comparing
the forecasting model with 16 historical tsunamis. The historical tsunamis’ run-up boundaries
were digitized while the most recent ones consisted of digital tide gauge data. The forecasting
model was found to produce an error of only ±1.74 feet in the most recent 2010 tsunami
generated by an 8.8. M earthquake in Chile. This was also the highest error found in any of the
comparisons.
Chapter 1 has provided a detailed description of the study area, background information
on tsunamis, and why tsunamis are an important aspect of Hilo, Hawaii. The remainder of this
19

thesis is divided into four chapters. Chapter 2 describes related works, relevant Geographic
Information System (GIS) models, and a tsunami study in great detail. Chapter 3 details the data
required for each model, the SCHEMA methodology developed as part of this study, and the
HAZUS methodology also performed as part of this thesis. Both the SCHEMA and HAZUS
damage matrixes are discussed in detail to provide oversight on how the dollar exposure is
measured. Chapter 4 reports the findings for each model scenario and a discussion about the
other model outputs. The accuracy of each model is described in relation to the historical tsunami
events. Chapter 5 points out the positives and negatives with each model, as well as the answer
to the two research questions. The potential future work on this research is also discussed.

20

Chapter 2 Related Works
To determine the most effective spatial modeling method using Esri ArcGIS, a review of three
studies was conducted. The first study review focused on an emergency management projects
dealing with tsunamis and the potential damage to buildings (Tinti et al. 2011). This project
methodology indicated that building damage can be modeled within a spatial computing program
when elevation values are available for the bathymetry, land surface, buildings and wave
inundation depths. The second relevant study described a theoretical tsunami event hitting
Seaside, Oregon and the damage predicted by such an event. That study used predetermined
inundation values that were imported into FEMA’s HAZUS program in order to perform
additional modeling (Schneider 2011). A third review included the HAZUS Flood Model
technical and user manuals from the Department of Homeland Security (DHS) and FEMA. This
article details the capabilities of the HAZUS Flood Model as well as the inputs required and the
potential outputs that can be produced by using this model to estimate structural damage and loss
of life due to a specified tsunami event (FEMA 2015).
2.1 JRC SCHEMA Methodology
In 2011 the Joint Research Centre (JRC) and the Institute for the Protection and Security
of the Citizen (IPSC) released the Scenarios for Hazard-induced Emergencies Management
(SCHEMA) project (Tinti et al. 2011). A total of 39 months of effort contributed to the
SCHEMA project to document and study tsunami scenarios and their potential damages.
According to Tinti et al. (2011), this is an important methodology to consider for assessing
tsunami damage, including building classifications and a damage matrix to assess the impact of a
tsunami based on the building class. This methodology is completely independent from FEMA’s
HAZUS program.
21

Utilizing SCHEMA, the level of inundation fluctuates throughout the inundated areas, so
the type of building on each parcel needs to be classified to determine the impact on each parcel
based on the height of the tsunami wave. The vulnerability of buildings is determined based on
factors such as the resistance of structures, proximity to shoreline, wave heights reaching
buildings, and the surrounding topography. To calculate building damage, a well-built SCHEMA
project requires a standardized building topology, a standardized economic damage scale based
on historical flood data, a damage function for each building related to water depth, and an
inventory of buildings. The Hilo SCHEMA modeling effort utilized all four factors and is thus
considered a good starting point for project methodology. Buildings are classified into seven
different letter grades based on their building types. A detailed listing of each class, building
type, and height can be found in Table 1 (Tinti et al. 2011).

22

Table 1. Building Classifications (Tinti et al. 2011)
Class Building Types Height or Stories
Light
A1 Beach or sea front light constructions of wood, timber, clay 0-1 level
A2
Very light constructions without any design. Very rudimentary
huts, built using wood or clay, timber, slabs of zinc
1 level
Masonry, and not
reinforced concrete
B1 Brick not reinforced, cement, mortar wall, fieldstone, masonry 1-2 levels
B2
Light and very concentrated constructions: wooden, timber and
clay materials
1-2 levels
C1
Individual buildings, villas: Brick with reinforced column &
masonry filling
1-2 levels
C2
Masonry constructions made of lava stones blocks, usually
squared-off, alternating with clay bricks
1-2 levels
D
Large villas or collective buildings, residential or commercial
buildings: Concrete not reinforced
1-3 levels
Reinforced
Concrete
E1
Residential or collective structures or offices, car parks, schools:
reinforced concrete, steel frame
0-3 levels
E2
Residential or collective structures or offices, car parks, schools,
towers: reinforced concrete, steel frame
3 levels
Other
F
Harbor and industrial buildings, hangars: reinforced concrete,
steel frames
Undifferentiated
G Other, administrative, historical, religion buildings Undifferentiated

In this study, the values in this table were used for the building classification process for
Hilo structures, to determine the structure on each parcel and the possible vulnerability to
tsunami wave heights. The elevation of each parcel was calculated and then subtracted from the
sea-level flood depth raster to determine the height of inundation. The building classification and
the height of the inundation are the only two values needed to determine the damage assessment
for each parcel. Additional detailed information about the SCHEMA methodology can be found
in Tinti et al. (2011).

23

2.2 Tsunami Modeling For Seaside, Oregon
The NOAA Tsunami Research Center developed a 500-year tsunami flood depth grid for
the town of Seaside, Oregon as a test trial for a tsunami workshop held in 2010 (Schneider
2011). This 500-year tsunami grid estimates the potential damages that could result from a
tsunami event with a 0.2 percent likelihood of impacting the Oregon coast. The 500-year tsunami
grid considers the worst case scenario for a tsunami, unlike the 500-year coastal flood grids that
only estimate floods without considering a potential tsunami. In the context of this study, a
coastal flood is defined as an oceanic event, such as a storm surge, tsunami, or sea level rise, that
causes a coastal area to become inundated with water for a period of time. As a part of
Schneider’s study, this tsunami flood depth grid was imported into the 2009 version of FEMA’s
HAZUS program for modeling purposes.
HAZUS is an effective tool to use in flood modeling because it evaluates many potential
direct and indirect values that need to be considered in a comprehensive model. HAZUS offers
information on buildings, infrastructure, population, hazards, economics, and other
environmental factors (FEMA 2015). Despite this wide variety of default HAZUS datasets, there
are some issues when trying to model a tsunami event within this system. HAZUS utilizes a
well-documented flood model but does not have a tsunami model integrated within the program.
HAZUS also does not incorporate velocity damage functions within the coastal flood
calculations, despite having such a methodology having been developed in the riverine model
(FEMA 2015). Velocity damage is a notable variable for estimating physical damage inflicted by
a tsunami, but it is a very complicated variable to model for ocean waves (Schneider 2011).
Schneider (2011) ran the flood model within HAZUS to evaluate the effectiveness of
modeling a tsunami using the 500-year tsunami flood depth grid, and to determine the
24

differences between using this custom 500-year tsunami grid and the 500-year coastal flood grid.
The initial observation of the author is that the 500-year tsunami grid produced a much more
significant flood in the region. The high water mark of the 500-year coastal grid was 11.1 feet,
while the 500-year tsunami grid produced a high water mark of 32.8 feet. This difference is
significant in terms of allowing local planners to strategize for a tsunami event with appropriate
data. A 15 feet seawall could be enough for a coastal flood but would easily be overtopped by
this estimated tsunami. As a comparison, a 100-year coastal flood was also modeled, and this
event only found a potentially high water mark of 6.5 feet (Schneider 2011).
Schneider (2011) found that residential dollar loss, which is the replacement value and
not the actual cash value of property, is another useful variable to compare the differences
between the two different flood events. The total residential exposure for the area is constant at
just over $1.1 billion. The residential loss for the 500-year tsunami was 2.4 times greater than the
500-year coastal flood, and 5.9 times greater than the 100-year coastal flood, as shown in Table
2.
Table 2. Tsunami Loss by Event (Schneider 2011)

The residential building damage assessment is another variable that is calculated in
HAZUS through a different methodology from the SCHEMA methodology. In HAZUS, the
building damage is computed with more damage tiers than SCHEMA, and with damage
calculated at each foot increment in inundation. In the context of this study, a damage tier is a
25

group of flood depths that create dollar exposures to buildings depending on the tier that they fall
within. In the 500-year tsunami example, it was determined that 0.5% of residences were
substantially damaged by the flooding, and no buildings escaped the inundation within the study
area. Whereas in the 500-year coastal flood a total of 22.7% of residences experienced no
damage, and in the 100-year coastal flood, 56.4% of residences were not damaged. Table 3
displays the breakdown between the damage increments and the three floods that were modeled
(Schneider 2011).
Table 3. Number of Residences Damaged by Event (Schneider 2011)

Schneider (2011) stated that the analysis in HAZUS only uses still-water damage
functions in coastal flooding and that the velocity variable is missing from the coastal flood
model. Without modeling the flow and velocity of the water, it must be noted that the damage
and loss are most likely underestimated in any tsunami estimation process. This is an important
variable that is missing in any analysis created in either ArcGIS using the SCHEMA
methodology, or using HAZUS. Velocity modeling would need to be added to future HAZUS
system models to more accurately assess loss in a given tsunami event (Schneider 2011).
Tsunami models that can utilize the NOAA tsunami flood grids can be developed for any
coastal area that has a tsunami flood grid generated for that location. These tsunami models
should be developed to support coastal planning and prevention strategies to potentially lessen
26

economic and loss of life. To create more accurate tsunami models, a wide range of variables
need to be incorporated into the analysis. Some of these variables include local bathymetry,
topography, debris assessments, and damage estimates for vehicles. Some of these components
are currently extremely difficult to model without knowledge of such real-time physical data of a
given location (Schneider 2011).
2.3 HAZUS Flood Technical Manual
FEMA’s HAZUS is a complex multi-hazard methodology modeling program used to
determine the impacts of floods, wind, and earthquake losses (FEMA 2015). The HAZUS
program is capable of producing real-time loss estimates directly following a real life event.
HAZUS has the capability to receive user-supplied input data to create a refined loss estimation
model. The flood model is in constant development and enhancement to provide floodplain
managers and other users with the ability to protect citizens and property from floods. The
HAZUS system was developed not only to estimate post-disaster losses, but also to provide an
analytical support tool so communities and individuals can make informed decisions (FEMA
2015).
The HAZUS program can create many outputs, so the focus of the flood model is greatly
determined by the expertise of the user. The flood model was created with a simple user interface
and minimal input data required, though the user has the option of adding personal data and
settings to customize the program for individual studies. HAZUS requires users to have the
proper version of ArcGIS installed with the spatial analyst extension provided. The user also
needs to supply a DEM since the coastal flooding model requires elevation data to determine
impacted areas. Users can acquire this DEM from the National Elevation Dataset (NED) within
the HAZUS interface, or provide their own DEM. With the elevation data provided users are
27

capable of determining their damage and loss requirements based on the specifics of a
customized flood model. Figure 7 displays the flowchart for the HAZUS Flood Model schematic
(FEMA 2015).

Figure 7. HAZUS Flood Model Schematic (FEMA 2015)
Building damage is determined through the HAZUS Flood Model by evaluating the
General Building Stock (GBS) default dataset provided within the program. The GBS contains
residential, commercial, industrial, and assorted infrastructure building data. Damage to these
buildings is determined based on a percent of impact from the inundation of flood waters. One
28

major difference compared to the SCHEMA methodology is that the GBS is aggregated for
census blocks, and each block is assumed to have an even distribution of building statistics
within it. Therefore, it is assumed that the SCHEMA methodology is better for evaluating
smaller study areas since individual buildings can be identified, whereas the GBS data
encompasses a significant amount of building inventory, by default.
In HAZUS, the building classification types and inundation level determines the
inundation damage (FEMA 2015). Specific building characteristics are required for an advanced
flood model to have accurate loss estimates. The GBS models for structural damage are not only
based on the inundation level but also on a wide range of variables. The building age is an
important variable that is used in the HAZUS model. While building structure type is important,
the age plays a significant role in estimating how vulnerable a building may be due to building
styles of a given time period. Foundation style and building materials are also found in the GBS
dataset to help determine the potential for debris from that structure. Buildings with weak
foundations or built on slabs are more likely to break away in a heavy flood and impact other
buildings. Lastly, the building model type and structure material are crucial to estimating the
building damage. Each building type has different components within the model that can
potentially increase debris, or prevent total destruction due to resistance to flooding. Table 4
displays the typical building types, and characteristics found within HAZUS (FEMA 2015).

29

Table 4. Model Building Types (FEMA 2015)

The HAZUS model determines direct physical damage to the GBS based on the
foundation type, inundation vulnerability level, and estimated water depths for a given area. The
vulnerability level is defined as the grouping of building characteristics for structures that
determines how likely they may be impacted by floods of various depths. These three variables
allow HAZUS to estimate the damage throughout a census block as previously mentioned.
However, the velocity variable is missing from these calculations for floods. Velocity has a huge
impact on tsunami models, and it is currently only mentioned in the methodology but not coded
into the HAZUS loss estimation calculations (FEMA 2015).
In the HAZUS Flood Model, the inundation vulnerability and water depth estimates are
used in a damage curve to find the percent of damage for each structure type within a given
census block, illustrated in Figure 8. FEMA Building Flood Depth-Damage Curve (FEMA
2015). These damage curves exist for many different types of structures, foundations, and
building material types. Each curve is customized based on the vulnerability of the building
30

characteristics and flood inundation. These curves are created also using information provided by
USACE districts, as well as the Federal Insurance and Mitigation Administration (FIMA).
FEMA initially created these damage curves for use in the actuarial rate settling process, but they
are also useful in estimating damage in a flood model. The curves were first generated in the
1970’s then were improved over time As additional damage information is collected, the curves
can be fine-tuned and estimate potential damage with increased accuracy. Figure 8. FEMA
Building Flood Depth-Damage Curve (FEMA 2015)demonstrates an example of what the
damage curve looks like (FEMA 2015).

Figure 8. FEMA Building Flood Depth-Damage Curve (FEMA 2015)
A major limitation of HAZUS is the lack of an environmental debris calculation within
the flood model. Debris disposal can be another costly expense after a flood, on top of the
economic loss from damaged buildings. But while the HAZUS flood debris calculation focuses
on building-related debris, it does not factor in environmental debris such as vegetation and
31

sediment. This flood model debris calculation is built similarly to the earthquake debris
calculation that determines the structural components and foundation materials for each
structure, while excluding vegetation and sediment. However, the flood-related debris model
focuses on the internal components of buildings, unlike the earthquake model that takes into
account both structural and internal components. This debris estimate is one important part of the
HAZUS flood methodology that is not currently available with the SCHEMA methodology.
Table 5 shows an example of the debris weight values based on each building type, occupancy,
and example depth of flooding (FEMA 2015).
Table 5. Debris Weight by Occupancy Class (FEMA 2015)

Even though there are some limitations, the outputs of HAZUS are very comprehensive.
Each scenario can generate a wide variety of data which may or may not be helpful to the user.
The flood model specifically generates statistics for the direct damage to buildings, essential
facilities, transportation and utility systems, and damage estimates for debris generation. Indirect
32

losses are also included with a focus on supply shortages, sales declines, opportunity costs, and
economic losses (FEMA 2015).
Overall the HAZUS flood methodology is very diverse and a great example of how a
wide variety of variables can be factored in to provide reasonably accurate estimates of damage
and economic loss. In this study, the HAZUS model was used to model the two historical
tsunamis that hit Hilo and the results were compared to each historical tsunami model using the
JRC’s SCHEMA methodology for building inundation damage sustained by a tsunami.
Schneider’s study in Oregon described some helpful ways to incorporate data from HAZUS into
ArcGIS and vice-versa. This technique is also employed by processing data in ArcGIS and using
it within the HAZUS Flood Model.

33

Chapter 3 Methodology
The SCHEMA methodology and the HAZUS Flood Model are comprehensive resources for
creating a spatial tsunami inundation flood model. Both have their strengths and weaknesses, so
this research was aimed at determining which methodology is more suitable for determining
physical economic loss from tsunami inundation in Hilo, Hawaii.
3.1 Data
The spatial data that was needed for this research is from the SCHEMA methodology,
HAZUS, the United States Geological Survey, and the State of Hawaii’s GIS Program. The
essential spatial files for the economic portion of the SCHEMA model are parcels downloaded
from the State of Hawaii (State of Hawaii 2015). The parcels were used to evaluate the economic
damage and inundation vulnerability associated with a given tsunami event. The necessary
building stock required for HAZUS are already included in the HAZUS package. This building
stock information was last updated in 2014 and is the only required economic spatial data used in
the HAZUS model.
In this study, the default inventory of building stock provided by HAZUS was used for
modeling purposes. Also, HAZUS provides advanced structural inventory modeling through the
Advanced Engineering Building Module (AEBM) which includes building-specific damage and
loss analysis (FEMA 2014). For the purposes of this study, it was determined that the AEBM
dataset was not needed since the default HAZUS building stock inventory is recent, last updated
in 2014. In the city of Hilo, damage and loss functions for generic building types are considered
to be reliable predictors of economic loss for large groups of buildings present today.
The parcel layer, as well as many other files used in the SCHEMA model, were found on
the State of Hawaii website. This parcel layer appears to accurately overlay with the aerial and
34

elevation raster of Hilo. This is important because this data must accurately overlap with the
digital elevation model (DEM) used in the analysis. This parcel layer contains not only the land
monetary value but also the building monetary values as well. For the purpose of estimating the
potential economic damage of a given tsunami, damage to the built environment was the main
focus of this study. Each parcel was manually classified into a building category based on the
construction methods and materials of the largest building on each parcel. This is required
because the SCHEMA damage matrix sets different levels of estimated damage depending on the
type of building construction and materials. This information comes from the County of
Hawaii’s Real Property Tax Assessment website that provides property taxes each year, recently
updated in 2015. Perhaps due to the variability of tax assessments, the exact day and month of
the most recent tax assessment was created for each parcel is not provided, so for consistency, it
is assumed that all values were updated in 2015.
Light Detection and Ranging (LIDAR) data is a remote sensing technology that collects
3-dimensional points of the Earth’s surface (USGS 2015). LIDAR is a helpful tool to estimate
the height of buildings by comparing LIDAR data against a DEM that was created around the
same time period as the LIDAR data was taken. Due to the SCHEMA models using historical
elevation data, it was not ideal to use LiDAR data that would be created from a time frame
different from the historical events. Also due to known terrain changes between 1946 and 1960,
the accuracy of the data would be questionable.
Non-economic input data required for the SCHEMA model included historical
topographic maps, historical wave heights, historical inundation maps, and the SCHEMA
Damage Matrix (FEMA 2015; NOAA 2015; State of Hawaii 2015; Shepherd, Macdonald, and
Cox 1949; USGS Hawaiian Volcano Observatory 2015). Historical documents were used to
35

ensure that each model was using input data from the same time period as the events. The
landscape and coastal area around Hilo have changed over the last century, so it was imperative
that historical elevation measurements were used. The historical USGS topographic maps were
manually digitized as part of this study, then converted into historical DEM files for use in the
1946 and 1960 tsunami event tests. A 1943 topographic map was used for modeling the 1946
event, while a topographic map dated 1963 was used to model the 1960 event.
Historical tsunami inundation maps were previously created after each historical tsunami
event and used as figures in websites and journal articles (Shepherd, Macdonald, and Cox 1949).
As part of this study, these historical tsunami inundation maps were manually digitized for the
1946 and 1960 events to create historical inundation polylines. The inundation polylines
represent the inundation boundary for the study area. Each historical map was georeferenced to
ensure that it was in the correct location using the coastline and major intersections as reference
points. Figure 9 displays the historical inundation map for the 1946 tsunami event.

Figure 9. Hilo 1946 Tsunami Inundation (Shepherd, Macdonald, and Cox 1949)
36

The 1946 historical inundation map extends to the city boundaries in both directions, so
additional information was not required to manually digitize the inundation polylines all the way
to the city boundary. Figure 10 displays the historical inundation map for the 1960 tsunami.

Figure 10. Hilo 1960 Tsunami Inundation (USGS Hawaiian Volcano Observatory 2015)
The 1960 historical inundation map does not extend to the city boundaries of Hilo in the
east. This was corrected by extrapolating the nearest point data from the historical tsunami wave
heights to estimate inundation to the city limits.
Historical tsunami wave height measurements were reported along the coastlines of
Hawaii after previous tsunamis (State of Hawaii 2015). Point spatial wave height data was
collected throughout Hilo Bay, and along the coastline to the east. This wave height data is
available from the State of Hawaii and was used with the elevation of the inundation boundaries
to create an accurate inundation raster file. This raster file was then used to calculate the
inundation of each parcel in order to determine the impact on buildings in each parcel.
37

There are fewer non-economic data required for the user to input into HAZUS due to the
comprehensive databases available with the program. The only required files not provided by
HAZUS were elevation data, inundation data, and the historical tsunami wave heights.
The same inundation data and historical tsunami wave heights files used in the SCHEMA
model calculations were used in the HAZUS model runs. These files were pre-processed before
using them as inputs in HAZUS. For the elevation data, both the 1943 and 1963 historical DEM
files were used to ensure that each model is using the same elevation data. Since HAZUS uses a
distinct wave height value to simulate flooding in coastal areas, it was also possible to create a
new scenario using the maximum values from each tsunami event. This new maximum event
was mapped using current elevation data to estimate potential maximum economic damage if a
large tsunami event were to occur in 2016. For this scenario, a recently updated DEM was
required. A 1/3 arc-second DEM from NOAA was available from NOAA’s Tsunami Inundation
DEM website (NOAA 2015). The DEM layer is considered a very accurate high-resolution
elevation model. The DEM file also contains bathymetric data that was helpful to visualize the
underwater terrain of Hilo Bay. This DEM was last updated in 2008.
The debris and turbidity of tsunami events were not measured in the SCHEMA model.
Inundation is the focus of calculating economic damage, and the SCHEMA methodology focuses
on structural damage caused by flooding from a tsunami. HAZUS has the capability of
estimating debris, but this information was ignored since it would create significant
discrepancies with the SCHEMA results. The additional debris information would give the
HAZUS model a significant advantage in estimating overall tsunami damage, even without
factoring in velocity, which the model is currently unable to do in the coastal flood model
(FEMA 2015).
38

3.2 Methodology
The methods used in this research are rooted in the research and data exploration stages
(Figure 11). After research was completed and the data was acquired and processed, the
economic impact model was built within ArcGIS according to the SCHEMA methodology, and
then the exiting flood model in HAZUS was utilized.
3.2.1. SCHEMA Model in ArcGIS
Each historical tsunami event first required an accurate elevation file to begin the
analysis. The historical topographic maps from 1943 and 1963 were georeferenced so the
contour lines could be digitized. Due to the 1943 topographic map having only 50 foot contour
lines, it was essential that the gaps were filled with the closed temporal elevation data. The 1963
20 foot contour lines were included with the 1943 contour lines to help create a more accurate
DEM. The contour lines generated were then used with the ArcGIS Topo to Raster tool to create
an elevation raster for each year.
The height of inundation is determined by finding the elevation of the historical tsunami
inundation polylines. After each historical tsunami polyline had been digitized, points were
constructed every 150 feet along the polylines to extract elevation values. These points were then
combined with the historical tsunami wave height values to create an inundation depth raster.
Next, a raster was created using the Empirical Bayesian Kriging method to ensure accurate
predictions throughout the study area (ESRI 2016). This inundation depth raster uses an assumed
elevation of sea-level so that it can be compared with the historical elevation and determine
specific levels of inundation on each parcel. Figure 11 provides a methodology flowchart
displaying the steps involved in running the SCHEMA Flood Model to produce tsunami
inundation loss estimates in the form of a vector dataset.
39

Figure 11. Overview of the SCHEMA Tsunami Flood Model. Green ovals are outputs, blue
squares are inputs, and yellow squares are tools or calculations.
The next step in this analysis was to calculate the inundation value for each parcel. First,
zonal statistics was used to find the mean elevation for each parcel using the historical DEM.
Next, zonal statistics was used again to attach the inundation elevation for each parcel from the
inundation depth raster file just created. The mean inundation elevation for each parcel was then
subtracted from the mean elevation to find the level of inundation for each parcel. The level of
inundation was used with the building classifications in Table 1 to find the potential dollar
40

exposure based on the SCHEMA methodology. To save time, only areas vulnerable to a tsunami
had their buildings classified. Figure 12 shows the tsunami evacuation zones generated with the
help of forecasting models and also historical tsunami measurements (Office for Costal
Management 2015). This tsunami evacuation zone forms the boundary for building
classifications.

Figure 12. Hilo Tsunami Evacuation Zones (Adapted from NOAA 2015).
The building classification process employed in this study was based on visual
confirmation using Bing Maps Bird’s Eye Imagery and Google Street View. Visual exploration
is the best way to assess building classification values when current documentation was not
available from current data sources. The visual confirmation was not foolproof, but the
construction methods appeared to be consistent throughout Hilo, and are generally easy to
41

identify through such brief visual inspection. Many blocks are made up of consistently
constructed buildings, and several blocks in from the ocean are now mostly residential buildings
that are universally categorized as B, light wooden or brick structures between one and two
stories high.
The SCHEMA methodology includes a building classification table, a damage scale, and
a damage matrix that determines the level of destruction for each building based on inundation.
The damage scale ranges from D0, no damage, all the way to D5, total collapse. Table 6 displays
the Building Damage Scale from the SCHEMA methodology.
The SCHEMA damage scale includes six different levels of damage that cover all
possible tsunami damage values. Note that D3, D4, and D5 all result in demolition and total
destruction of a building. Only values of D0, D1, and D2 prevent total economic loss for a given
parcel or structure. Once the buildings are classified and the inundation of each parcel is known,
the damage scale can be used in the building damage matrix to determine which damage level
each parcel falls into. Table 7 displays the building damage matrix and the corresponding values
of economic loss. The percentage of economic loss refers to the percentage of the repair costs in
relation to the total value of the structure.

42

Table 6. Building Damage Scale. Source: (Tinti et al. 2011)
Damage Scale
Damage Level Damage on Structure
Use as shelter /
post crisis use
Detection by Earth observation
D0 No damage No significant damage
Shelter /
immediate
occupancy
No sign of damage visible on
building and surrounding
environment. The absence of
damage cannot be proved only
through space imagery.
D1 Light damage
No structural damage - minor damage,
repairable: chipping of plaster, minor visible
cracking, damage to windows, doors.
Shelter /
immediate
occupancy
Barely visible
D2 Important
damage
Important damage, but no structural damage:
out-of-plane failure or collapse of parts of
wall sections or panels without
compromising structural integrity, leaving
foundations partly exposed.
Unsuitable for
immediate
occupancy, but
suitable after
repair
Damage on roof hardly visible.
Other damage not visible.
D3 Heavy
damage
Structural damage that could affect the
building stability: out-of-plane failure or
collapse of masonry, partial collapse of
floors, excessive scouring and collapse of
sections of structure due to settlement.
Evacuation /
Demolition
required since
unsuitable for
occupancy
Not or hardly visible if roofs
have not been removed
D4 Partial failure
Heavy damages compromising structural
integrity, partial collapse of the building
Evacuation /
Complete
demolition
required
Visible
D5 Collapse
Complete collapse: foundations and floor
slabs visible and exposed.
Evacuation Very Visible
Table 7. Building Damage Matrix. Source: (Tinti et al. 2011)
Building Class and Inundation Depths
Damage Level A B C D E Economic Loss
D0 No damage
0 0 0 0 0
0%
0 0 0 0 0
D1 Light damage
0 0 0 0 0
30%
5.9 6.6 8.2 6.6 9.8
D2 Important
damage
5.9 6.6 8.2 6.6 9.8
60%
7.2 9.8 13.1 14.8 19.7
D3 Heavy damage
7.2 9.8 13.1 14.8 19.7
100%
8.5 13.1 19.7 21.3 31.2
D4 Partial failure
8.5 13.1 19.7 21.3 31.2
100%
12.5 16.4 26.2 29.5 41.0
D5 Collapse >12.5 >16.4 >26.2 >29.5 >41.0 100%
43

The inundation depth value in the damage matrix (Figure 7) provides a set of inundation
ranges that define economic loss based on the building class and damage level. The heights for
each value are in feet, as is the DEM. Also, all calculations completed in this study were in feet.
The building classes are assigned a building class letter, while the number attached to each letter
is not used for calculating damage. There are also F and G values in the damage scale, but these
were not used in this study because of invulnerability to tsunami inundation or unquantifiable
financial values. Most buildings are residential homes and fell under the B classification. These
homes are vulnerable to complete destruction by inundation above 9.8 feet. A measurement of
9.8 feet is generally considered high enough to inundate a building’s first floor. Several homes in
Hilo are vulnerable to complete collapse or demolition from inundation starting at 19.7 feet in
height. A 19.7 foot wave is high enough to cover a two story structure in Hilo, and while that
may not destroy a well-built hotel, it is likely to cause severe structural integrity damage (Tinti et
al. 2011).
The economic loss column of Table 7 was created in this study to provide a quantitative
value for economic impact calculations. Damage levels D3, D4, and D5 all require full
demolition so all of a given parcel’s (building) value would be considered lost. The D2 damage
level mentions serious damage but does not require demolition due to the comparatively greater
integrity of the structure. Integrity is determined by the ability of a structure to be repaired
without worry of structural failure or collapse. Damage could vary depending on the structure.
In the context of this study D2 level damages were estimated at just over half the cost of the
home’s assessed value because most of the first floor would have been underwater and thus
severely damaged. D1, or light damage, varies as well, but can range anywhere from minor
cosmetic damage, to replacing the exterior and much of the interior of the structure. In this study
44

D1 level damage was estimated to be roughly a third of the value of a structure. Due to the
variety of building types and economic damage levels, the loss estimation step of the SCHEMA
model was computed manually in the attribute table of each respective tsunami event shapefile
using ArcMap’s default Calculation tool.
3.2.2. HAZUS Flood Model
The HAZUS Flood Model required less initial input data preparation because the
program is by default comprehensive. The inputs required for HAZUS to model a flood,
representing tsunami inundation, include a DEM and tsunami wave height calculations (FEMA
2015). The inundation can be created within the HAZUS program using the Stillwater Flood
Elevation (SWEL) setting with the tsunami wave height calculated values. The SWEL setting
determines the height of coastal flooding in comparison to sea-level elevation. This value
dictates the amount of flooding that HAZUS will create when it delineates the floodplain. As
previously stated, it was anticipated that the SCHEMA Flood Model damage results would fall
within the HAZUS ranges for a given year.
The initial steps in the HAZUS Flood Model involve the same pre-processing of data as
the SCHEMA Flood Model. The historical inundation map was manually digitized to create the
inundation boundary, while the historical topographic map was digitized to create the historical
contour lines (Shepherd, Macdonald, and Cox 1949; USGS Hawaiian Volcano Observatory
2015; USGS 1943; USGS 1963). These historical contour lines are presented in Chapter 4. This
data was input into the ArcMap Topo to Raster tool to generate two historical DEMs. Each
scenario, 1946 and 1960, had a historical inundation polyline and a historical DEM created for
analysis. The inundation polyline was used to construct points along the inundation boundary.
These points were used to create a depth raster through interpolation methods. The elevation of
45

inundation in each historical tsunami event was calculated by adding the elevation from the
historical DEMs to the constructed points along the inundation polyline.
The HAZUS Flood Model allows a user to input an inundation depth raster, which is a
raster file that determines the depth of flooding at each pixel location. The depth raster generated
as input to the SCHEMA flood model could also have been used as an input in HAZUS. Since
one research goal of this study was to compare both methodology’s flood models as well as the
building stock information in HAZUS versus the parcel information in SCHEMA, the SCHEMA
depth raster was not used as an input in HAZUS. Since the HAZUS analysis did not utilize a
depth raster, the SWEL setting within HAZUS was used to flood the project area of Hilo with
determined inundation values from the historical tsunami information. These values were static,
and therefore, it was deemed best to determine the mean values for each year, from which a
range within one standard deviation of these values could be created. The ranges provide a
68.27% chance of a similar event occurring within one standard deviation of the mean historical
inundation.
In addition, to ensure accurate sampling of inundation depths, wave heights for each
event were broken into two values for each tsunami strike in Hilo Bay, to take into account the
influence of local topography on the wave heights inside and outside of Hilo Bay. Two transects
were manually digitized that represent the coastal area outside of Hilo Bay and the coastal area
within Hilo Bay. Since the topography of the northwest border of Hilo Bay has been shown to
amplify the height of tsunamis, it was considered inaccurate to use a mean inundation depth for a
single event in this study.
The same flood depth values from the SCHEMA model were used in the HAZUS model
to determine the mean inundation depth in each transect. This is different from wave height in
46

that inundation depth is the amount of water above ground level at a particular location, while
wave height is the height of a wave measured along the coast. The mean inundation calculations
also returned a standard deviation for each mean in each transect. These standard deviations were
used to determine a minimum and maximum wave height by subtracting or adding the standard
deviation to the mean, respectively. By using standard deviations, these minimum and maximum
values are intended to produce an output of each model that will cover approximately 70 percent
of outcomes for a similar tsunami event. The 1946 and 1960 events each have a minimum and
maximum wave height range for the inside of Hilo Bay, and outside of Hilo Bay. The minimum
inundation value is one standard deviation below the mean, used to determine a lower dollar
exposure outcome, while the maximum inundation value is one standard deviation above the
mean, intended to estimate a higher dollar exposure outcome. HAZUS allows one inundation
value for each run, so each two calculations were performed for both 1946 and 1960, once for a
minimum inundation value and once for a maximum inundation value. This differs from the
SCHEMA Flood model, which results in a range of damage values as the final output, versus an
exact number of damage values produced using HAZUS. Figure 13 displays the likelihood of an
outcome based on the standard deviation value.
47

Figure 13. Standard Deviation Bell Curve (Source: Massachusetts Institute of Technology). This
bell curve displays the statistical percentage of an event based on the distance from the mean.
The first step in running the HAZUS Flood Model was to create the two shoreline
transects for Hilo to divide the study area into two distinct topographical areas inside and outside
of Hilo Bay. After the transects had been created, HAZUS was implemented using the minimum
and maximum flood depth calculation values to create an inundation raster of the coastal area.
These flood depth values were placed into the 100-year SWEL field that allows the user to input
a custom value. HAZUS allows the user to provide a different value for each transect created, so
the minimum or maximum values were provided for each 100-year SWEL field. In effect, these
inputs delineated the floodplain and created the depth of flooding raster. This depth of flooding
raster was used within HAZUS to determine the degree of flooding for each census block. Figure
14 provides a methodology flowchart displaying the steps involved in running the HAZUS Flood
Model to produce tsunami inundation loss estimates in the form of a vector dataset.
48

Figure 14. Overview of the HAZUS Tsunami Flood Model. Green ovals are outputs, blue
squares are inputs, and yellow squares are tools or calculations. Transects were created during
the Depth of Flooding Raster step.
After the flood depth raster is created, the building damage can be estimated with the
HAZUS building inventory and HAZUS damage matrix described in the literature review.
HAZUS has a comprehensive building dataset that summarizes building information, for
example, based on critical infrastructure within census blocks. The previously created historical
1946 and 1960 DEMs were used as the input land elevation layer for each specific year. It is
important for consistency that the same DEM be used for wave heights and building elevations.
The HAZUS methodology has a different dollar exposure calculation for every additional foot of
49

flooding, while the SCHEMA methodology includes only a few classifications for dollar
exposure.
It is important to note that the census area analyzed in HAZUS is not the exact city
boundary of Hilo, and does not match up directly with the SCHEMA Flood Model’s study area.
The reason for the mismatch is that the census tracts in HAZUS do not share the exact same
border as the city limits of Hilo. To rectify this issue, the census blocks that were not within the
city limits of Hilo were omitted after the file was imported into ArcGIS. The output from
HAZUS was brought into ArcGIS to perform a select by location with the Hilo City boundary.
Any areas not within the city boundary were deleted to keep the study areas consistent.
By studying the outputs of both the SCHEMA and HAZUS models, this research is
intended to reveal the strengths and weaknesses of both methodologies for future tsunami flood
modeling. Both models produce economic loss information as outputs that were compared with
the historical loss information in order to address the original research questions. As mentioned
previously, inflation is also considered regarding each monetary result, and these calculations
were completed in a post-processing step. This inflation computation step ensured that historic
economic losses are accurately compared to today’s costs.

50

Chapter 4 Results
The results of this study indicated that there are several important differences between modeling
tsunamis using the SCHEMA methodology versus FEMA’s HAZUS Flood Model. Both models
have their strengths and weaknesses, and both are very useful for different reasons. The results
chapter consists of three subsections; a discussion of the SCHEMA model outputs, the results of
the HAZUS model, and a historical comparison of the results of both models.
In both models, the monetary values were inflated to 2016 dollar values for structural
inventory to provide a more accurate comparison between all outputs. The same elevation values
were used for each comparison, 1946 and 1960, between the models to so that ground elevation
would be a held constant, rather than consider it as another variable.
4.1 SCHEMA Model
During the parcel building classification, a new point file of high cost buildings was
created to represent the highest economic cost structures to repair. This layer is displayed with
the hotel layer on several of the maps to help explain why that area may have a high dollar
exposure to tsunami inundation. Figure 15 displays the hotels and high value buildings most at
risk in the Hilo costal area.
51

Figure 15. Hilo Buildings at Risk. Examples of high value buildings and areas potentially at risk.
The hotels and buildings at risk indicate that there are a number of areas of economic
importance that may be impacted by tsunami events. The government district of Hilo is near the
old downtown, which was destroyed in both the 1946 and 1960 tsunamis. However, this area has
taller concrete buildings and is farther away from the ocean than in previous years. These
buildings are likely to be impacted by future tsunamis, though not destroyed unless the wave
height exceeds the historical event considered. There is also an oil and gas terminal with large
storage tankers right next to the bay. This terminal is adjacent to the cruise terminal within the
main port area of Hilo. This area is commercial and is also economically vulnerable with large
buildings next to the ocean. The rest of Hilo has several large hotels and apartment complexes
52

within several hundred feet of the ocean in some cases. These are sturdy buildings yet still
vulnerable to a large tsunami event due to their close proximity to the bay.
4.1.1. 1946 tsunami event
The 1946 tsunami event in Hilo originated in Alaska and traveled directly south into Hilo
Bay. This tsunami devastated the downtown area of Hilo but left the Waiakea Peninsula, which
extends into the bay, mostly untouched. Figure 16 shows the inundation line of the 1946 tsunami.
Figure 16. 1946 Tsunami Inundation
The 1946 tsunami extended far into the main downtown area of Hilo, which is directly
south of the bay. The Waiakea Peninsula experienced inundation along the shoreline, but the
structures there were largely untouched. However, the pier complex was devastated, and appears
to still be vulnerable. The eastern area of the city outside of the bay was also inundated during
53

this tsunami event. The inundation was reported as wave height almost uniformly all along the
coast of Hilo, with the one exception of the Waiakea Peninsula. Figure 17 shows the dollar
exposure of the 1946 tsunami event with 2015 parcels predicted for this event using the HAZUS
Flood Model.

Figure 17. 1946 Hilo Dollar Exposure
The dollar exposure of the 1946 event in today's’ costs shows a drop-off of economic
impact compared with the historical incident (Pararas-Carayannis 1977). The historical incident
reportedly cost over $317 million, while the same event with 2015 parcels resulted in a dollar
exposure of just over $74 million. Some of the historical dollar exposure reported probably
included damage from debris and physical objects like vehicles, as well as buildings that were
54

damaged or destroyed. Table 8 displays the dollar exposures for the 1946 tsunami event. The
reasons for the difference in these costs are hypothesized in the following discussion.
Table 8. 1946 Dollar Exposures. Source: (Pararas-Carayannis 1977)
1946 Historical
Adjusted to
2016 value
Current
Dollar Exposure $317,509,330 $74,384,564

The SCHEMA model for the 1946 tsunami event shows very little damage in the area
south of Hilo Bay. This area comprised main downtown Hilo up until the 1960 tsunami. These
results show that the hesitance to rebuild the old downtown area of Hilo paid off in fewer
damages incurred due to the 1960 event. This area is now open greenery for several blocks along
the coast. However, the present-day Waiakea Peninsula shows some areas of high dollar
exposure, unlike in 1946. Several of these parcels have large hotels or apartment complexes that
may be vulnerable to a large tsunami similar to the 1946 event. The old pier area also shows a lot
of high dollar exposure to this event. These parcels have several large commercial businesses
right next to the bay, which would be heavily impacted by a similar event. This could also
become a larger-scale environmental disaster due to the oil and gas terminal at the pier. The
eastern portion of Hilo, outside of the bay, was also heavily inundated. Today the parcels that
were affected are mostly low in dollar exposure, but this area is now heavily residential and thus
could experience many fatalities. Several parcels adjacent to the shoreline also have large hotels
or apartment buildings that show a high dollar exposure potential.

55

4.1.2. 1960 tsunami event
The 1960 tsunami event in Hilo originated in Chile and traveled around the Big Island
and into Hilo’s Bay. This tsunami also devastated the downtown area of Hilo, but unlike the
1946 event, it also destroyed much of the Waiakea Peninsula. Figure 18 shows the inundation
line of the 1960 tsunami.

Figure 18. 1960 Tsunami Inundation
The 1960 tsunami was very similar to the 1946 event in terms of overall height, yet it
impacted different areas of Hilo. The downtown area of Hilo was destroyed again, and this time,
it was not rebuilt. The main difference compared to the 1946 event is that the Waiakea Peninsula
was hit the hardest of any area in Hilo. In 1946 the waves wrapped around the peninsula but did
not touch the center of the peninsula. In 1960, the waves completely overtopped the peninsula
56

and greatly damaged the entire area. Many of the lives lost in the 1960 tsunami event were due to
Hilo citizens believing they would be safe in the peninsula as they were in 1946. This tsunami
was less destructive than the 1946 event, partly due to areas not being rebuilt and increased
building standards. The tsunami did not inundate as much area to the east of the bay, most of the
physical damage was within the bay. Figure 19 shows the dollar exposure for the 1960 tsunami
event using 2015 parcels.

Figure 19. 1960 Hilo Dollar Exposure
The dollar exposure of the 1960 event in today’s costs shows another huge drop-off of
economic impact compared to the 1946 incident. The 1960 event had a dollar exposure of over
$185 million, while the same event with 2015 parcels only had a dollar exposure of just under
$50 million (Pararas-Carayannis 1977). Again, this can partly be explained by the lack of
57

vehicles and debris being calculated in the ArcGIS model, but the difference is also likely due to
both improved building standards and avoidance of building in areas of low elevation vulnerable
to tsunamis. Table 9 displays the dollar exposures for the 1960 tsunami event.
Table 9. 1960 Dollar Exposures. Source: (Pararas-Carayannis 1977)
1960 Historical
Adjusted to
2016 value
Current
Dollar Exposure $185,035,000 $49,475,963
4.2 HAZUS Model
HAZUS is capable of modeling coastal flooding, but only with set, or know input
flooding values that are static across a given study area. Due to this issue, each HAZUS scenario
was run with a minimum and maximum value that is one standard deviation from the mean
inundation level in order to understand a range of possible damage costs. Also, each HAZUS
scenario was created using transects for calculations inside and outside of the bay because the
tsunami heights within the bay tend to differ dramatically from those outside of the bay. Figure
20 displays the two transects used for each scenario.
58

Figure 20. HAZUS Transects
4.2.1. 1946 HAZUS Inundation Scenario
The 1946 SCHEMA output suggested that the 1946 tsunami had a much larger impact on
the historic downtown area of Hilo, missing most of the Waiakea Peninsula. The tsunami also hit
the port area hard, as well as the area to the east of the bay. The HAZUS inundation values were
calculated from the mean inundation elevation for each transect. A range of values was then
created using the standard deviation of these mean values to create a minimum and maximum
inundation value to use in HAZUS. Table 10 shows the inundation values used for the 1946
event.

59

Table 10. 1946 HAZUS Inundation Values
1946 HAZUS Inundation Values (in feet)
Location In Bay Outside Bay
Minimum Inundation Depth 4.7 5.3
Maximum Inundation Depth 17.5 15.7

It is important to note that the values displayed on the maps and tables for HAZUS are in
feet because that is the unit of measure used in HAZUS. The minimum inundation depths and the
maximum inundation depths are both within just a few feet regardless of their location. It is an
indication that the 1946 tsunami was fairly uniform when it made landfall. The tsunami came
directly from the north in Alaska, which might indicate that the momentum of the water was not
diverted much on its way to Hawaii. Figure 21 displays the 1946 minimum inundation depth
from the HAZUS model. The amount or degree of refraction of these initial tsunami waves off of
the northwestern topographic highs is not documented in the references sited in this study.

60

Figure 21. 1946 Minimum Inundation Depth
The 1946 minimum inundation scenario in HAZUS appeared mild in comparison to the
SCHEMA scenario, as expected. This scenario shows inundation of 4.7 feet inside the bay and
5.3 feet outside of the bay. Hilo Bay experienced very little flooding outside of the immediate
area near the coast. Much of this land is now parkland and is, therefore, unlikely to have a high
economic impact. The tsunami did impact the port area, which would lead to moderate dollar
exposure. The eastern area of Hilo, outside of the bay, experienced mild flooding along the coast
which may impact several hotels or apartment buildings. Figure 22 displays the dollar exposure
of the 1946 minimum inundation scenario.
61

Figure 22. 1946 Minimum Dollar Exposure
The 1946 minimum dollar exposure for Hilo amounted to just over $46 million. This is
quite a bit more than expected, considering the inundation levels are quite low. However, Hilo is
particularly susceptible to coastal flooding because of its relatively low elevation. The areas of
high dollar exposure seem to point to the same areas where hotels and buildings at risk are still
located. The government district in downtown Hilo would experience some moderate dollar
exposure, as do parts of the Waiakea Peninsula. The port would be the hardest hit area in this
scenario as it covers a large portion of the eastern Hilo. The flooding does not seem to penetrate
very deep into the mainland. Figure 23 shows the inundation depth for the 1946 maximum
inundation scenario.
62

Figure 23. 1946 Maximum Inundation Depth
In contrast, the 1946 maximum inundation scenario in HAZUS appears to have
penetrated much of the coastal areas of Hilo. The minimum inundation scenario had moderate
flooding right along the coast, but this output shows extreme flooding several blocks into Hilo.
This scenario shows inundation of 17.5 feet inside the bay and 15.7 feet outside of the bay. Hilo
Bay would be subject to a great deal of flooding, and the Waiakea Peninsula would be
completely inundated. The middle area of the peninsula may experience less inundation as
opposed to the rest of the bay, which would be similar to what happened in 1946. The port area
would be significantly flooded, as well as the government district in Hilo. The eastern portion of
Hilo would also be inundated far into the island. Figure 24 shows the dollar exposure of the 1946
maximum inundation scenario.
63

Figure 24. 1946 Maximum Dollar Exposure
The 1946 maximum dollar exposure scenario for Hilo amounted to over 242 million
dollars in today costs. This is more in line with the historical accounts (Pararas-Carayannis
1977). As previously mentioned, inflated to 2016 dollars, the historical event claimed over $317
million, while including some property damage not estimated in HAZUS or SCHEMA. This
brings the HAZUS estimate in line with historical estimates. The maximum scenario shows
portions of the airport impacted that could wreak havoc on the local economy as well. While the
inundation did not appear to reach the airport, the airport parcel is very large and extended into
part of the inundated area. As a result, it is possible that several buildings on the airport grounds
could be impacted. And, as expected, the government district, port terminals, and coastal area
64

appear to be devastated in this scenario. Table 11 shows the historical, minimum, and maximum
dollar exposure estimates.
Table 11. 1946 HAZUS Dollar Exposures. Source: (Pararas-Carayannis 1977)
1946 Dollar Exposure
HAZUS Minimum
HAZUS
Maximum Historical
$46,277,435.00 $242,421,900.00
$317,509,330.00

The results for HAZUS appear to be similar to expectations. The maximum scenarios
being modeled with current building parcels and census blocks would likely show less dollar
exposure than the previous historical events. Many of the hardest hit areas in Hilo were either not
rebuilt after 1960 or were built to withstand further tsunami events. The 1946 event was
particularly devastating for the downtown area of Hilo, which no longer exists in its historical
location. This seems to be a primary reason for the lower dollar exposure amounts. The 1960
tsunami event was equally devastating for Hilo, but it impacted different areas of the town.
4.2.2. 1960 HAZUS Inundation Scenario
The 1960 SCHEMA output suggests that the 1960 tsunami had a much larger impact
within Hilo Bay than did the 1946 tsunami. This may be due to the increased magnitude of the
tsunami originating earthquake, or possibly the direction that the tsunami waves were traveling
around Hawaii. The 1960 event also had less inundation outside of Hilo Bay, to the east of Hilo.
The 1960 event destroyed much of the port area and Waiakea Peninsula, which was largely
spared from the 1946 tsunami. Table 12 shows the inundation values used for the 1960 event.

65

Table 12. 1960 HAZUS Inundation Values
1960 HAZUS Inundation Values (in feet)
Location In Bay Outside Bay
Minimum Inundation Depth 4.6 1.3
Maximum Inundation Depth 21.2 7.1

Again, these units are in feet, which is the unit of measure used in HAZUS. A quick
observation with the previous 1946 values is that the values outside of the bay are much lower
than in 1946. The minimum inundation values are also both lower than their respective
comparison values from 1946. The maximum inundation depth within the bay, however, is
nearly 4 feet higher than in 1946. These values created stark differences between the areas
impacted in each model. The 1960 tsunami did not uniformly impact the cost of Hilo, with some
areas being much harder hit than others. Figure 25 displays the 1960 minimum inundation depth
from the HAZUS model.
66

Figure 25. 1960 Minimum Inundation Depth
The 1960 minimum inundation scenario in HAZUS also seems to be very mild in
comparison to the SCHEMA scenario results. With much lower values of inundation than seen in
the 1946 model, the impact was relegated to only the coastal areas of Hilo. The flooding did not
appear to impact anything other than beachfront properties. This scenario shows inundation of
4.6 feet inside the bay and 1.3 feet outside of the bay. Hilo Bay would experience very little
flooding outside of the immediate area near the coast. The tsunami does seem to impact the port
area, which would have led to moderate dollar exposure in that area. The eastern part of Hilo
experienced very mild flooding along the coast, which is not likely to impact hotels or apartment
buildings. Figure 26 displays the dollar exposure of the 1946 minimum inundation scenario.
67

Figure 26. 1960 Minimum Dollar Exposure
The 1960 minimum dollar exposure scenario for Hilo amounted to just over 21 million
dollars. This is a minor flooding event even for a large coastal community like Hilo. The dollar
exposure seems to come from only a few locations. The government district shows moderate
damage from inundation. These buildings are typically reinforced concrete and therefore are only
expected to incur superficial damages. The port area is again one of the harder hit areas. This is a
recurring theme and is expected for a coastal flooding event. A surprise in the 1960 minimum
scenario is some moderate damage seen outside of the bay. One particular parcel that was hit
hard, shown in the darkest color (reddish-orange), is situated at a very low elevation. This dollar
exposure could be due to buildings or docks built right on the ocean. The flooding around Hilo is
68

very mild and would be a best case scenario for a large tsunami event. Figure 27 shows the
inundation depth for the 1960 maximum inundation scenario.

Figure 27. 1960 Maximum Inundation Depth
The 1960 maximum inundation scenario in HAZUS penetrates much of the coastal areas
within Hilo Bay. The flooding outside of the bay is very moderate in comparison to the 1946
scenario. The 1960 minimum scenario had very mild flooding throughout, but the 1960
maximum scenario flooded much of the urban area around Hilo Bay. This scenario has an
inundation of 21.2 feet inside of the bay, and 7.1 feet outside of the bay. The government district
appears to be heavily inundated, as is much of the Waiakea Peninsula. In 1960 the historical
tsunami event was well known for destroying much of Waiakea Peninsula. Since the 1946
tsunami heavily hit downtown but spared much of the peninsula, Hilo citizens who experienced
69

the 1946 event thought that a similar event would take place and felt safe within Waiakea
Peninsula in 1960. As this flooding scenario shows, the peninsula was greatly impacted and saw
many lives lost throughout Hilo in the 1960 tsunami. This event overall had a much lower dollar
exposure outside of the bay but saw a similar total to 1946 due to the heavy inundation within
Hilo Bay. Figure 28 shows the dollar exposure of the 1960 maximum inundation scenario.

Figure 28. 1960 Maximum Dollar Exposure
As expected the 1960 maximum inundation scenario causes much of the dollar exposure
within the Hilo Bay, while outside of the bay there is much less destruction. The 1960 maximum
dollar exposure for Hilo amounted to just over 236 million dollars. The historical amount,
inflated to 2016 dollars, comes to just over $185 million. The 1960 HAZUS maximum scenario
shows a higher dollar exposure than the actual historical event. Part of this may be due to the
70

maximum inundation traveling farther into the heart of Hilo than the actual tsunami was likely to
have done. Due to the varied areas of flooding in 1960, this tsunami was particularly hard to
accurately model. The areas of the highest dollar exposure to this tsunami event are again the
government district and the port area. The downtown area of Hilo, which was pushed further
inland, also appears to be moderately impacted by this scenario. The Waiakea Peninsula saw
heavy inundation in this scenario but did not have a high dollar exposure. This is due to the type
of buildings found on the peninsula. Most of the construction on the peninsula is reinforced
concrete hotels and apartments which are likely to withstand even a large tsunami with only
moderate dollar exposure. Table 13 shows the historical, minimum, and maximum dollar
exposure values.
Table 13. 1960 HAZUS Dollar Exposures. Source: (Pararas-Carayannis 1977)
1960 Dollar Exposure
HAZUS Minimum
HAZUS
Maximum Historical
$21,128,929.00 $236,187,331.00
$185,035,000.00

These results for the HAZUS scenario are mostly in line with expectations. The historical
amount was expected to be higher than the maximum inundation scenario. However, it is
reasonable to expect a higher number for the maximum scenario due to an increased area of
inundation inside Hilo Bay. After the 1960 event, downtown was not rebuilt in the same location
and was moved several blocks further south. This new downtown area appears to be moderately
impacted by the maximum inundation scenario, which may explain the higher dollar exposure.
These historical tsunami events are important to study for planning purposes, but there are many
unknown risks throughout the Pacific Ocean from tsunami-earthquakes that can strike at any
moment.
71

4.2.3. 2016 HAZUS Inundation Scenario
Hilo is always at risk of a disastrous tsunami event, but history does not always point to
future results. Due to the nature of variation in earthquake occurrences, it is unlikely that
historical tsunamis would happen in the same way. There are some large faults in the Pacific
Ocean that have not yet created recordable tsunami events in Hilo. The Cascadia fault has not
produced a large scale earthquake in hundreds of years, but it is expected to create one in the
next several hundred (Atwater, Yamagushi, and Satoko 2005). This area has the potential to
create a devastating tsunami that may hit Hilo. To model a worst-case scenario, the highest
values from the two historical tsunamis are used in a 2016 maximum inundation scenario. This
2016 scenario uses the inside bay value from 1960 and the outside bay inundation value from
1946. Table 14 shows the inundation values for a 2016 worst-case scenario.
Table 14. 2016 HAZUS Inundation Values
2016 HAZUS Inundation Values (in feet)
Location In Bay Outside Bay
Inundation Depth 21.2 15.7

These values of inundation display the highest historically referenced values for a
potential maximum inundation event. These values, like the previous maximum inundation
scenarios, are one standard deviation above the mean for each area. The intent is to show the
impact of a plausible worst-case tsunami event that could impact Hilo. This scenario shows the
devastating potential of a significant tsunami event, with a more uniform inundation throughout
the entire Hilo area. Figure 29 displays the inundation from a worst-case scenario.
72

Figure 29. 2016 Maximum Inundation Depth
This 2016 maximum inundation scenario uses a modern DEM to ensure that the elevation
correctly reflects current conditions. This current DEM may decrease the dollar exposure
considerably as the coastal area has been reshaped and rebuilt over the years to prepare for future
tsunami events. Despite using a different DEM, the scenario results are very similar to 1960
inside Hilo Bay. The downtown area still seems to be partly impacted, while the government
district is heavily inundated. The Waiakea Peninsula and port area are also under a considerable
amount of water. Outside of the bay, the inundation appears to be similar to 1946, with less area
of inundation. The residential area north of the airport still appears to be heavily inundated, but
some of the more severe flooding missed several of the larger buildings. Figure 30 shows the
dollar exposure of the 2016 maximum inundation scenario.
73

Figure 30. 2016 Maximum Dollar Exposure
The 2016 maximum inundation scenario did not result in the highest dollar exposure by a
wide margin. This was expected due to the high inundation values of the historical events,
although the updated DEM seemed to contribute to the analysis predicting more disastrous
inundation in the eastern portion of Hilo. The downtown area and government district seem to be
the hardest hit areas in the 2016 scenario. The Waiakea Peninsula and port area both seem to
have significant inundation similar to the 1960 maximum inundation scenario. The real
difference between the historical models and the 2016 scenario is the eastern part of Hilo, which
was not hit as hard as expected. Figure 31 shows the maximum dollar exposures for each year
using the HAZUS models.
74

Figure 31. HAZUS Maximum Dollar Exposures
These maximum dollar exposure results were as expected. The updated DEM seems to
contribute to the analysis predicting more severe inundation in Hilo, while still having the
highest dollar exposure due to higher inundation values. These results suggest that despite
careful planning, Hilo is still at the mercy of tsunamis.
4.3 Historical Comparison
HAZUS is a comprehensive program designed to plan for natural disasters, but it is not
currently built to model tsunamis. ArcGIS is the framework for HAZUS, but it does not include
the technical modeling framework within HAZUS. Before the modeling was completed, it was
expected that the SCHEMA models would fall within the HAZUS output ranges for each
scenario. This was proven to be correct, with each SCHEMA scenario analyzed coming closest
to the outputs of the minimum inundation scenarios. Each model has its benefits but also
negatives when attempting to accurately model a historical tsunami.

$242,421,900
$236,187,331
$319,136,863
$0.00
$50,000,000.00
$100,000,000.00
$150,000,000.00
$200,000,000.00
$250,000,000.00
$300,000,000.00
$350,000,000.00
1946 Maximum 1960 Maximum 2016 Maximum
HAZUS Maximum Dollar
Exposures
75

4.3.1. 1946 Tsunami
The 1946 scenarios results indicated uniform inundation throughout the study area. Areas
both inside and outside of the bay saw moderate to severe inundation as Hilo’s coastal defenses
proved inadequate. Figure 32 shows the total dollar exposure for the 1946 tsunami event in each
model.

Figure 32. 1946 Tsunami Dollar Exposures
The SCHEMA model provided a promising output for total dollar exposure of buildings.
It was well within the range of the HAZUS models, and significantly lower than the Historical
dollar exposure total, which was expected. If an identical tsunami was to hit Hilo today, it would
likely cause damages within the range of those predicted by HAZUS, and also close to the
SCHEMA model outputs when considering building losses. The lower dollar exposures for each
model, compared to the historical dollar exposure, suggests that Hilo has done well in
preparation for similar tsunami events.

$46,277,435.00
$74,384,564.00
$242,421,900.00
$317,509,330.00
$0.00
$50,000,000.00
$100,000,000.00
$150,000,000.00
$200,000,000.00
$250,000,000.00
$300,000,000.00
$350,000,000.00
HAZUS Minimum ArcGIS SCHEMA HAZUS Maximum Historical
1946 Tsunami Event Dollar Exposure
76

4.3.2. 1960 Tsunami
The 1960 scenarios resulted in very uneven inundation throughout the study area. Areas
inside Hilo Bay saw the highest levels of inundation, while areas to the east of the bay only saw
moderate inundation levels. It would be tough for any town to significantly prepare for
inundation to the scale of the 1960 tsunami event. Figure 33 shows the total dollar exposure for
the 1960 tsunami event in each model.

Figure 33. 1960 Tsunami Dollar Exposures
Again the SCHEMA model provided a promising output for total dollar exposure of
buildings. The accuracy of the dollar exposure for the SCHEMA is very similar to the 1946
output. It is well within the range of the HAZUS models, and relatively close to the minimum
inundation HAZUS output. It is also significantly lower than the historical dollar exposure total.
A similar event to 1960 tsunami would likely cause less than $100 million in damage, closer to
the SCHEMA model prediction when considering just building losses, as previously stated.
$21,128,929.00
$49,475,963.00
$236,187,331.00
$185,035,000.00
$0.00
$50,000,000.00
$100,000,000.00
$150,000,000.00
$200,000,000.00
$250,000,000.00
HAZUS Minimum ArcGIS SCHEMA HAZUS Maximum Historical
1960 Tsunami Event Dollar Exposure
77

The HAZUS maximum value is a moderate cause for alarm, as a somewhat larger
tsunami event may result in new records of dollar exposure due to the number of large buildings
located near the ocean. It would take a very large event to impact many of these new buildings,
but it is possible.

78

Chapter 5 Conclusions and Future Work
Research question 1) was to discover if the SCHEMA and HAZUS methodologies can be used to
estimate damage to the built environment of historical tsunami events in the city of Hilo, Hawaii.
It is clear that both SCHEMA and HAZUS are successful in providing relevant results in
modeling the impact of tsunami inundation on the Hilo built environment. These methodologies
tested the possibility of accurately modeling the tsunami inundation to estimate potential
economic losses to the built environment caused by a large tsunami event. This research suggests
that not only does the SCHEMA model compliment and compare to the HAZUS Flood Model,
but both HAZUS and the SCHEMA models provided outputs comparable to economic costs
incurred due to historical tsunami events.
Research question 2) was to determine if the SCHEMA model results in calculating
potential tsunami inundation damage to the built environment are similar or different compared
to results obtained using the HAZUS coastal flood model for a given event. There are many
variables in modeling tsunamis that can make it challenging to test all possible variations within
a given model. Focusing on inundation and building dollar exposure appears to have been
successful using the SCHEMA model. A practical and suitable model was built to model tsunami
inundation using the SCHEMA methodology was built in ArcGIS as part of this thesis work.
Both SCHEMA outputs for the 1946 and the 1960 tsunamis fell within the HAZUS range of
dollar exposure outcomes. The coastal flooding model within HAZUS is an accurate and proven
model, which suggests that the SCHEMA model should also be a possible tool to model coastal
flooding from a tsunami. Modeling tsunamis is still a work in progress due to the fact that
tsunamis are not static disasters and require more comprehensive modeling techniques to
increase the accuracy of models.
79

5.1 Recommendations
Tsunami preparedness is essential to both mitigate property loss and the loss of lives in a
large tsunami event. For many coastal communities that cannot be relocated, mitigation is their
only option. For most communities in the Pacific basin, the threat of a tsunami is uncertain, and
in most cases the probability of occurrence and recurrence intervals are not known (Eisner 2005).
Detecting areas of vulnerability is crucial for the future of coastal communities around the
Pacific Ocean. Proactive coastal communities are the ones who plan for the worst but hope for
the best.
Neither the HAZUS model nor the SCHEMA model factored in direction or velocity of
tsunami waves when estimating the economic loss. Due to the absence of modeling the flow and
velocity of waves, it is unclear if the current breakwater in Hilo would have any impact on a
tsunami event, or if additional breakwaters would be beneficial. In 1946 it was determined that
the breakwater, though partly destroyed, helped to absorb some of the tsunami’s energy (Miller
2011). While not modeled in this study, it is likely that an upgrade to the existing seawall, or an
additional seawall extended from the cliffs Northwest of Hilo Bay, would also absorb some of
the impacts of future tsunamis. If it is determined that the aesthetic beauty of Hilo Bay should
not be compromised by an onshore seawall, then an additional breakwater may help to mitigate
future tsunamis economic impact without significantly blocking the views of the Pacific Ocean.
A highly relevant case study, in the most recent Japanese tsunami in 2011 a tsunami wave
as high as 50 feet destroyed many towns along Japan’s coast. Also, tsunami events in 1896 and
1933 killed a total of 439 in the small fishing village of Fudai (NBC News 2011). However, the
major of the town of Fudai, Kotaku Wamura, learned from the past and decided that this would
not happen again. Kotaku Wamura fought for 40 years for many years to obtain the funds to
80

build an immense seawall construction project to project Fudai. Indeed, this 51 foot high seawall
protected and saved the town from major damage due to the 2011 tsunami, while communities
nearby were completely destroyed and thousands of people lost their lives (CBS News 2011).
Figure 34. Fudai Seawall (Adapted from the Associated Press 2011) shows the seawall that
Fudai built under the guidance of Kotaku Wamura.

Figure 34. Fudai Seawall (Adapted from the Associated Press 2011)
Since Hilo was heavily economically impacted by several tsunamis in the last 75 years,
this research strongly indicates that the city of Hilo should consider building a seawall. If a
seawall is determined to be the most prudent mitigation technique in Hilo, there could be a
number of ways to fund such a project. A special sales tax over several decades, such as for
touristic goods and services, could help fund a large seawall. Also encouraging a local tax
referendum could increase tax spending to create a new revenue stream to be used for tsunami
81

mitigation. Nevertheless, it is also important to understand the negative impacts of a seawall.
Hilo may see a reduction in tourists if the views of Hilo Bay become obstructed by a new
seawall. The citizens of Fudai, Japan understood their vulnerable coastal community. It is
important to note that until the seawall was tested by the 2011 tsunami, it was locally despised
and viewed as a waste of tax dollars (NBC News 2011). A new seawall in Hilo might also be
seen as a waste of money and an eyesore until a tsunami is held at bay by such a structure in the
Hawaiian Islands or other South Pacific island cluster.
The importance of this research is rooted in the knowledge that it could impart to urban
planners in coastal communities in understanding the potential damages that a wide range of
tsunamis may have on a coastal community. Regardless of the location, even a relatively small
tsunami can cause millions of dollars-worth of damage. Urban planners with access to ArcGIS
can create their rudimentary tsunami inundation models to see roughly where their communities
stand against a tsunami threat. They can also use a HAZUS Flood Model to determine potential
dollar exposure if they are aware of the height of flooding.
If a seawall and breakwater are not the desired mitigation techniques, increased scrutiny
of construction practices and more restrictive building codes could be another positive mitigation
option. Water just 7 feet deep will have pressure of 450 pounds per square foot. The deeper the
water, the greater the pressure on the walls of a structure (Reid Steel 2015). According to Reid
and Steel (20150, for example, if buildings are going to be built at low elevation along a
shoreline, it is better to build them such that so that water can flow under them. Suspended floors
with concrete framing on stilts can survive some of the force of an unexpected wave. In contract,
although timber-framed buildings are good construction material in earthquake prone areas
because they are light and thus the effects of earthquakes are reduced, timber is the worst
82

possible choice in tsunami-prone areas. The wood in these buildings is easy to lift off of a
foundation and will float just like a ship if not adequately tied down. These wooden structures
also create floating debris which can then destroy other buildings and cause additional loss of
life.
To build tsunami resistant buildings, wave surges should be avoided by building the
buildings out of the projected wave paths (Reid Steel 2015). A building may need to be
suspended on stilts to prevent a lower floor being inundated with water. Foundations should also
be deeper than usual and braced down to the footings of the foundation. Lower floors should be
made with concrete to give the building some weight, while steel frames should be used to resist
substantial loads in the case of a wall collapse. Tsunami-prone areas tend also to be in Hurricane
and Earthquake-prone areas as well. It is important to try and build structures that can be
structurally safe in any disaster environment.
HAZUS has many capabilities that were not fully explored in this research. The research
was more focused on the viability of a SCHEMA model, and HAZUS was used primarily as a
way to compare and validate the output data. HAZUS also allows post-disaster debris
calculations that could be completed with any coastal flooding model. Adding in this component
analysis and comparing it to historical information could provide more insight to potential future
dollar exposure overall. FEMA has also been working on including turbidity in more aspects of
the HAZUS Flood Model, which would be crucial for modeling tsunamis. HAZUS also does not
currently have the capability to integrate bathymetry data in a coastal flooding scenario. HAZUS
is more focused on terrain above sea level, so bathymetry analysis capabilities would need to be
created before HAZUS can model tsunamis (FEMA 2015).
83

For emergency management purposes, HAZUS could continue to be the go-to model
when time is crucial as long as long as HAZUS default datasets are deemed adequate for the task
at hand. ArcGIS models such as a SCHEMA implementation can become more comprehensive
than HAZUS, but that only happens with research, access to quality input data, and time spent
refining the model. There are several levels of HAZUS models that include more comprehensive
datasets and input data, but the scope of HAZUS is still limited in comparison to the entire
ArcGIS suite of tools and software. As additional tsunami data becomes available, both ArcGIS
and HAZUS should become more accurate at replicating tsunamis and also predicting potential
economic loss. HAZUS is currently more efficient at modeling specific disasters using static
values, whereas with the advanced interpolation tools within ArcGIS there may be room for
improvement using a customized ArcGIS model, as opposed to the closed backend data
processing system of HAZUS. Most importantly, tsunamis are not static waves (floods) and are
better modeled using a dynamic method. On the upside, FEMA’s HAZUS team is refining their
flooding models at every update and appear on track to be eventually able to model tsunamis
accurately within HAZUS.
Throughout the modeling process, a range of issues arose that either redirected the project
or slowed it down. Creating historical elevation models to use as a base for each historical
scenario was not originally planned when starting this project. However, tsunamis and local
planning projects can dramatically shape the landscape over the decades. It was essential that
historical elevation models were utilized to make sure that the model was being run using
historical terrain conditions.

84

5.2 Future Work
Future amendments to this study could be numerous depending on the goals of future
research. Specifically, using data from HAZUS within a SCHEMA study, and vice-versa, could
provide additional insights into the strengths of each model and the accuracy of the output data.
In the future, to further this study it is recommended that an up-to-date AEBM dataset is
incorporated into the HAZUS models. The AEBM provides the most accurate building-specific
HAZUS analysis (FEMA 2014). For mitigation purposes, it would be useful for users to be able
to create building-specific damage and loss functions that could be used to assess potential losses
at a very high accuracy. This would provide a more specific economic loss estimate compared to
the general building functions found in the default data for HAZUS.
Inundation depth rasters that were created in ArcGIS for the SCHEMA model could be
used in HAZUS to simulate the same flooding conditions, and to evaluate the building stock in
HAZUS with the ArcGIS parcel information. This was not done in the initial research because
that would only provide insight into the differences between the parcel layer and the building
stock layer. It was imperative for this project to create a range, or base, of results that the
SCHEMA model could be compared to. Using the building stock layer in ArcGIS could also
provide more insight with how each model calculates dollar exposure. The census blocks in
HAZUS may be too large regarding spatial extent to be used accurately in the SCHEMA model
methodology, but it may be worth attempting. The accuracy of the SCHEMA building
classification is always fair to question as the input data was compiled using visual inspection of
recent aerial imagery and street view images. Finding or creating a more comprehensive building
database would be essential for additional modeling.
85

Additional research could also include study areas of the Pacific basin where large
earthquakes are likely to occur that could impact Hilo. The Cascadia tectonic boundary off the
coast of Oregon and Washington has long been suspected of being capable of producing large
tsunamis. There has been sediment found throughout the Pacific basin from a large tsunami
event that occurred in the 1600’s that may point to a future Cascadia earthquake and resulting
tsunami. Further research into potential sizes of historical Cascadia tsunamis and the likelihood
of future earthquakes would make for a comprehensive extension of this research.
In-depth terrain changes throughout Hilo since the 1946 tsunami would also be a
worthwhile addition to this research. Aside from tsunamis changing the landscape, Hilo has built
different variations of seawalls, extended the beach into the ocean, and turned previous
commercial and residential areas into green space along the coast. Some of these were created to
mitigate tsunamis, but the full impact has not been studied. Continuing this research by
evaluating terrain changes and building code changes over the years would be helpful in finding
out what may be working, and what could be useful for other coastal communities as well to help
them mitigate potential tsunamis. Related to terrain changes, one challenge is to find accurate
DEMs to use as historical references. Historical DEM’s often were created from topographic
maps that either had too few contour lines or questionably accuracy. These historical topographic
maps were still very useful in creating historical DEM’s as long as their limitations are
understood, but there may be more efficient or accurate ways to determine historical elevation
levels.
Thought this study required a great deal of time and effort to analyze the impact of such a
random disaster, tsunamis are some of the most destructive natural disasters we face. Luckily,
coastal communities typically receive a warning before impacts, but the physical damage cannot
86

be mitigated in a short time. These model results appear to be indicative of stricter building
codes and better urban planning and zoning. The town of Hilo seems to understand that they
cannot deter future tsunamis, only try to avoid inundation or build structures that can withstand
it. Even the smallest coastal communities should not overlook a tsunami preparedness plan, and
should be aware of how vulnerable they may be to a tsunami event. For the safety of these
communities, damage estimation models should be nurtured, and resources should be shared to
educate anyone potentially impacted. It would be useful to replicate this study with refined tools,
and updated data, to extend this research and help perfect the dynamic tsunami modeling
process.

87

REFERENCES
AIA. 2009. “A Community-Based Vision and Living Action Plan for Downtown Hilo.” Paper
presented at the AIA SDAT conference, Hilo, HI, May 5-8. Accessed February 25, 2016.
http://www.aia.org/aiaucmp/groups/aia/documents/pdf/aiab080963.pdf.

Atwater, Brian F., David K. Yamaguchi, Musumi-Rokkaku Satoko. 2005. “Orphan Tsunami of
1700: Japanese Clues to a Parent Earthquake in North America.” Professional Paper
1707: 1-42.
Bernard, Eddie N., 2005. “The U.S. National Tsunami Hazard Mitigation Program: A Successful
State-Federal Partnership.” Natural Hazards 35, no. 1: 5-24.
BLS. 2016. “CPI Inflation Calculator.” Accessed February 23, 2016.
http://www.bls.gov/data/inflation_calculator.htm.
Britannica Academic. 2015. "Chile earthquake of 1960," Accessed September 14,
2015. http://academic.eb.com/EBchecked/topic/1421130/Chile-earthquake-of-1960.
Bryant, Edward. 2008. Tsunami: The Underrated Hazard. Chichester, UK, Praxis Publishing.
CBS News. 2011. “The Man Who Stopped The Tsunami.” Accessed April 24, 2016.
http://www.cbsnews.com/news/the-man-who-stopped-the-tsunami/.
Cheung, Kwok Fai, Yefei Bai, Yoshiki Yamazaki. 2013. “Surges around the Hawaiian Islands
from the 2011 Tohoku Tsunami.” Journal of Geophysical Research 118: 5703-5719.
Davidson, Sarah. 2004. “Mystery of Deadly 1946 Tsunami Deepens.” Accessed February 15,
2016. http://www.livescience.com/3732-mystery-deadly-1946-tsunami-deepens.html.
Dominey-Howes, Dale, Paula Dunbar, Jesse Varner, Maria Papathoma-Kohle. 2010. “Estimating
probable maximum loss from a Cascadia tsunami.” Natural Hazards 53: 43-61.
Eaton, J. P., D. H. Richter, W. U. Ault. 1961. The tsunami of May 23, 1960, on the island of
Hawaii. Bulletin of the Seismological Society of America 51: 135–157.
Eisner, Richard K. 2005. “Planning for tsunami: Reducing future losses through
mitigation.” Natural Hazards 35, no. 1: 155-162.
ESRI. 2016. “How Kriging Works.” Accessed March 21, 2016.
http://desktop.arcgis.com/en/arcmap/10.3/tools/3d-analyst-toolbox/how-kriging-
works.htm.
FEMA. 2014. Advanced Engineering Building Module (AEBM) Manual – HAZUS MR1.
Washington, D.C., Department of Homeland Security.
———. 2015. Flood Model Technical Manual. Washington, D.C., Department of Homeland
Security.
88

———. 2015. Flood Model User Manual. Washington, D.C., Department of Homeland Security.
Greenslade, Diana J. M., Vasily V. Titov. 2008. “A Comparison Study of Two Numerical
Tsunami Forecasting Systems.” Pure and Applied Geophysics 165: 1991-2001.
King, Hobart. 2016. “Tsunami Geology- What Causes a Tsunami?” Accessed April 25, 2016.
http://geology.com/articles/tsunami-geology.shtml.
Macdonald, Gordon A., Chester K. Wentworth. 1954. “The tsunami of November 4, 1952, on the
island of Hawaii.” Bulletin of the Seismological Society of America 44, no. 3: 463-469.
Mazzotti, Stephane, John Adams. 2004. “Variability of Near-Term Probability for the Next Great
Earthquake on the Cascadia Subduction Zone.” Bulletin of the Seismology Society of
America 94, no. 5: 1954-1959.
Miller, Holly. 2011. “Waves of Change The Role of Tsunamis in the Changing Built Landscape
on Mokuola Island in Hilo Bay, Hawai’i.” Hohonu 9:19-28.
National Ocean Service. 2015. “What is a tsunami?” Accessed March 13, 2016.
http://oceanservice.noaa.gov/facts/tsunami.html.
Natural Hazards Big Island. 2015 “Tsunami Hazards Mitigation” Accessed October 1, 2015.
http://hilo.hawaii.edu/~nat_haz/tsunamis/.
NBC News. 2011. “How one Japanese village defied the tsunami.” Accessed April 24, 2016.
http://www.nbcnews.com/id/43018489/ns/world_news-asia_pacific/t/how-one-japanese-
village-defied-tsunami/.
NOAA. 2015. “NOAA Tsunami Inundation Digital Elevation Models (DEMs).” Accessed
December 15
th
, 2015. https://www.ngdc.noaa.gov/mgg/inundation/tsunami.
Office for Costal Management. 2015 “Hawai’i Tsunami Hazard Information Service.” Accessed
October 1, 2015. https://coast.noaa.gov/regions/pacific/resources/tsunami/.
O’Sullivan, Mike. 2015. “Hawaii, Vulnerable to Tsunamis, Prepares for the Worst.” Accessed
April 24, 2016. http://www.voanews.com/content/hawaii-vulnerable-yo-tsunamis-
prepares-for-the-worst/2631262.html.
Palmer, Robert Q., Michael E. Mulvihill, Gerald T. Funasaki. 1965. Hilo Harbor Tsunami Model
– Reflected Waves Superimposed. Honolulu, HI, American Society of Civil Engineers.
Pararas-Carayannis, George. 1977. Catalog of Tsunami in the Hawaiian Islands. Boulder, Co,
World Data Center A for Solid Earth Geophysics.
Peters, Robert, Bruce Jaffe, Guy Gelfenbaum. 2007. “Distribution and sedimentary
characteristics of tsunami deposits along the Cascadia margin of western North
America.” Sedimentary Geology 200: 372-386.
Reid Steel. 2016. “How to Make Buildings Safer in Tsunamis.” Accessed May 20, 2016.
http://www.reidsteel.com/information/tsunami_resistant_building.htm.
89

Satake, Kenji, Kelin Wang, Brian F. Atwater. 2003. “Fault slip and seismic moment of the 1700
Cascadia earthquake inferred from Japanese tsunami descriptions.” Journal of
Geophysical Research. 108, no. 11: 1-17.
Satake, Kenji, Kunihiko Shimazaki, Yoshinobu Tsuji, Kazue Ueda. 1996. “Time and size of a
giant earthquake in Cascadia inferred from Japanese tsunami record of January 1700.”
Nature 379, no. 6562: 246-249.
Schneider, Phillip. 2011. Solutions to Coastal Disasters 2011. Washington, DC. American
Society of Civil Engineers.
Shepherd, F., G. Macdonald, D. Cox. 1949. “The Tsunami of April 1, 1946.” Bulletin of the
Scripps Institution of Oceanography 5, no. 6: 6-33.
State of Hawaii. 2015. “Hawaii Statewide GIS Program.” Accessed December 15, 2015.
http://planning.hawaii.gov/gis.
Synolakis, Costas E., Eddie N. Bernard, Vasily V. Titov, Utku Kanoglu, Frank I. Gonzalez.
2008. “Validation and Verification of Tsunami Numerical Models.” Pure and Applied
Geophysics 165: 2197-2228.
Tang, Liujuan, Vasily V. Titov, Christopher D. Chamberlin. 2010. “A Tsunami Forecast Model
for Hilo, Hawaii.” NOAA OAR Special Report, PMEL Tsunami Forecast Series 1: 1-96.
Tinti, Stefano, R. Tonini, L. Bressan, A. Armigliato, A. Gardia, R. Guillande, N. Valencia, S.
Scheer. 2011. Handbook of Tsunami Hazard and Damage Scenarios. Luxembourg,
Publications Office of the European Union.
USGS Hawaiian Volcano Observatory. 1998. “Tsunami Generated by Magnitude 7.7 Earthquake
on November 29, 1975.” Accessed October 19, 2015.
http://hvo.wr.usgs.gov/earthquakes/destruct/1975Nov29/tsunami.html.
———. 1998. “The 1960 Tsunami, Hilo.” Accessed November 12, 2015.
http://hvo.wr.usgs.gov/volcanowatch/archive/1994/94_05_20.html.
USGS. 2015. “Light Detection and Ranging (LIDAR).” Accessed May 20, 2016.
https://lta.cr.usgs.gov/lidar_digitalelevation.
USC Tsunami Research Group. 2002. “1957 Aleutian Tsunami.” Accessed October 9, 2015.
http://cwis.usc.edu/dept/tsunamis/alaska/1957/webpages/index.html.
World Atlas. 2015. “What and Where is the Ring of Fire.” Accessed September 27, 2015.
http://www.worldatlas.com/aatlas/infopage/ringfire.htm.
Wu, Tso-Ren, Hui-Chuan Huang. 2009. “Modeling tsunami hazards from Manila trench to
Taiwan.” Journal of Asian Earth Sciences 36: 21-28.
Xie, Jinsong, Ioan Nistor, Tad Murty. 2011. “Tsunami risk for Western Canada and numerical
modelling of the Cascadia fault tsunami.” Natural Hazards 60:149-159.

Abstract (if available)

Abstract The city of Hilo, Hawaii is more vulnerable to tsunamis than any other location in the United States. Due to the unique bathymetry, topography, and location relative to the Cascadia Subduction Zone, in the future, Hilo could be struck by a large tsunami similar to the historic 1946 and 1960 events. The Cascadia Subduction Zone can produce a 9.5 M earthquake with the potential of generating a tsunami with maximum wave heights of over 29 feet. Before devastating economic loss occurs, it is imperative that such potential flood inundation and consequent dollar exposure are understood. This study compares the Joint Research Centre’s (JRC) Scenarios for Hazard-induced Emergencies Management (SCHEMA) flood model implemented using ArcGIS with the Federal Emergency Management Agency’s (FEMA) Hazards-United States (HAZUS) flood model to simulate the potential impact of a large-scale tsunami on the city of Hilo. The SCHEMA and HAZUS models, the National Oceanic and Atmospheric Administration (NOAA), and the State of Hawaii provided the spatial data required to build the financial and structural inventory database for these analyses. Field measurements recorded during the 1946 and 1960 tsunamis and corresponding historical inundation maps provided input into the models. The results of this research suggest that although the SCHEMA model has the benefit of being more customizable, the HAZUS inundation scenario can be implemented with fewer input data and produce results comparable to historical damages. Future work will involve refining the inundation scenarios to include more detailed input data such as historical terrain (digital elevation models), field-verified updates to the structural inventory database, and an increased number of predicted events based on wave height.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A comparison of two earthquake events in the City of Downey: the Puente Hills and Whittier faults at 7.0 and 6.8 magnitudes

PDF

Assessment of the FEMA HAZUS-MH 2.0 crop loss tool Fremont County, Iowa 2011

PDF

Preparing for the next major southern California earthquake: utilizing HAZUS with soils maps and ShakeMaps to predict regional bridge damage and closures

PDF

Validating the HAZUS coastal surge model for Superstorm Sandy

PDF

Risk analysis and assessment of non‐ductile concrete buildings in Los Angeles County using HAZUS‐MH

Asset Metadata

Creator Kline, Matthew John (author)

Core Title Modeling potential impacts of tsunamis on Hilo, Hawaii: comparison of the Joint Research Centre's SCHEMA and FEMA’s HAZUS inundation scenarios

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 06/15/2016

Defense Date 05/20/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Cascadia,dollar exposure,earthquake,economic loss,FEMA,flood,Geography,GIS,Hawaii,HAZUS,Hilo,inundation,JRC,Modeling,OAI-PMH Harvest,schema,subduction zone,tsunami,tsunami-earthquake

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Swift, Jennifer (committee chair), Fleming, Steven (committee member), Loyola, Laura (committee member)

Creator Email matthejk@usc.edu,mjkline@rocketmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-250674

Unique identifier UC11280256

Identifier etd-KlineMatth-4432.pdf (filename),usctheses-c40-250674 (legacy record id)

Legacy Identifier etd-KlineMatth-4432.pdf

Dmrecord 250674

Document Type Thesis

Format application/pdf (imt)

Rights Kline, Matthew John

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA