Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Preparing for earthquakes in Dallas-Fort Worth: applying HAZUS and network analysis to assess shelter accessibility
(USC Thesis Other)
Preparing for earthquakes in Dallas-Fort Worth: applying HAZUS and network analysis to assess shelter accessibility
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Preparing for Earthquakes in Dallas-Fort Worth:
Applying HAZUS and Network Analysis to Assess Shelter Accessibility
by
Crystal Eden Curtis
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
Copyright © 2016 by Crystal Curtis
ii
Contents
Contents..........................................................................................................................................iii
Chapter 1 Introduction.....................................................................................................................1
1.1 Motivation............................................................................................................................2
1.2 Goal of this research............................................................................................................3
1.3 Outline of this document......................................................................................................3
Chapter 2 Background.....................................................................................................................5
2.1 Study Area...........................................................................................................................5
2.2 North Texas Geology...........................................................................................................6
2.3 Earthquakes in Texas...........................................................................................................8
2.4 HAZUS..............................................................................................................................10
2.4.1. HAZUS for earthquakes studies..............................................................................11
2.5 Locating Emergency Shelters............................................................................................13
2.5.1. Using GIS to identify shelter locations....................................................................13
2.6 Location-Allocation...........................................................................................................13
2.6.1. Location-allocation for emergency shelter planning...............................................14
2.7 Summary............................................................................................................................15
Chapter 3 Data and Methodology..................................................................................................16
3.1 Data Used...........................................................................................................................16
3.1.1. HAZUS census tract base data.................................................................................17
3.1.2. HAZUS inventory data............................................................................................18
3.1.3. Fault Lines...............................................................................................................20
3.1.4. Historic earthquakes.................................................................................................22
3.1.5. Roads........................................................................................................................23
3.1.6. Shelters.....................................................................................................................23
iii
3.2 Methodology......................................................................................................................24
3.2.1. Running HAZUS scenarios......................................................................................26
3.2.2. Extracting data from HAZUS results.......................................................................27
3.3 Using ArcMap for Network Analysis................................................................................28
3.3.1. Creating the Network...............................................................................................28
3.3.2. Location-Allocation.................................................................................................30
3.4 Summary............................................................................................................................31
Chapter 4 Results...........................................................................................................................32
4.1 HAZUS Results.................................................................................................................32
4.2 Allocation of Population Centers.......................................................................................36
4.3 Summary............................................................................................................................40
Chapter 5 Conclusion....................................................................................................................42
5.1 Implication of these results................................................................................................42
5.2 Limitations.........................................................................................................................43
5.3 Future Research.................................................................................................................45
References......................................................................................................................................47
iv
List of Figures
Figure 1 Study Area...................................................................................................................................6
Figure 2 Balcones Fault Zone....................................................................................................................7
Figure 3 Barnett Shale / Fort Worth Basin................................................................................................8
Figure 4 Azle faults and earthquakes. The fault used in this project is the Antithetic fault
identified on the center left side of the map...................................................................................9
Figure 5 Annual Earthquake Loss............................................................................................................11
Figure 6 Population Density for Dallas and Tarrant County..................................................................18
Figure 7 Azle and Irvine faults as mapped for this project......................................................................21
Figure 8 Post-2009 earthquakes in the study area...................................................................................22
Figure 9 Potential shelter locations, including schools, recreation centers, Salvation Army
shelters and sports centers............................................................................................................24
Figure 10 Flowchart of the methodology followed.................................................................................25
Figure 11 Attenuation menu....................................................................................................................26
Figure 12 Small section of the road network...........................................................................................30
Figure 13 Bridge damage in Dallas County.............................................................................................33
Figure 14 Shelter needs in Dallas............................................................................................................34
Figure 15 Bridge damage in Tarrant County...........................................................................................35
Figure 16 Shelter needs for Tarrant County............................................................................................36
Figure 17 Allocation of tract centroids to shelters in Tarrant County.....................................................37
Figure 18: Allocation of tract centroids to shelters in Dallas County......................................................38
Figure 19: Unallocated points in crucial census tracts.............................................................................39
Figure 20: Allocation of census tracts by distance in meters...................................................................40
v
List of Tables
Table 1 Sample Individual Structures in a HAZUS Essential Facilities Category.................................19
Table 2 HAZUS Facility Classification Definitions...............................................................................19
vi
List of Abbreviations
CEUS: Central and Eastern United States
CDMS: Comprehensive Data Management System
DFW: Dallas-Fort Worth
FEMA: Federal Emergency Management Agency
HAZUS-MG: Hazards United States Multi-Hazards
NEHRP: National Earthquake Hazards Reduction Program
NTCOG: North Texas Council of Governments
OKC: Oklahoma City
PESH: Potential Earthquake Science Hazard
USGS: United States Geological Survey
WUS: Western United States
vii
Abstract
Earthquakes, which were previously rare events in the Dallas-Fort Worth (DFW) Metroplex, have
become increasingly common in the last five years. In a five-month period, Irving alone had 26
earthquakes over magnitude 2.0. Damage has been minimal, but larger earthquakes have not been ruled
out as new fault lines have been discovered and their precise structure is only beginning to be
understood. This project's goal was to show how HAZUS can be used to demonstrate possible
earthquake effects on the new fault lines, and how the results from HAZUS can be used to assess the
impact of damaged bridges on the road network and shelter accessibility. Two fault lines discovered in
the last couple of years were digitized and loaded into the HAZUS model. Historical earthquake data
were used to form a hypothetical earthquake scenario that characterizes local conditions. The scenario
was run twice, once on each fault.
To explore how these results can inform emergency planning, output from the HAZUS
scenarios regarding the amount of people needing shelter per census tract, as well as damage to bridges
and their location, was imported into ArcMap. A road network was built to support a location-
allocation model to assign people needing shelter to potential shelters and the damaged bridges were
added as barriers. A centroid was calculated for each census tract to provide population source points.
Lacking data on the location of existing emergency shelters, all schools throughout the two counties
were designated as potential shelter destinations. Then location-allocation analysis was conducted on
each county’s data to determine the closest accessible shelters with available capacity. The
demonstration scenario resulted in not enough shelters, as several source points were unallocated. It is
hoped that the demonstrations provided by this study will encourage city planners to begin to address
emergency planning in the region for these newly identified hazards.
viii
Chapter 1 Introduction
Several cities in the Dallas-Fort Worth (DFW) Metroplex, a historically quiet area geologically, have
suffered multiple, frequent earthquakes whose cause has been linked to wastewater injection wells that
accompany fracking for natural gas (Hornsbach et al. 2015). Cities and counties in the Metroplex area
have shown little consideration for earthquake-related issues, with the Texas state hazard plan labeling
an earthquake event as unlikely (Texas Department of Public Safety 2015). This thesis demonstrates
how HAZUS can be used to evaluate possible damage from an earthquake, and how the results from
that analysis can be used to plan for the now likely earthquakes.
The DFW Metroplex encompasses 12 counties in North Texas, but the core with the highest
population are Dallas, Tarrant, Denton, and Collin counties. DFW has a population of 5.92 million,
with Tarrant and Dallas having the highest population densities. The area is also one of the fastest
growing in the nation (Young 2015). These core counties sit on unique geologic structures that are now
known to make an earthquake possible in what should be a geologically quiet area (Jordan 1977).
Because the DFW Metroplex sits on the eastern edge of the Barnett Shale, a hydrocarbon rich
formation, many new wells have been dug to extract oil and gas since fracking makes previously
impossible extractions much more profitable (Galbraith 2013). Fracking is the process of injecting high
pressure water, sand and chemicals into the well to allow gas to pour out (BBC 2013). The waste water
and chemicals coming out of these wells are subsequently disposed of in injection wells. Recent
concern comes from the increase in earthquakes attributed to injection wells being positioned near
previously undetected fault lines and the injected liquid resulting in fault slippage. Previous studies
have pointed to wastewater injection wells as a possible cause of these earthquakes (Hornsbach et al.
2015)
1
The goal of this project was to demonstrate how HAZUS and ArcMap can be used to assist in
disaster planning for this new hazard in the DFW area. This was accomplished by showing how
accessible shelters for displaced residents can be identified using a road network that incorporated
damaged bridges predicted by HAZUS modeling. The 5.5 magnitude for the modeled earthquake
scenario was determined by recent statements from researchers who cautioned that an earthquake
stronger than the recent uptick in 2 to 4 magnitude earthquakes cannot be ruled out (Stump 2015).
1.1 Motivation
From October 2014 to March 2015, the city of Irving in Dallas County recorded over 20
earthquakes. While the majority were under magnitude 3.0, six were between 3.0 and 3.6 (Earthquake
Hazards Program 2016), which can cause slight damage close to the epicenter. These stimulated
discussions on earthquake frequency, and potential future consequences within the community (Selk
2015). Researchers at Southern Methodist University (SMU) are doing research into these local
earthquakes, and their conclusions have linked the earthquakes to new faults that are potentially
affected by nearby oil wells (Frolich et al. 2010). Their statement that stronger future earthquakes
cannot be ruled out is the motivation for this project.
Responding to this concern, this research explored the potential damage that might occur to
bridges from such larger earthquakes, the amount of people affected that will need short term shelter,
and whether this damage would affect their ability to get to shelter locations. This analysis used the
HAZUS (HAZards United States) model which was created by FEMA (Federal Emergency
Management Agency) to help localities prepare for hazards in the future and plan ways by which to
mitigate losses. In such an earthquake scenario, it is critical that the shelters be accessible by road,
which was examined by using the Network Analyst toolset in ArcMap. By using these tools, it
addresses the main questions of this project: Following a 5.5 earthquake along these newly discovered
2
faults, how many people may need shelter due to damaged homes and might there be an impact on the
road network caused by damage to bridges that will make shelters inaccessible? Practically, can
ArcMap be used to model access to shelters using damage results from HAZUS?
HAZUS-MH 2.2 has been used in Texas for analyzing damage from flooding and hurricanes,
however, earthquakes have traditionally been very rare and of small magnitude, which makes planning
for a moderate sized quake a low priority. Tarrant County does not list earthquakes as a hazard in their
hazard planning guide (Tarrant County 2015), and Texas’s statewide emergency planning guide only
casually mentions earthquakes (Texas Department of Public Safety 2013). The most likely threat
according to Texas's planning guide comes from the New Madrid fault, which lies at the border of
Missouri, Kentucky, Arkansas and Tennessee. The Texas guide does suggest that if earthquakes persist,
local planners would benefit from adjusting building codes to increase structural stability. This study
demonstrates this gap in hazard analysis by demonstrating a methodology that can be used to analyze in
advance potential damage to crucial road networks and to show how this damage might affect the
ability of people in the surrounding areas to travel to shelters.
1.2 Goal of this research
This project’s goal is to help encourage disaster planning in Texas based on the recent
occurrences of earthquakes. Shelters are crucial if people’s homes are too damaged to live in, so this
project sought to demonstrate how accessible shelters can be identified using a road network that
incorporated bridges too damaged for travel.
1.3 Outline of this document
This document next moves into Chapter 2 to a review of the background research that guided its
development, which is important since Texas has experienced few earthquakes and guidance must be
found elsewhere. Chapter 3 describes the research method that was split into several parts: analyzing
3
historical earthquake data; creating and loading faults; applying the HAZUS model; and using network
analysis to address shelter needs. Chapter 4 explains the results from running HAZUS and the location-
allocation tool. From this, conclusions were drawn and these are discussed, along with suggestions for
future research and policy development in DFW in the final chapter.
4
Chapter 2 Background
In order to understand the reason for the project, it is important to understand the Texas earthquake
context and several related concepts. The geology of the state is the fundamental base that makes
earthquakes possible. The history of these earthquakes is important to document to address potential
impact in the study area. HAZUS was a key program used in this project, and its usefulness is detailed
through several studies as a guide to how it should be used. Studies for locating emergency shelters
using location-allocation are discussed as a guide to constructing the methodology.
2.1 Study Area
The Dallas-Fort Worth Metroplex refers to a group of 12 counties in North Central Texas. The
study area for this project covers Dallas and Tarrant counties, which houses 71 towns and cities, as
shown in Figure 1. The Metroplex is home to 5.92 million inhabitants and is one of the four largest
metro areas in the United States. It was second nationwide in growth rate between 2013 and 2014
(Young 2015). With continued growth comes concern about the increase in the level of damage from
natural hazards. This is especially important since Dallas County houses a fault near Irving, and Tarrant
County has a fault near Azle.
5
Figure 1 Study Area
2.2 North Texas Geology
What makes earthquakes in Texas possible, despite millions of years of inactivity, is the unique
geology of the state. The Balcones Fault Zone runs from southwest Texas near Del Rio to the
northwest, close to Dallas (Figure 2). It is a collection of many smaller faults and was the result of
strain during the Tertiary era approximately 66 million years ago (Jordan 1977). The faulting ended in
the Miocene epoch between 23 and 5 million years ago, with relatively little activity since.
6
Figure 2 Balcones Fault Zone (Source: University of Texas 2016)
The other major geological feature in the study area is the Barnett Shale. It is a large natural gas
producing formation that encompasses 18 counties (Figure 3). It is normally associated with the Fort
Worth Basin. It provides a large economic incentive to drill the area, with royalties, taxes and fees
providing a great boon to cities and their residents. Fracking and horizontal drilling made efficient
drilling possible starting in 2002.
7
Figure 3 Barnett Shale / Fort Worth Basin (Source: University of Texas 2007)
2.3 Earthquakes in Texas
Researchers at Southern Methodist University (SMU) in Dallas have been recording and
analyzing these North Texas earthquakes since they became a prominent concern in 2009. Their goal
was to determine what the cause was in order to better address the issue going forward. An initial study
first looked at how they could determine earthquake location and explored a correlation between oil
exploration and earthquakes (Frohlich et al. 2010). They measured P and S wave arrival times, as well
as the locations of several wells and the amount of brine injection per day.
After a series of quakes in 2013 near Azle, the group of SMU researchers looked more in depth
to the causes. They located an ancient buried fault line (Figure 4), and then measured both natural
stress changes from water table changes and tectonic stress. This was then compared to stress changes
due to the oil and gas industry's extraction and injection methods. They concluded that the human-
8
induced activity produced the most significant stress changes (Hornbach et al. 2015). However, why
some areas with wells have earthquakes and others do not was reported as unclear.
Figure 4 Azle faults and earthquakes. The fault used in this project is the Antithetic fault
identified on the center left side of the map. (Source: Hornbach 2015)
In 2016, the USGS released a report by Peterson and other researchers detailing earthquake
hazard areas that included human-induced earthquakes for the first time. It focused on large areas of
Oklahoma, but also included the DFW area. Their study used five different parts of the seismic model
created for their report to evaluate risk, which includes earthquake maximum magnitudes and
earthquake locations (Petersen et al. 2015). The report details hazard risk over the course of a year
instead of 50 years, as policies that can affect induced earthquakes can vary year to year. In a summary
of the report, the USGS added that while human-induced earthquakes may have lower magnitude than
natural earthquakes, induced quakes may trigger larger quakes (Fitzpatrick and Petersen 2016).
9
2.2 HAZUS
The HAZUS model was created by FEMA in 1997 to estimate potential losses from
earthquakes, floods, and hurricanes. This was to help local officials plan for the after effects of
disasters and mitigate the losses. HAZUS is a large complex model framework composed of many
different modules that simulate the impact of hurricane, flood, and earthquake scenarios on a study area
created by the user. It and the state datasets are free to download from FEMA’s Flood Map Service
Center HAZUS page. It incorporates ArcGIS to visualize the impacts of a hazard on any given
category, such as hospitals or utility systems. It also can estimate the amount of casualties that a
population may suffer, as well as how many may need shelter. It has three different levels of usage,
with level 1 using only default data, and level 3 being heavily customized with local knowledge. As
noted above, in Texas, the most used applications of HAZUS are for understanding the impacts of
flooding and hurricanes, with earthquakes not yet being seen as a large issue.
HAZUS uses ArcMap as a means by which to map results and to add data such as custom fault
lines from outside the model itself. It comes with an extensive database inventory that includes details
about residential, commercial, and industrial buildings, as well as transportation networks and schools.
This inventory can be expanded by the user by importing their own data. Damage is measured as direct
damage to structures, and direct and indirect social and economic losses.
The HAZUS earthquake module uses the Physical Earth Science Hazard (PESH) model to
measure ground motion and ground failure. There are 59 equations available to determine ground
failure, including the Atkinson and Boore equation that is used in one of the studies discussed below.
The actual equations used in a particular model run are filtered based on the study area created and can
be chosen by the user. Ground motion is determined by attenuation models for Western United States
(WUS) and by ground-motion prediction equations for Central and Eastern United States (CEUS), as
10
well as spectral contour and probabilistic hazard maps by the USGS. Scenarios are either pre-made or
custom set by the user
2.2.1. HAZUS for earthquakes studies
In 2008, FEMA determined the average Annual Earthquake Loss (AEL) and average Annual
Earthquake Loss Ratio (AELR) risks for all states. Texas fell in the bottom half of both measurements
(FEMA 2008). While the FEMA study showed the AEL is moderate for this project's study area with
an AEL rating of 1 to 5 million dollars, the higher rating is because of the large population and
abundance of structures involved. Figure 5, which was drawn straight from the FEMA study, shows the
AEL for the United States. Tarrant County used HAZUS in their latest hazard mitigation action plan,
but did not include earthquakes (Tarrant County 2015), as to date most earthquakes have been minor to
light.
Figure 5 Annual Earthquake Loss (Source: FEMA 2008)
11
To see how HAZUS has been used to understand earthquake damages, it is necessary to look at
areas with a high potential of earthquakes. Ploeger (Ploeger, Atkinson and Samson 2010) studied
earthquake potential in Ottawa since it is the capital of Canada, has nearby fault lines, and a history of
moderate historical earthquakes. Ploeger's study uses the Atkinson and Boore equation in HAZUS to
determine ground shaking for the earthquake model as it provides amplitude as a part of the earthquake
distance and magnitude measurements. This predication was compared to expected ground motions
included in the 2005 National Building Code of Canada. The study also included soil classes as key
input as soils can amplify movement. The soil classification system used is from the National
Earthquake Hazard Reduction Program (NEHRP), and they used the SITEAMP program to convert
velocity and density into amplification. Data preparation was extensive, but HAZUS is not as complete
for Canada as it is for the US so Ottawa buildings were classified by visual inspection. In an earlier
study by the same author (Ploeger et al. 2008), they got around the lack of Canadian ground motion and
geographic conditions by using New York settings and replacing all the building data with Ottawa data.
PESH, which needs soil classification, liquefaction and landslide susceptibility, and water depth data,
was a guideline used by the author to collect the necessary Ottawa data.
The USGS applied HAZUS to King County in Washington due to the numerous faults, distance
to edge of a tectonic plate, and susceptibility to large earthquakes. They also emphasized ground
motion through use of the PESH model (Neighbors et al. 2013). They created seven scenarios to
compare and contrast results, split between shallow and deep earthquakes. They used three sources of
input: one from HAZUS historic catalog and two from the researcher's own input. Each of these were
run in both HAZUS default hazard input mode and user input mode influenced by NEHRP. Variation
in ground motion was compared to economic building damage. They noted higher values in peak
ground acceleration and economic building damage compared to using user-sourced input. This was
credited to inaccurate flaws in the default settings for earthquakes in HAZUS and the authors
12
commented that it is important to use information from previous studies and local data instead. By
using user supplied data, the values for damage and ground motion magnitude were reduced.
2.3 Locating Emergency Shelters
Emergency planning is a crucial function of municipalities, and identifying locations that can be
used as shelters is a key part of that process. Schools, community or recreation centers, and churches
are the most commonly identified shelters in Texas (TXDPS 2015). Also, as seen in New Orleans
following the Katrina hurricane, indoor sports arenas can also house people temporarily. Private
entities such as the Salvation Army can also provide shelter if the people come from an emergency
shelter and/or can be regarded as homeless, according to their regulations (Salvation Army 2016). GIS
has played an important role, using location-allocation and other models, to determine if there are
shelters in places where they are needed.
2.3.1. Using GIS to identify shelter locations
Kar and Hodgeson (2008) used GIS with a Weighted Linear Combination (WLC) model and
combined it with a Pass/Fail technique to identify which shelters were in suitable and non-suitable
areas. Values for the factors were determined by a rule-based technique that included criteria such as
proximity to highway and not being within 10 miles of a nuclear power plant. Weights were
determined by an opinion-based technique. The final value was determined as weight multiplied by
factor value. Excluded variables had a factor value of 0. This was then mapped in raster format due to
the large number of facilities and the large study area of Southern Florida.
2.4 Location-Allocation
Location-allocation chooses the best site(s) based on pre-determined criteria, such as travel
time, for the surrounding population situated at demand points. It does this by allocating the demand
13
points to the best fit facility(s) (Esri 2016). Such models requires a road network built from facility and
road data, and identification of any impedances to road travel. In any implementation of location-
allocation, there are several options for the Problem Type setting but Minimize Impedance and
Maximize Coverage are the best for planning public services such as shelters (Polo et al. 2016). To run
the tool, the facility destination points, the demand points, the road network, and any properties
essential to the specific project must be provided. The result is a map of both candidate and chosen
facilities, with lines showing which demands points are allocated to each facility.
2.4.1. Location-allocation for emergency shelter planning
Previous studies have applied location-allocation specifically to exploring emergency shelter
locations. Wu and others focused on applying this technique to the Yushu Tibetan prefecture in China
in order to help human life, property, and guide the government’s planning to protecting its citizens
(Wu et al . 2015). The authors extracted alternative shelters from a digitized map, identified and drew
important features such as first aid centers and water sources, and created the road network from
scratch by shaping and drawing the roads. The authors also established an index system that prioritized
safety, accessibility, and if it is fit for use. After building the network, the authors determined the
weight for each index using an analytic hierarchy process. They used arithmetic average as the
optimization strategy of the model, and natural breakpoint to divide the level. Anything above .5 was
determined to be the best location, and below .3 was not recommended to be a shelter location.
Kongsomsaksakul, Yang and Chen (2005) used location-allocation for locating shelters,
focusing on the implementation of the technique rather than on a specific practical application. He used
bi-level programming, with the assumption that the city administrators would designate shelters, while
the citizens would choose from those shelters. He used multiple mathematical formulas, the city of
Logan’s road network in Utah as a case study, and 10 pre-defined shelter locations. As would be
14
expected, shelter capacity had a large effect on the travel time, with smaller capacities increasing the
time it took by redirecting to the next closest shelter with capacity. He also looked at the impact of
shelter location on travel time, showing that the removal of two shelters increased the travel time by
several minutes. Combining both Kongsomsaksakul and Kar's studies, while out of date with current
GIS technology, gives guidance on shelter choice and road network impact for the location-allocation
model in this project.
2.5 Summary
The importance and motivation for this project is fueled by the geologic and demographic
context of the state of Texas. The populous DFW area sits on the Balcones Fault, which is an essential
concern as it houses many fault lines. Earthquakes have been rare until 2009, and this recent increase is
thought to be connected to wastewater injection wells. HAZUS can simulate the effects of earthquakes
in order to visualize the possible effects on an area, which allows for the study of a specific scenario to
occur in the DFW area. Previous studies provided guidance on how to execute HAZUS accurately. The
results from HAZUS can be incorporated into ArcMap to assist in planning how to handle a population
with no home to return to. To this end, location-allocation is useful as it locates the best facilities for
the demand points to travel to given particular objectives such as minimum travel time. Previous
studies provided detail on how it should be done and how shelters should be allocated. In the next
chapter, how these methodologies were applied to an examination of the earthquake hazards in the
study area is described.
15
Chapter 3 Data and Methodology
This chapter begins with a description of the data used. Then, the methodology is discussed in two
parts: using HAZUS and using ArcMap. HAZUS was run using a single scenario on two different
faults, Irving and Azle. Bridge damage and shelter needs results from HAZUS along with potential
shelter locations and road data were then incorporated into ArcMap in order to build a road network
with the damaged bridges as barriers. From this, location-allocation was run in order to determine the
closest shelters to meet the needs of the newly homeless in each census tract.
3.1 Data Used
Most of the data needed for this research was readily available, thanks to the federal
government. As noted above, HAZUS comes preloaded with a large data inventory, which is expansive
for the study area since it has a high population. The inventory can be further customized with local
knowledge if needed. The USGS offers earthquake point data for free through their Earthquake
Hazards Program, which can quickly be loaded into ArcMap.
The fault lines proved to be the trickiest data to create. Hornbach et al. (2015) published the
locations of the faults, although only in descriptive format. USGS does provide fault line data, but
neither the Azle nor Irving fault lines were included. The Irving fault line was only discovered in
February 2015, and while Azle was known prior to that, the study noting its location came out in 2015.
Thus as HAZUS does not include these fault lines, multiple steps, described below, were required in
order to use them in in this research.
School locations, bridge locations and damage estimates, and shelter needs, data necessary for
the Location-Allocation component, were obtained from the Results tab in HAZUS. These results can
be exported as shapefiles and loaded into ArcMap for manipulation and analysis. The location of
potential non-school shelters, such as recreation centers and sports arenas, were obtained through city
16
and county websites which list recreation centers and nearby sports centers. The only sports centers that
were included were fully enclosed buildings such as the AT&T Stadium in Arlington.
Road data was acquired from the North Texas Council of Government’s (NTCOG) website. It
was split into two shapefiles: streets and highways. Lakes were included as background data because
Eagle Mountain Lake serves as a barrier to population growth to the east of the Azle fault. It also limits
travel as only a few bridges cross it. More on each of these data sets is provided in the following
sections.
3.1.1. HAZUS census tract base data
HAZUS uses census tracts as the base polygon layer to identify at risk areas and to compile
damage statistics. The population and population density per square mile for 2010 for each census tract
are included in the inventory data. Figure 6 shows population density for the two counties in the study
area. These were classified using the natural breaks method as there were breaks in the data, with
Dallas having noticeable break points at 10000 and 25000 while Tarrant increased a lot after 7000. In
addition, Dallas was classified into six classes because there were two census tracts with abnormally
high densities, compared to the rest of the county, and should be their own class. All inventory data and
their specific characteristics are associated with a census tract, although essential facilities also have
addresses and names which allow for point data creation. Residential and commercial buildings only
have codes that identify building type. This allows HAZUS to calculate the effect of a scenario per
each census tract, creating a targeted analysis and view of the study area.
17
Figure 6 Population Density for Dallas and Tarrant County
3.1.2. HAZUS inventory data
Building type, construction materials, height, occupancy type, and address are just some of the
many building attributes that must be present in the inventory for HAZUS to conduct an accurate
analysis. HAZUS comes with an extensive list of structures in its default database, with user defined
buildings incorporated as General Building Stock (FEMA 2015). Some of these structures are stored in
the aggregate by census tract, others are stored individually. Although residential housing is critical
when calculating the damage estimates and determining how would many people need shelter if their
homes are too damaged, residences are only identified by code, depending on building type and
18
construction, and cannot be mapped individually as they do not have an address. Instead, each building
is associated with a census tract, with results showing damage per building category per census tract.
On the other hand, critical facilities such as schools and hospitals are stored individually. Table
1 shows a sample of Essential Facilities, which is one of many categories of inventory. Each building
has a class designation, where the definition of each is preset by FEMA as seen in Table 2. All
classification codes are listed in the data dictionary from FEMA. For HAZUS to calculate various
effects on the study area, each piece of inventory data must be associated with a census tract.
Table 1 A Sample of Individual Structures in a HAZUS Essential Facilities Category
ID Number Class Tract Name
TX000004 EFHL 48113014601 Baylor Medical CTR at Irving
TX000026 EFHM 48113006301 Dallas Southwest Medical Center
TX000031 EFHM 48113019211 Baylor/Richardson Med Center
TX000045 EFHL 48113008100 Doctors Hospital of Dallas
TX000062 EFHS 48113010000 Mental Health Connections
TX000063 EFHL 48113000401 Parkland Health & Hospital
Table 2 HAZUS Facility Classification Definitions
Class Description
EFClass EFHL = Large hospital
EFHM = Medium hospital
EFHS = Small hospital
EFMC = Medical clinic
Bridge locations and schools were able to be plotted directly into ArcMap, but the residential
data could only be aggregated at the census tract level. Bridge damage and shelter need attributes can
be associated with the bridges and census tracts, respectively, after a scenario is run. Bridge data can
also be acquired by running a scenario and selecting transportation results. Shelter needs can be
mapped per census tract, and are summarized in a table.
19
3.1.3. Fault Lines
Fault lines are typically where earthquakes happen, as they are the result of slippage and
releasing of pressure. While the USGS makes these available, the Azle and Irving faults have not been
loaded into HAZUS as they were only discovered in 2013 and 2015, respectively. Both of these are
also sub-surface faults, which makes precise locations difficult to measure. Seismometers and tracking
of earthquake epicenters are used to determine the approximate location, but the margin of error for
Irving is approximately half a mile due to it still being poorly understood (Deshon et al. 2015). It is
described as “a narrow two-mile line which indicates a fault extending from Irving into West Dallas,
running north-by-northeast from TX Highway 114 to Walnut Hill Road along the Trinity River”
(NBCDFW 2015, p. 1) in Dallas County. The Azle fault in Tarrant County is better mapped, as seen
above in Figure 4. Lines representing these faults were built in ArcMap from the available maps and
descriptions using the edit tool to create shapefiles (Figure 7). Their precise location is not critical in
this study.
20
Figure 7 Azle and Irvine faults as mapped for this project. Bottom images show the entire counties.
21
3.1.4. Historic earthquakes
It is important to know the attributes of historical earthquakes in the area in order to create a
scenario that represents local conditions. Epicenter location, depth, and magnitude are all important
factors for an accurate scenario. The earthquake data for earthquakes is made public by the USGS and
can be downloaded in CSV format, with location given as latitude and longitude coordinates in WGS
1984 datum. The data for earthquakes that occurred in the area post-2009 were downloaded, imported
into ArcMap and converted into a point shapefile ().
Figure 8 Post-2009 earthquakes in the study area
22
In order to run HAZUS, an earthquake scenario must be created that specifies the horizontal
epicenter, magnitude and depth. Using statements from Heather Deshon et al. (2015) from SMU as
justification, a 5.5 magnitude event was used. Depth for the scenario was determined by averaging the
depths of all the earthquakes recorded by the USGS in Tarrant and Dallas counties post-2009.
Epicenters for the earthquakes for the scenarios were located directly on the Azle and Irving fault lines.
Both Azle and Irving are short faults of two miles or less and their locations are not accurately known,
so an imprecise determination of the location of the scenario epicenters is sufficiently representative of
potential earthquakes in the area.
3.1.5. Roads
In order to create a location-allocation model, a topological road network is required. Roads and
highway line data were downloaded from the North Texas Council of Government (NTCOG) website.
Using the network construction tool in ArcMap, these line segments were converted into a fully
connected network dataset. Based on local knowledge of the area, speed was assigned as 40 miles per
hour for segments classified as roads and 70 miles per hour for highways. This information was used
by the location-allocation model to calculate the time it would take to travel each segment and
ultimately to determine the closest shelters to each census tract.
3.1.6. Shelters
Shelters are any building that is designated by either state or local municipalities to house
people temporarily. For this project, it is where people can go after their homes are damaged or
destroyed by an earthquake. Schools are a common type of shelter, and their location can be directly
downloaded from the HAZUS inventory as point data. There were approximately 2700 schools within
Tarrant and Dallas counties, and several hundred additional were included from the bordering counties
to see if they would be closer to travel to for the border census tracts.
23
Recreation centers and indoor sports arenas were also included as shelters in the analysis,
although they were not present in the default HAZUS data. Therefore, it was necessary to retrieve the
names and addresses of such facilities from the city websites, recording each facility listed in a
spreadsheet. Salvation Army and sports arena addresses were also acquired through their websites. All
of these were then geocoded using their addresses, and a point shapefile was created. This was done for
each county. Error: Reference source not found shows the distribution of all shelter points compiled.
Figure 9 Potential shelter locations, including schools, recreation centers,
Salvation Army shelters and sports centers
3.2 Methodology
The methodology has two parts: hazard modeling using HAZUS and network analysis using
ArcMap. HAZUS results provided input for the ArcMap analysis step. HAZUS used earthquake data to
24
predict damage from a hypothetical earthquake scenario, in particular which bridges would be too
damaged for travel, and how many people would need shelter after an earthquake event. These results
were then exported into ArcMap where a transportation network was built and a location-allocation
analysis performed. outlines the methodology.
Figure 10 Flowchart of the methodology followed. Colors indicate the separate sections of the process
described below.
25
3.2.1. Running HAZUS scenarios
For a scenario to be run, it requires the correct modules to be selected and a study area created.
Tarrant and Dallas counties were loaded separately to accommodate two different fault lines and two
different damage estimates. A scenario can be created from scratch by selecting the Arbitrary option, or
a historic earthquake and its characteristics can be used by selecting the Historical event option.
Under the Hazards menu, a single scenario was created for each fault line. Each scenario used
the same settings. Arbitrary was used instead of historical, as arbitrary allows for a hypothetical
scenario with customized parameters. However, for scenarios in the Central U.S., the ability to
customize is limited. For example, the attenuation menu (shown in Figure 11) does not allow any
definition of fault type, likely due to the area not being known for earthquakes until the last 5 years.
Fault rupture was also not available for specification.
Figure 11 Attenuation menu
In the Scenario Wizard, the parameters for the earthquake on each fault were defined in the
following steps: 1) fault lines were loaded into the scenario map, 2) the epicenter was chosen on the
26
fault line, 3) magnitude was defined as 5.5, and 4) the depth was entered as 7 km (the average depth of
historic earthquakes in the area). The fault lines for the study area were loaded into this map simply to
guide placement of the epicenter which was chosen by clicking a point on the fault. The Azle fault was
modeled only using Tarrant County, and the Irving fault only using Dallas County.
Once the scenarios were set, the Run menu was opened. All options for buildings other than
user-defined and military installations (of which there were none) were selected as the interest was in
the effects on all types of buildings and structures. For each county, the analysis took 7 to 10 minutes to
run.
3.2.2. Extracting data from HAZUS results
After the results are ready in HAZUS, the map is not automatically shown as there are many
options and variables that can be displayed. In the Results menu, options are given for every category.
Any of these can be mapped by opening the attribute table selecting a column, and hitting the Map
button. The maps produced in HAZUS are simple, such as shelter needs per census tract. More
complex analysis requires ArcMap itself. For this project, Shelter needs and Road Transportation
results were mapped and then exported into separate shapefiles for each scenario. The data inside these
shapefiles include location, name, damage severity, percentage of damage that was a certain severity or
above, and whether it is expected to be functional after a few days to a few weeks.
Bridges are the most important structure in determining which shelters are accessible, so
determining at what damage threshold a bridge is unusable was critically important. Bridge data could
only be acquired after running the scenario by mapping an attribute which creates a layer that includes
all the data, and then exporting the whole file into shapefile format. In the bridge attribute table
produced by a HAZUS model run, damage is categorized as slight, moderate, and severe. There are
also fields that show the total percentage of damage at and above a severity level, such as At Least
27
Slight. For this project, bridge usability was determined by the At Least Moderate field with a value of
20% damaged. A separate shapefile for damaged bridges was created by selecting only the bridges with
damage above the threshold stated previously from the overall bridge count. The road shapefiles
include bridges as part of the road so including all bridges was not necessary. The damaged bridges, as
well as shelters and census tract centroids, were then snapped to the combined road and highway
shapefile. All this now allowed the creation of a network dataset, which is necessary for location-
allocation modeling.
3.3 Using ArcMap for Network Analysis
Once HAZUS scenarios were run, the results were saved and exported as separate shapefiles,
with all bridges, shelter, and school shapefiles generated for each county. Non-school shelter points
collected earlier were joined with the school shelter points extracted from the HAZUS inventory. The
road and highway shapefiles from the NTCOG were also combined. A lakes shapefile from ArcGIS
Online was included for display purposes because Eagle Mountain Lake is a prominent feature near
Azle and serves as a limiter for population and a barrier to transportation.
3.3.1. Creating the Network
A network allows for analysis that requires traveling, such as identifying routes and locating
best facilities. Networks require road shapefiles, as well as any facilities and objects that may affect the
road network. All point shapefiles must be snapped to the road shapefiles to be read as part of the
network. It uses connectivity settings in order to determine the behavior of each structure on the
network. Properties such as speed and length are used to determine travel time, while impedances block
certain routes. Once the road’s properties and general attributes are set, the network dataset can be
built. The output dataset has junctions, which are the damaged bridges and schools, and edges which
are the roads and highways. System junctions were where the roads intersect.
28
Potential damage to the schools was assessed as a limiting factor to their designation as shelters,
but all had usability ratings of 80% or higher for the first day after the disaster. Thus, given that only
15% of the schools incurred any damage at all, potential damage was not considered relevant when
assessing shelter accessibility. All potential shelters were included in the modeling.
To create the network, a database was built that included the locations of the damaged bridges,
shelters, and road shapefiles for each county. Damaged bridges were selected from the original bridge
shapefile using a moderate damage threshold of 20% and made into a separate shapefile. This was so it
could be treated as an impediment. Generally, bridges were simply part of the road rather than a
separate feature since a bridge just allows a road to cross over an obstacle. However, the damaged
bridges serve as an impediment to travel, completely blocking a route that requires crossing the bridge.
Thus they were junctions in the network acting as barriers. This is because if a bridge is damaged
enough, it is not safe to drive over and alternative routes will need to be found.
Network edges have time and length attributes. Time is used to calculate the fastest route, while
length is for calculating the shortest route. Other considerations normally assigned when building a
network such as one way roads and U-turns were not included as police can override travel directions
in case of emergencies. A section of the resulting network is shown in Figure 12. Damaged bridges
were symbolized on top of junctions to show which junctions were the bridges and which were schools.
29
Figure 12 Small section of the road network
3.3.2. Location-Allocation
Location-allocation calculates the best destination point from each starting point given the
objective of minimizing travel time over the network and accounting for capacity constraints at the
destination. If a facility gets full, it will choose the next closest facility with available capacity. The
built in impedance in the network may alter the destination by preventing the best route from being
used since that route involves a barrier.
For this project, centroids of the census tracts were used as starting points since individual
houses are not available in HAZUS. Thus they serve as a generalization of the starting point for all
people within each census tract and are used to calculate the distance to shelter. The number of people
30
needing shelter in each census tract was assigned to its centroid as the shelter demand quantity. Each
shelter was given a capacity of 250. The damaged bridges shapefile was used as the point barriers so
that travels times would be calculated correctly if it required going around an affected bridge. Prior to
running the location-allocation tool, tract centroids, damaged bridges and shelter points must be
snapped to the road network in order that road connections between origin and destination could be
calculated.
Due to edge effect, it was anticipated that census tracts along the edges of each county would be
allocated to the opposite side of the county where spare capacity may remain. Thus, additional schools
from adjoining counties were exported from HAZUS, and then combined with the Tarrant and Dallas
County schools before they were added to the network.
3.4 Summary
HAZUS was used to produce data to be input to the location-allocation model in ArcMap. An
earthquake scenario was created for each fault. Bridge damage and shelter needs were acquired after
the scenario was run. After exporting the bridge and school data, the schools were combined with the
other potential shelter points collected earlier, and a road network was created that included the
damaged bridges as point barriers. Census tract centroids were created and snapped to the network and
a location-allocation analysis was run using the centroids as starting points and the set of potential
shelters as the destination points. The results are shown in the next chapter.
31
Chapter 4 Results
This chapter is split into two parts: the results from HAZUS and the results from the location-allocation
tool. HAZUS results include how many people need shelter, as well as the location of and damage
estimates for bridges. Location-allocation results show the closest shelter location by travel time from
each population starting point and available capacity.
4.1 HAZUS Results
Scenarios were run for both faults. Comparing the results, the 5.5 magnitude earthquake has a
higher predicted impact on the Irving fault area in Dallas County area than on the Azle fault area in
Tarrant County. Many of the bridges in Dallas County were predicted to have damage, with a large
portion of the bridges in the northwest corner having a predicted damage rating of 70% for at least
moderate damage, as seen in Figure 13. The fault runs right through this heavily populated area (shown
in the upper left in the figure), which is home to multiple highways and a large amount of local streets.
Figure 14 shows the number of people in each census tract in Dallas County who may need
shelter. This map is classified using natural breaks as there were breaks in the data, with people
needing shelter jumping from 173 to the next closest number at 264. This was adjusted so that the
lowest value is two or less since it was decided that with only a small number in a tract, they are not
likely to need help finding shelter. Overall, the scenario estimated that a total of 12,820 people in this
county would need temporary shelter if this event happens on the Irving fault. Shelter needs were
projected to be significant in some areas, with one census tract possibly having 279 people with short
term shelter needs, although this was also the largest tract. Three smaller tracts also had over 175
possibly needing shelter. On the other hand, most of the tracts have under 175 people needing shelter,
which allows multiple tracts to be assigned to the same shelter with a capacity of 250.
32
Figure 13 Bridge damage in Dallas County
33
Figure 14 Shelter needs in Dallas. Map shows the number of people requiring shelter in each tract.
In Azle, the damage to the infrastructure and the population at large was predicted to be less
severe. This is due to less population overall (10,000 as of 2010 Census), less population density, the
lake being a natural barrier to growth and less transportation infrastructure. There is a distinct lack of
bridges near the fault area in the northwest as well. As Figure 15 shows, the vast majority of bridges
are predicted to be fine, but there are some areas where the bridges were given a moderate damage
rating of 70% or more. A total of 1,011 people will need shelter as a result of the hypothetical
earthquake event, with the highest number of 56 in one census tract (Figure 16). This map was also
34
classified using natural breaks, as there were breaks in the data, going from 15 to 20 and again from 27
to 41 of people needing shelter.
Figure 15 Bridge damage in Tarrant County
35
Figure 16 Shelter needs for Tarrant County. Map shows the number of people requiring shelter in each
tract.
4.2 Allocation of Population Centers
While most tracts were allocated to shelters, several in Tarrant County and many in Dallas
County were not. In Tarrant County, the remoteness of the area hit means there are less ways to get
around, and thus one is more likely to be heavily affected if a bridge is unusable. Figure 17 shows
several census tracts were not matched, even with the inclusion of shelters outside the county. Census
tracts in the remote areas close to where the earthquake is located have low population densities of
199.5 and 402.1 per square mile respectively. The most heavily damaged bridges are around this area,
36
as seen in Figure 15. The shelters that would be closer also most likely filled up from other, closer
census tracts. The map output shows all candidate shelters, with the designated destinations symbolized
differently. It also includes allocation lines to show which census tract was allocated to which shelter.
Unallocated starting points are also shown.
Figure 17 Allocation of tract centroids to shelters in Tarrant County
Dallas County also had unmatched points, as seen in Figure 18. These tracts were not edge
cases, so it is likely the accessible shelters filled up. Many, however, have moderate shelter needs, as
seen in Figure 19, so identifying shelters reachable by foot or ideal locations for more shelters would be
the next step. Most other census tracts with moderate to high shelter needs were allocated within one to
two tracts, so shelters would be reachable. Edge tracts were assigned to outside shelters, but many of
37
them are far away from the origin point. Several of these have moderate shelter needs, so additional
shelter locations may need to be identified in the future to avoid long travel times.
Figure 18: Allocation of tract centroids to shelters in Dallas County
38
Figure 19: Unallocated points in crucial census tracts
Some census tracts in both Dallas and Tarrant Counties were allocated outside, and quite far
into, those other counties. This analysis was conducted on time in seconds, not on length of travel. Due
to highways having an average of 30 miles per hour increase over normal roads, these may have been
quicker once the closer ones filled up. This may be because frequent stops on slower roads due to
traffic lights and stop signs may not be as efficient as a fast, continuous highway if the highway is not
blocked. Figure 20 shows how the tracts are allocated when using meters instead of seconds.
39
Figure 20: Allocation of census tracts by distance in meters
4.3 Summary
The HAZUS analysis showed that damage from these earthquake scenarios could be significant.
Dallas County had many of its bridges damaged, and over 12,000 people would need shelter. While
many of the shelter needs in census tracts that were impacted moderately or above were able to be
allocated, others were not. This may be due to capacity being filled at all the reachable shelters. Azle
was less impacted, likely due to a decreased population density, but 1,000 people still needed housing
in shelters. There are far fewer bridges in the area but the ones that are there had mostly moderate or
above damage. Some census tracts in Tarrant County were also unallocated, also likely due to capacity
issues combined with whether it could be reached at all. Census tracts allocated far into other counties
40
may be due to it being faster to travel to them by highway with no damaged bridges rather than use
many small roads to shelters that are less far away. These results may have significant implications if
an event were to actually occur, and this is discussed in Chapter 5.
41
Chapter 5 Conclusion
HAZUS showed that there is potential for damage from a moderate earthquake on the newly
discovered faults. From this, the combination of HAZUS and location-allocation can be useful to
understand shelter needs as a result of the simulation, and which shelters should be used. Results from
HAZUS were incorporated into ArcMap to build a road network and run the location-allocation model.
This allowed a visualization of which shelters were allocated to which census tracts, and whether some
could not find any available shelter. Such an approach can assist city emergency response planners to
better address their community shelter needs in the event of a disaster, such as the earthquake in this
project.
5.1 Implication of these results
Running HAZUS showed that the damage from a 5.5 magnitude earthquake could be significant
to the road network as well as to residential homes. A total of over 12,000 were predicted to need
shelter in Dallas County, compared to just over 1,000 for Tarrant County. Many more bridges were
predicted to suffer from a significant amount of moderate or above damage in comparison as well.
Bridges are a crucial component in transportation, and are especially vital in areas such as North
Texas that rely on a lot of highways and have a dense road network. Running HAZUS on a simulated
5.5 earthquake showed the possibility of many of the bridges in Dallas County suffering significant
moderate damage. This was assumed to have made the bridges unusable in the analysis conducted. If
the simulation holds true, strengthening key bridges should then be prioritized to increase route
availability. If the bridge data in remote areas are accurate in where they are damaged, local
government could send out engineers to determine if it is possible at all to get people across. This is
because simulation is not necessarily fact, and bridges may be affected differently than the simulation
implies. There could also be routes built specifically for such emergencies that wouldn't need to travel
42
over or under bridges at all. This would allow people to travel to shelters without encountering road
blocks from out of service bridges.
While most census tracts were able to be matched, several were not and this is concerning.
Damaged bridges in remote areas where there aren't a lot of alternative routes had a significant effect.
Including shelters from outside the county did not seem to help either these or the census tracts on the
edge of Tarrant county, even though it is the next most logical step if local shelters are unavailable.
This suggests a need to address transportation issues at the local level to get to acceptable shelter.
Otherwise, most of the other census tracts with heavy shelter needs were able to find shelter within a
reasonable distance of one to two census tracts. Census tracts being allocated farther away than one or
two census tracts may suggest that it would be better to open up new shelters close by that wont be
slower to get to than hopping on a useable highway and driving farther, but quicker. Local governments
could identify which structures would serve well as shelters that also avoid as many bridges as
possible. If none could be identified, they could also build or designate additional structures that
community's needs. This last is especially important for rural areas, and may be the main way to
address the completely unallocated census tracts. Overall, the results showed there may not be enough
shelters without extensive driving.
5.2 Limitations
The project does have some limitations. Primarily, scenarios and models can only approximate
damage and its impact, with the accuracy depending on how much the data reflects reality in ground
movement and the composition of buildings. It cannot say what will happen, only what can happen. It
is then up to city planners whether the possible risk is worth acting on or not. HAZUS inventory is
extensive, but it is updated only every few years. This limits accuracy without extensive local
knowledge, and so damage estimates may be more inaccurate than what would be acceptable for
43
municipalities. FEMA does allow user input into HAZUS and adjusting the state dataset though their
Comprehensive Data Management System (CDMS) tool, but that was not utilized in this project.
Ideally, this project would have used that tool to make a more accurate representation of the study area.
In addition, it would be beneficial to include the counties around the study area to see how far the
damage may have extended into those areas.
The location-allocation tool does not provide the routes between source and shelter, only which
available location is the closest destination. Including the routes through use of other tools would be
beneficial in a future study for those who may not know how to get to the suggested designated shelter
for their census tract.
The results for distance versus time location-allocation show that there is a large discrepancy on
closest available shelters between the methods. This project focused on time as the measure of
closeness, which would be appropriate for people with vehicles. The distance solution that was briefly
tested did allocate some people more than a census tract or two away, which might be a very long walk.
Future efforts should explore the distance solution further in an effort to constrain the travel distance by
foot for those without a vehicle.
The value of this research is in informing city planners of what may happen in the event of a
moderate earthquake, where acceptable shelters are for each census tract, and if there would be any
census tracts that may need additional shelters from not being allocated to any current shelters. Census
tracts that were unallocated would be the most important to address, with the planners taking steps to
address the problem. The HAZUS simulation can also be used by engineers to identify bridges that
may be issues after a disaster, and to fortify them before an event occurs. Overall, it serves as a way to
demonstrate to city planners the possible effects of an earthquake, possible shelter needs, and where or
if people can get to shelter
44
5.3 Future Research
There are several ways that further studies can expand on this project. While this document
focused on the DFW area, Oklahoma has had the most severe and most numerous of the earthquakes
that have occurred. The highest was a magnitude 5.6 44 miles East – Northeast of Oklahoma City
(USGS 2011).With a magnitude close to this project's scenario having actually occurred, identifying
shelters would be very important if the epicenter happened closer to Oklahoma City (OKC), which has
a higher population than the rural area where the earthquake occurred. This project could serve as a
framework to apply the same methodologies to OKC.
The HAZUS portion could also be used to identify possible problematic bridges for engineers to
fix. With American infrastructure not being adequately kept up, and bridges having a C+ grade with the
occasional collapse (Swearington 2015), it is critically important to know what bridges may suffer
enough damage to make it impassable, how important that bridge is to the overall traffic flow, and to
fortify it ahead of time.
While this study is useful as a demonstration of capabilities, in order to turn this study into a
highly detailed, localized Level 3 HAZUS analysis appropriate for full emergency preparedness
exercises, some additional information would need to be acquired and added. These include:
1. Detailed soil maps that can give a better damage analysis resulting from ground shaking, as
the default in HAZUS uses very generalized soil descriptions and classifications.
2. Ground motion equations appropriate to the area should be used instead of the default
provided for the CEUS area.
3. The user should also load their own inventory of local buildings for improved accuracy, as
HAZUS only updates every few years.
Overall, this project showed how it is possible to determine which shelters may be accessible
using the location-allocation tool, with HAZUS providing the damage impact to the road structure and
45
population displacement. Counties are interconnected through their road networks, so it was necessary
to include outside shelters for edge census tracts to travel to if possible. City planners and engineers can
use the study as demonstration of how to assign people to shelters and to identify places where shelters
may not be available, whether due to not enough capacity or too much damage to the road network.
Whether it is better to find a close shelter or to get to a shelter as soon as possible should also be
explored. It is the hope that this project can serve as a framework to encourage other areas vulnerable
to earthquakes where they have been historically very rare, such as Oklahoma City, to explore the
potential damage through tools such as HAZUS and Location-Allocation.
46
References
BBC. 2013. "What is fracking and why is it controversial?" BBC News. June 27. Accessed April 29,
2016. www.bbc.com/news/uk-14432401.
Deshon, Heather, Robert Williams, Brian Stump, Michael Blanpied, Chris Hayward, M. Beatrice
Magnani, and Mathew Hornbach. 2015. Accessed November 9, 2015. "Preliminary Report -
Irving, Texas, earthquake sequence." SMU. Februrary 6.
http://www.smu.edu/~/media/Images/News/PDFs/Irving%20Quake%20Interim%20Report
%2006feb2015.ashx?la=en.
Esri. 2016. Location-Allocation. Accessed April 29, 2016.
http://desktop.arcgis.com/en/arcmap/latest/extensions/network-analyst/location-allocation.htm.
FEMA. 2008. “HAZUS MH Estimated Annualized Earthquake Losses for the United States” FEMA.
April. Accessed April 29, 2016. http://www.fema.gov/media-library-data/20130726-1627-
20490-0934/fema366b.pdf.
FEMA. 2015 Appendex F: HAZUS-MH Data Dictionary. Accessed April 29, 2016.
http://www.fema.gov/media-library-data/20130726-1800-25045-8485/hazus2_appf.pdf.
FEMA. 2016. HAZUS-MH Overview. Accessed April 29, 2016. http://www.fema.gov/hazus-mh-
overview.
Fitzpatrick, Jessica and Mark Petersen. 2016 Induced Earthquakes Raise Chances of Damaging
Shaking in 2016. Accessed April 29, 2016.
http://www.usgs.gov/blogs/features/usgs_top_story/induced-earthquakes-raise-chances-of-
damaging-shaking-in-2016/.
Frolich, Cliff, Eric Potter, Chris Hayward, and Brian Stump. 2010. “Dallas-Fort Worth earthquakes
coincident with activity associated with natural gas production.” The Leading Edge 29(3): 270-
278
Galbraith, Kate. 2013. "Ready (or Not?) for a Great Coming Shale Boom." The New York Times. April
27. Accessed April 29, 2016. http://www.nytimes.com/2013/04/28/us/time-for-texas-to-get-
ready-for-the-shale-boom.html.
Hornsbach, Matthew J., Heather R. Deshon, William L. Ellsworth, Brian W. Stump, Chris Hayward,
Cliff Frolich, Harrison R. Oldham, Jon E. Olson, M. Beatrice Magnani, Casey Brokaw, and
James H. Leugert. 2015. "Casual Factors for seismicity near Azle, Texas." Nature
Communications 6: 6728.
47
Jordan, M.A. 1977. "Chapter 3: The Balcones Fault Zone of Austin." In Guidebook to the Geology of
Travis County. Austin: Student Geology Society of University of Texas. Accessed April 29,
2016. https://www.lib.utexas.edu/geo/ggtc/ch3.html
Kar, Bandana and Michael E. Hodgson. 2008. "A GIS-based model to Determine Site Suitability of
Emergency Evacuation Centers." Transactions in GIS 12(2): 227-248.
Kongsomsaksakul, Sirisak, Chao Yang, and Anthony Chen. 2005. "Shelter location-allocation model
for flood evacuation planning." Journal of East Asia Society for Transportation Studies 6:
4237-4252.
NBCDFW. 2015. "Recent Quakes Reveal Fault Line from Irving to West Dallas: SMU." NBCDFW.
Februrary 6. Accessed April 29, 2016. http://www.nbcdfw.com/news/local/Recent-Quakes-
Reveal-Fault-Line-From-Irving-to-West-Dallas-SMU-291059741.html
Neighbors, C.J., E.S. Cochran, Y. Caras, and G.R. Noriega. "Sensitivity Analysis of FEMA HAZUS
Earthquake Model: Case Study from King County, Washington." Natural Hazards Review
14(2): 134-146.
Petersen, Mark D., Charles S. Mueller, Morgan P. Moschetti, Susan M. Hoover, Justin L. Rubinstein,
Andrea L. Llenos, Andrew J. Michael, William L. Ellsworth, Arthur F. McGarr, Austin A.
Holland, and John G. Anderson. 2015. "Incorporating Induced Seismicity in the 2014 United
States National Seismic Hazard Model - Results of 2014 Workshop and Sensitivity Studies."
U.S. Geological Survey Open-File Report 2015-1070. http://pubs.usgs.gov/of/2015/1070/
Ploeger, S.K., G.M. Atkinson, and C. Samson. 2010. "Applying the HAZUS-MH software tool to
assess seismic risk in downtown Ottawa, Canada." Natural Hazards 53(1):1-20.
Ploeger, S.K., R. O'Connell, G.M. Atkinson, and C. Samson. 2008. "Earthquake loss estimation data
collection and preparation for urban disaster management: A case study from the City of
Ottawa, Canada." Paper presented at the 4th Canadian Conference on Geohazards. Quebec City,
Canada. May 20-24.
Polo, Gina, C. Mera Acosta, Fernando Ferreria, and Ricardo Augusto Dias. 2015. "Location-Allocation
and Accessability Models for Improving the Spatial Planning of Public Health Services." PLoS
ONE 10(2). http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119190.
Salvation Army. 2016. Salvation Army Carolinas. Accessed April 29,2016.
http://www.salvationarmycarolinas.org/programs/programs-that-help/shelter.
Selk, Avi. 2015. "SMU seismologists discuss earthquakes with residents at Irving City Hall." The
Dallas Morning News. January 15. Accessed April 29, 2016.
http://www.dallasnews.com/news/community-news/northwest-dallas-
county/headlines/20150115-experts-discuss-earthquakes-with-irving-residents.ece.
48
Swearington, Jake. 2015. "This Is What Crumbling Infrastructure Looks Like." Popular Mechanics.
July 20. Accessed April 29, 2016. http://www.popularmechanics.com/technology/
infrastructure/news/a16532/america-crumbling- infrastructure-interstate-10-bridge-collapse/
Tarrant County. 2015. "Local mitigation Action Plan - June 2015." City of Fort Worth. Accessed April
29, 2016. http://fortworthtexas.gov/emo/hazard-mitigation-plan.pdf.
Texas Department of Public Safety. 2013. "State of Texas Hazard Mitigation Plan 2013 update."
Accessed April 29, 2016. https://www.txdps.state.tx.us/dem/Mitigation/txHazMitPlan.pdf.
University of Texas. 2016. Geologic Wonders of Texas. Accessed April 29, 2016.
http://www.beg.utexas.edu/UTopia/centtex/centtex_where.html
USGS. 2011. Magnitude 5.6 - Oklahoma: 2011 November 6 3:53:10 UTC. November 6. Accessed
April 29, 2016. http://earthquake.usgs.gov/earthquakes/eqinthenews/2011/usb0006klz/.
USGS. 2015. Soil Type and Shaking Hazard in the San Francisco Bay Area. Accessed November 1,
2015. http://earthquake.usgs.gov/regional/nca/soiltype/.
Wu, Ping, Hong Shao, Kening Wu, and Bin Cai. 2015. "The Study on Optimization Locaiton-
Allocation of Emergency Shelter for Earthquake." Paper presented at the annual meeting of
International Federation of Surveyors. Sofia, Bulgaria, May 17-21.
Young, Michael. 2015. "Houston Area and Dallas-Fort Worth top nation's fastest growing list." The
Dallas Morning News. March 26. Accessed April 29, 2016.
http://www.dallasnews.com/news/metro/20150326-dallas-fort-worth-trails-only-houston-area-
on-nations-fastest-growth-list.ece.
49
Abstract (if available)
Abstract
Earthquakes, which were previously rare events in the Dallas-Fort Worth (DFW) Metroplex, have become increasingly common in the last five years. In a five-month period, Irving alone had 26 earthquakes over magnitude 2.0. Damage has been minimal, but larger earthquakes have not been ruled out as new fault lines have been discovered and their precise structure is only beginning to be understood. This project's goal was to show how HAZUS can be used to demonstrate possible earthquake effects on the new fault lines, and how the results from HAZUS can be used to assess the impact of damaged bridges on the road network and shelter accessibility. Two fault lines discovered in the last couple of years were digitized and loaded into the HAZUS model. Historical earthquake data were used to form a hypothetical earthquake scenario that characterizes local conditions. The scenario was run twice, once on each fault. ❧ To explore how these results can inform emergency planning, output from the HAZUS scenarios regarding the amount of people needing shelter per census tract, as well as damage to bridges and their location, was imported into ArcMap. A road network was built to support a location-allocation model to assign people needing shelter to potential shelters and the damaged bridges were added as barriers. A centroid was calculated for each census tract to provide population source points. Lacking data on the location of existing emergency shelters, all schools throughout the two counties were designated as potential shelter destinations. Then location-allocation analysis was conducted on each county’s data to determine the closest accessible shelters with available capacity. The demonstration scenario resulted in not enough shelters, as several source points were unallocated. It is hoped that the demonstrations provided by this study will encourage city planners to begin to address emergency planning in the region for these newly identified hazards.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Preparing for the next major southern California earthquake: utilizing HAZUS with soils maps and ShakeMaps to predict regional bridge damage and closures
PDF
Risk analysis and assessment of non‐ductile concrete buildings in Los Angeles County using HAZUS‐MH
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
Modeling potential impacts of tsunamis on Hilo, Hawaii: comparison of the Joint Research Centre's SCHEMA and FEMA’s HAZUS inundation scenarios
PDF
Assessing the impact of a web-based GIS application to promote earthquake preparation on the University of Southern California University Park Campus
PDF
Dasymetric mapping of building stocks within HAZUS-FL
PDF
Generating trail conditions using user contributed data through a web application
PDF
Analyzing earthquake casualty risk at census block level: a case study in the Lexington Central Business District, Kentucky
PDF
Spread global, start local: modeling endemic socio-spatial influence networks
PDF
Installing public electric vehicle charging stations: a site suitability analysis in Los Angeles County, California
PDF
A model for emergency logistical resource requirements: supporting socially vulnerable populations affected by the (M) 7.8 San Andreas earthquake scenario in Los Angeles County, California
PDF
Designing an earthquake preparedness web mapping application for the older adult population of Los Angeles, California
PDF
Preparing for immigration reform: a spatial analysis of unauthorized immigrants
PDF
Network accessibility and population change: historical analysis of transportation in Tennessee, 1830-2010
PDF
A unified geodatabase design for sinkhole inventories in the United States
PDF
Finding environmental opportunities for early sea crossings: an agent-based model of Middle to Late Pleistocene Mediterranean coastal migration
PDF
Modeling patient access to point-of-care diagnostic resources in a healthcare small-world network in rural Isaan, Thailand
PDF
Demonstrating GIS spatial analysis techniques in a prehistoric mortuary analysis: a case study in the Napa Valley, California
PDF
Out-of-school suspensions by home neighborhood: a spatial analysis of student suspensions in the San Bernardino City Unified School District
Asset Metadata
Creator
Curtis, Crystal Eden
(author)
Core Title
Preparing for earthquakes in Dallas-Fort Worth: applying HAZUS and network analysis to assess shelter accessibility
School
College of Letters, Arts and Sciences
Degree
Master of Science
Degree Program
Geographic Information Science and Technology
Publication Date
06/23/2016
Defense Date
05/06/2016
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
Bridges,Earthquakes,hazard prepardness,HAZUS,network analysis,OAI-PMH Harvest,shelters
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Kemp, Karen K. (
committee chair
), Swift, Jennifer N. (
committee member
), Yang, Wei (
committee member
)
Creator Email
crystaec@usc.edu,crystalc13@hotmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-257112
Unique identifier
UC11281124
Identifier
etd-CurtisCrys-4471.pdf (filename),usctheses-c40-257112 (legacy record id)
Legacy Identifier
etd-CurtisCrys-4471.pdf
Dmrecord
257112
Document Type
Thesis
Format
application/pdf (imt)
Rights
Curtis, Crystal Eden
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
Tags
hazard prepardness
HAZUS
network analysis
shelters