University of Southern California Dissertations and Theses

Exploring global natural disaster and climate migration data: a Web GIS application

(USC Thesis Other)

Exploring global natural disaster and climate migration data: a Web GIS application

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Exploring Global Natural Disaster and Climate Migration Data: A Web GIS
Application

by

Declan Forberg

A Thesis Presented to the
Faculty of the USC Dornsife College of Letters, Arts and Sciences
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
Master of Science
(Geographic Information Science and Technology)

December 2020

Copyright ® 2020 by Declan Forberg

ii

To Cecelia, Julia, my family, best friends, and Hannah: Thank you for everything.

iii

Acknowledgements
I am extremely grateful for the guidance provided by Dr. Bernstein throughout my thesis, as well
as my committee members Dr. Ruddell and Dr. Vos for their earnest and thoughtful feedback.
Lastly, I am thankful for the Internal Displacement Monitoring Centre for not only providing the
data for this project, but for continuing to fight for and care about the millions of people affected
by climate change.

iv
Table of Contents
Acknowledgements ........................................................................................................................ iii
List of Tables ................................................................................................................................. vi
List of Figures ............................................................................................................................... vii
List of Abbreviations ..................................................................................................................... ix
Abstract ........................................................................................................................................... x
Chapter 1 Introduction .................................................................................................................... 1
1.1 Motivation ............................................................................................................................2
1.2 IDMC Dataset ......................................................................................................................4
1.3 Project Goals ........................................................................................................................6
1.4 Methods ................................................................................................................................8
Chapter 2 Related Work ................................................................................................................ 10
2.1 Benefits of Providing a Spatial Context .............................................................................10
2.2 Web Map User Interface and User Experience ..................................................................12
2.3 Mapping Natural Disasters ................................................................................................13
2.4 The Importance of Preparing for Displacements ...............................................................16
Chapter 3 Methods ........................................................................................................................ 20
3.1 Data Cleanup ......................................................................................................................20
3.2 R Script ..............................................................................................................................22

v
3.3 Data Parsing .......................................................................................................................23
3.4 Data Mapping .....................................................................................................................27
3.5 Web Application & Story Map Creation ...........................................................................33
Chapter 4 Results .......................................................................................................................... 36
Chapter 5 Conclusion .................................................................................................................... 45
5.1 Summary of Web Application Development .....................................................................45
5.2 Difficulties Encountered During Development .................................................................46
5.3 Future Development ...........................................................................................................47
References ..................................................................................................................................... 50

vi
List of Tables
Table 1. A description of each field within the master IDMC dataset ......................................... 21

vii
List of Figures
Figure 1. The IDMC’s data exploration tool .................................................................................. 5
Figure 2. GDACS 4-day disaster alert map .................................................................................. 14
Figure 3. BestPlaces map showing US cities with highest risk to experience natural disasters ... 15
Figure 4. A snapshot of the master dataset compiled by the IDMC ............................................. 21
Figure 5. Snapshot of the 2008 sub-dataset for the total IDP per country category ..................... 24
Figure 6. Snapshot of the 2008 sub-dataset for the total natural disaster event category ............. 25
Figure 7. An example of the point data produced from geocoding the sub-datasets .................... 28
Figure 8. Pop-up for Guatemala’s IDP Total symbol for the year 2008 ....................................... 30
Figure 9. The configuration of the pop-up title for the IDP Total layers ...................................... 30
Figure 10. Pop-up for the United States Natural Disaster Total symbol for the year 2008 .......... 31
Figure 11. The Web AppBuilder displaying the theme, web map, and widget options ............... 34
Figure 12. View of the web application with the IDP Total Layer List widget activated ............ 37
Figure 13. The web application’s time slider activated for Total Natural Disasters layer ........... 38
Figure 14. View of the Caribbean islands’ total IDP per natural disaster type in 2018 ............... 39
Figure 15. View of the Summary widget while utilizing the filter feature. .................................. 40
Figure 16. The Natural Disaster Total Summary widgets found in the “more” tab. .................... 41

viii
Figure 17. Summary widget used in Pacific island countries for the 2018 dataset ...................... 42
Figure 18. View of the Bookmark widget and the original extent shortcut. ................................. 43
Figure 19. The landing page for the web application on ESRI’s Story Map. ............................... 44

ix
List of Abbreviations
GDACS Global Disaster Alert and Coordination System
GIS Geographic information system
SSI Spatial Sciences Institute
USC University of Southern California
IDMC Internal Displacement Monitoring Centre
IDP Internally Displaced Person/People
IOM International Organization for Migration
NCDC National Climatic Data Center
NOAA National Oceanic and Atmospheric Administration
NRDC National Resources Defense Council
USGS United States Geologic Survey

x
Abstract
Natural disasters have always influenced migration, whether international or within one
country’s borders. However, as the effects of climate change continue to cause irregular weather
patterns and stronger, more frequent natural disasters, the number of individuals at risk of being
displaced from their home due to natural disasters is poised to substantially increase. Given the
millions of people on the brink of needing to relocate due to natural disasters, as well as the
potential billions of dollars needed to repair the resulting damages, there exists a need to better
understand trends in weather and migration patterns. Such an understanding would allow for
governments and emergency response teams to be more prepared to face sudden onset disasters.
The Internal Displacement Monitoring Centre (IDMC) published a dataset detailing the number
of internally displaced people (IDP) per country per year between 2008-2018, and the specific
natural disaster event associated with each IDP. This project utilized the IDMC dataset to create
a web map application using ArcGIS Online that will organizes and visualizes the data in a
spatial context. The original IDMC dataset was broken down into smaller thematic datasets using
the R programming language, which were subsequently visualized using the ESRI products
ArcGIS Pro and ArcGIS Online. The application was designed for ease of use, thus allowing for
new trends and potential patterns to be discovered far more easily. The resulting web application
includes widgets and tools that allow users to manipulate the dataset in meaningful ways unique
to their needs.

1
Chapter 1 Introduction
Historically, natural disasters have always played a role in human movement and where
populations choose to live. Natural disasters have devastated entire cities and can demolish the
homes of hundreds of thousands of people in a matter of minutes. We see disasters dramatized in
summer blockbuster movies: giant waves crashing over New York City, volcanic eruptions
wiping out entire villages, or an earthquake splitting California in half. Seemingly every few
years, news networks dive into non-stop coverage about a major earthquake or tsunami,
providing updates on the death toll every few hours. Pop culture may lead individuals to believe
natural disasters are limited to large and deadly events. The reality is somewhere in between –
while some natural disasters like the ones seen on the news are colossal in nature and carry high
death tolls, natural disasters happen much more often than just once or twice a year and bear a
multitude of different socioeconomic consequences in addition to injury and death (Coronese et
al. 2019). Around the world, people and the economies they inhabit are affected by natural
disasters of all scales and types multiple times a month. The resulting socioeconomic damages
are not always covered in the news and are rarely ever seen in theaters. Nonetheless, millions of
people all over the world feel the direct effects of these damages, and with the constantly
increasing effects of climate change, countries are experiencing natural disasters at an even more
frequent and costly rate (NOAA 2020).
In order to decrease damage caused by natural disasters and create better safeguards to
protect vulnerable populations, there needs to be a better understanding of data surrounding
natural disasters. Governments, international agencies, and scientific organizations have all been
tracking the occurrence of natural disasters and their subsequent effects on society. When
properly utilized, the datasets produced by these different entities can begin to provide insights

2
into important questions. For example: which types of natural disasters are occurring the most
frequently, and where? How many individuals are being displaced as a result? What patterns and
trends are emerging from this data? Introducing an interactive and spatial representation of these
otherwise dense, table-based datasets allows for these questions to be answered much more
quickly and easily. This thesis introduces a web application aimed at providing a visual and
spatial analysis of natural disaster events between 2008-2018, including which countries are
experiencing the highest levels of population displacement (measured as internally displaced
people, or IDP), and a ten-year historical analysis of which weather-related events are occurring
in each country during this specific time period. The foundation of the web application is built
upon data compiled by the Internal Displacement Monitoring Centre (IDMC). The web
application is designed with a simplistic, accessible user interface to ensure a wide audience can
interact with the extensive IDMC dataset – government agencies, international refugee and
disaster relief organizations, emergency response teams, journalists, or even interested private
citizens should all be able to utilize the application with little to no prior GIS experience.
1.1 Motivation
Since 2008, an annual average of 25.3 million people was displaced from their homes as
a result of natural disasters such as flooding, earthquakes, storms, and volcanic activity (Internal
Displacement Monitoring Center 2019). As the effects of climate change continue to worsen, the
frequency of natural disasters continues to grow on a yearly basis, and on a global scale (Centre
for Research on the Epidemiology of Disasters 2019). The increased severity of natural disasters
such as heat waves, severe winter conditions, and drought have the potential to destroy billions
of dollars’ worth of infrastructure populations depend on for survival, including crops and
livestock (Zagorsky 2017). Millions of people are therefore poised to be displaced from their

3
homes as a result of these socioeconomic repercussions. While individuals being displaced by
natural disasters is reason enough to explore the data further, there also exists an immense
economic impact. In the United States, for example, the National Oceanic and Atmospheric
Administration (NOAA) produced a report showing the number of natural disasters causing over
$1 billion dollars in damages has steadily increased since 1980. For instance, any given year in
the 1980s would typically experience an average of about 2.7 disasters resulting in over $1
billion dollars in damages (inflation adjusted). Conversely, the last five years in the US (2015-
2019) saw an average of 13.8 events per year. Furthermore, since 1980 the US experienced 258
weather-related natural disaster events that caused over $1 billion in damages. The cumulative
cost of these damages exceeds $1.75 trillion (Smith 2020). Without proper analysis of natural
disaster and displacement data, the task of accurately allocating resources to mitigate the human
and economic risks associated with natural disasters becomes increasingly difficult.
A cumulative, global economic cost resulting from natural disaster damages may not be
as easily quantifiable globally as it was in the US. However, the effects of natural disasters on
humans have been carefully studied and monitored by the IDMC, a Swiss agency that is part of
the Norwegian Refugee Council. The IDMC partners with national governments, UN agencies,
and other various international organizations to provide accurate data regarding the number of
IDP on a global scale each year as a result of natural disasters (IDMC 2019). The data produced
by the IDMC encapsulates the magnitude and scale of the humanitarian effects of natural
disasters – millions of individuals are displaced each year in multiple countries throughout the
world, and in certain countries, the number continues to increase. There is no simple solution that
will reduce the number of IDP, as natural disasters are inevitable, and the causation behind each
IDP is extremely complex. However, the approach and response to each natural disaster can be

4
improved if each country prepared the necessary resources and infrastructure to vulnerable
communities. This process starts at the statistical level – governments and emergency response
organizations must first know which types of natural disasters are most likely to affect their
communities, and how many individuals are at risk. The web application outlined in this thesis
offers a tool that transforms a large Excel spreadsheet into an interactive, visual map. Such a map
can help users decipher large amounts of information much easier than when presented in
traditional spreadsheet formats.
1.2 IDMC Dataset
The foundation of the web application is a dataset created by the IDMC. The organization
collected data for every country over a ten-year period (2008-2018) documenting each natural
disaster event and the number of IDP associated with each event. Each row in the dataset
represents a natural disaster event. The IDMC monitors and reports all natural disaster events by
partnering with different sources and confirming the disaster event data with government
agencies, UN agencies, news outlets, civil society organizations, and various other international
organizations (IDMC 2019). By partnering with these various organizations, the IDMC is able to
cross-check the disaster and IDP data among multiple sources in order to confirm the data’s
accuracy. The natural disaster events were collected on an event-by-event basis, with no specific
criteria reported by the IDMC to constitute whether or not a disaster event should be added to the
dataset. Given the IDMC’s website did not specify the criteria for a disaster to be included in the
dataset, there are disaster events documented in the dataset that did not result in any
displacements. As a result, there are rows within the dataset detailing natural disaster events that
have 0 associated IDP.

5
The IDMC’s dataset is available as a downloadable spreadsheet containing over 6,400
unique records that lacks any type of spatial reference or tool that would allow for users to easily
gain insights into any spatial or temporal patterns or trends within the data. The dataset can be
downloaded from the IDMC’s website in an Excel spreadsheet format. Table 1 in Section 3.1
contains a detailed breakdown of the 8 columns of information contained within the dataset. The
IDMC does provide a data exploration tool allowing for users to generate their own charts based
on the data held in the spreadsheet (Figure 1). This tool provides users no spatial reference,
however, and can only generate two-dimensional charts on an x and y axis. Given that there are
over 100 countries included in the dataset, the exploration tool is also difficult to interpret with
multiple countries overlapping one another. As a result, the charts are challenging to read and do
not provide a user-friendly experience when attempting to understand any one country within the
group. Thus, there exists a need for an improved visual interface for the dataset.

Figure 1. The IDMC’s data exploration tool (Source: IDMC)

6
1.3 Project Goals
The IDMC dataset in its current form is difficult to filter and organize without breaking it
up into separate, smaller datasets. The web application detailed in this thesis builds upon the
dataset and created a web map offering a singular and more visual, intuitive, and interactive
method of data exploration. The detailed methodological chronology can be beneficial for a
broader audience, rather than just experts in the field of migration, disaster relief, or database
administrators. The symbology, pop-up information, and widgets included on the web
application provide an in-depth examination of the data. The web map will also allow for users
with little to no knowledge of geographic information systems (GIS) to utilize one with relative
ease due to the simple layout, symbology, and widgets.
This web application was built with multiple different audiences in mind. Potential end
users can span from government officials, international agencies such as the IDMC or United
Nations, emergency response teams, educators, students, journalists, or simply an interested
private citizen. By utilizing GIS, the dataset is transformed into an interactive tool that allows
users to explore and filter the data however they see fit. For example, governments can use the
application to analyze trends in the number of IDP produced within their borders, as well as
surrounding countries’ borders. The ability to quickly decipher these types of trends will allow
for government and international agencies to be more prepared to allocate resources when
assisting IDP. Natural disaster trends can be discovered and analyzed to assist emergency
response teams as well. Knowing the type, location, and frequency of natural disasters – as well
as the number of people displaced as a result of each event – would allow for the managers of
emergency response teams to better anticipate natural disaster events and which populations are
at risk. Examining IDP totals throughout the ten-year time frame also allows for both

7
government agencies and emergency response teams to prepare for adequate resource allocation
to aid those affected by natural disasters.
When properly organized and visualized, the information in the IDMC dataset should be
able to provide these types of organizations with the insights necessary to build or improve upon
a robust disaster relief framework. The ability to quickly determine which types of disasters are
having the most effects on a country’s population can, in turn, ideally allow for the improvement
of disaster response infrastructure. However, the design of the web application is not tailored
towards government agencies or disaster relief organizations alone. Instead, the web application
can be utilized as an educational tool for educators and journalists. The user interface allows for
non-experts and otherwise inexperienced GIS users to have the same level of access and
functionality in regard to data exploration. In an educational setting, the web application can
provide students a unique tool to explore natural disaster and IDP data. Rather than approaching
these two subjects from a more traditional perspective – a static, data-driven spreadsheet format
focused on raw statistics – the web application acts a dynamic, interactive, and customizable tool
in which students can discover trends and information based on their own inputs. Outside of the
classroom, journalists can use the web application to supplement their articles about climate
change or refugee crises. The software in which the web application was built (outlined in
Section 1.4) is extremely easy to share – thus, the web application can be embedded into a
variety of media, including online articles and websites. Rather than journalists citing statistics
from the IDMC, they are instead able to offer readers hands-on access to the dataset itself via the
web application. Ultimately, there is not one specific use-case for the web application. The web
map was designed to encompass a wide variety of users, and to provide each user with a useful
and intuitive new means to view an otherwise large, dense dataset.

8
The web application is able to filter the data in numerous different ways that would
otherwise require a script or function when utilizing the original spreadsheet format. Each
attribute within the dataset will be able to be combined or isolated from the other attributes.
Below is a list of functionalities the web application offers the user:
• Time-lapse of natural disaster totals, 2008-2018
• Time-lapse of IDP totals per country, 2008-2018
• Remove categories that do not meet user-specified criteria
• Count the number of instances of natural disaster types in a given year per country
• Count the number of IDP of a given country or region in a given year
• Count the number of IDP per natural disaster of a given country, countries or region in a
given year
• View the number of IDP produced by specific type(s) of natural disaster(s)
• View charts containing IDP per natural disaster and natural disaster frequency per country
per year
1.4 Methods
In order to create a web map that included the majority of the IDMC dataset while still
maintaining accessibility, separate custom data layers were created using the IDMC dataset.
Each newly created data layer serves to enable the user to dive deeper into the attribute data
within the IDMC dataset, including features like natural disaster category, type, and date. Each
dataset was uploaded into ArcGIS Pro, where it was visualized and then published to ESRI’s
web platform, ArcGIS Online. Next, the data layers were imported into the web application
itself, which was created using ESRI’s Web AppBuilder tool. The Web AppBuilder tool was
utilized for its built-in widgets that allow for users to customize the data layers as they see fit.

9
Lastly, the web application created in Web AppBuilder was embedded into ESRI’s Story Map
software, which contained a brief background on the topic of natural disasters and IDP, as well
as instructions on how to use the web application.
In the web application, certain countries cannot display any information if they do not
meet the user-specified criteria. For example, if users only want to examine natural disaster data,
they are able to remove IDP data from the web map. As a result, countries that experienced
natural disasters will populate the web map – if a country did not experience a natural disaster, it
will not be represented with any symbology on the map.
The following chapter outlines similar research surrounding climate change, natural
disasters, and IDPs, and outlines a more in-depth breakdown of the methodology used to create
the resulting web application.

10
Chapter 2 Related Work
This chapter builds a case for the importance of mapping natural disasters and internally
displaced people by examining previous work and data surrounding the two topics. The ensuing
section will first demonstrate the need for a spatial context in regard to these subjects by
detailing the benefits of data visualization. Section 2.2 details existing natural disaster maps and
their use cases, and Section 2.3 outlines the importance of preparing for migrations as a result of
natural disasters.
2.1 Benefits of Providing a Spatial Context
The IDMC’s foundational dataset used in this thesis’ web application contains rich
attribute data, such as event name, date, hazard type, hazard category, and the resulting IDPs. By
monitoring natural disasters as well as IDPs, this thesis explores which natural disasters are
causing population displacement and visualizes these conclusions within an accessible spatial
context. This is in contrast to a richly populated data table that may otherwise be difficult for
non-experts in the field to uncover trends within the data. The following section examines past
work focused on the phenomenon of climate migration, and how the increased effects of climate
change can result in millions more displaced people throughout the world. As a result, the
section concludes with a need for more research regarding the spatial context of climate
migration.
The Migration Data Portal is a comprehensive resource for projects and case studies on
migration. Funded primarily by Germany and Switzerland via the International Organization for
Migration (IOM), the portal provides summaries on each theme driving immigration (human
trafficking, family migration, gender and migration, labor migration, etc.). The portal contains a
number of publications and case studies surrounding environmental migration since 1980. One

11
major takeaway demonstrates that of the 522 case studies analyzed by the organization, only
11% of them include spatial analysis (IOM 2018). Given that only 11% of case studies include
spatial analysis, the web application outlined in this thesis would provide a more unique
approach to natural disaster and migration data. Thus, if the application were to be utilized by
educators or journalists, the userbase would be exposed to a more distinct and engaging medium
to interact with the large dataset.
The IOM’s summary of environmental migration case studies was a key guide in
determining the gaps in knowledge this application aimed to fill. The goal of this web application
is to build upon the existing natural disaster and internal migration data provided by the IDMC
by adding a visual and interactive spatial component. Overall, the web application as a whole has
an equal emphasis on mapping both IDP and natural disaster trends. In order for a broad
audience to comprehend the intersection of these two phenomena, the application will be in the
form of an interactive web map.
ESRI’s Story Maps team created a Story Map aimed to educate viewers as to the
increased effects of climate change, resulting in increased vulnerability in populations
throughout the world. However, the effects of climate change are exacerbated by a combination
of non-climate related factors such as resource constraints, population growth, and unprepared
governments. The results is that communities are forced to leave their homes in search for a safer
place to live. ESRI’s Story Map, entitled “Climate Migrants”, details case studies of
communities throughout the world at risk of displacement due to the impending destruction of
their homes. Brought on by effects of climate change, this poses a risk to millions of people. The
Story Map presents the data clearly so end users of all backgrounds are able to understand the
issue. Presenting the data in a spatial format means end users without a technical, GIS-based

12
background – such as government decision-makers, for example – can each access and
comprehend the datasets and issues at hand. It is important for government officials in particular
to be able to access and study the intersection of these phenomena as well, because the migration
of millions of people places a burden on the governments accommodating the climate migrants;
certain country’s infrastructure and resources may not be able to handle such a dramatic influx.
As such, this thesis aims to provide a tool that can allow for non-technical government officials
to view an otherwise large and complex dataset, ideally allowing for easier comprehension and a
better understanding of populations at risk.
2.2 Web Map User Interface and User Experience
An important component of any web application is the user interface (UI) and user
experience (UX). As web maps increase in popularity due to convenience, scalability, and
accessibility, it is paramount for the UI to be uniformly accessible from a variety of different
devices and resolutions. However, there is not yet a uniform standard for any given web map’s
interface. The UI is therefore up to the map’s designer, who is tasked with retaining as much
simplicity as possible without sacrificing efficiency (You et al. 2007, 16).
One standard that has emerged in web applications is the grouping of similar tools in the
same area. Users of web applications demonstrate higher accuracy when controls are grouped
together (You et al. 2007, 17). Furthermore, in addition to accuracy, cartographers at a Polish
university conducted an eye-tracking experiment that determined the grouping of similar tools on
a web map increases user speed and overall efficiency. Based on their study’s results, users tend
to first examine the corners of a web map before finding their desired tool of choice (Cybulski
and Horbiński 2020). Thus, in order to design an accessible and efficient UI/UX without
sacrificing any access to the foundational data within the web map, the web map in this thesis

13
was designed to have all tools and widgets grouped as close together as possible. The top right-
hand corner of the web application contains a cluster of the Summary widgets, with each widget
clearly marked with the corresponding year it calculates (as seen in Figure 16). This design
element was done purposefully with the UX in mind. By keeping each widget grouped together,
a non-technical user is able to quickly navigate through the map’s tools and execute functions
with the same efficiency as an experienced GIS-user.
2.3 Mapping Natural Disasters
Natural disasters can occur at any moment. According to the European Space Agency,
earthquakes registering over a 5 on the Richter scale happen about 1,000 times in a year. Thus, it
is important to track their frequency, type, scale, and residual effects. Government agencies and
emergency response teams must be able to quickly sift through the data in order to determine
which earthquakes had tangible effects on a population. The Global Disaster Alert and
Coordination System (GDACS) produces a web map that tracks natural disaster alerts within a
four-day time span (Figure 2). Each natural disaster within the last four days is categorized and
placed in its respective location on the map and is provided a GDACS score in terms of the
overall impact on the surrounding area. Some natural disasters have extremely detrimental
effects to the surrounding area, while others are relatively harmless storms in the middle of the
ocean. Web maps like those created via this project are efficient ways to demonstrate the
frequency and general impact of natural disasters within a short time span and can be useful for
emergency response teams to gauge the severity of such an event. While similar in scope – both
the web application proposed in this thesis and the GDACS web map categorize and assess
natural disaster events – the web map proposed in this thesis aims to provide a more historical

14
look at the frequency and impact of each natural disaster between 2008-2018, while providing
more specific data regarding the number of people displaced within a country’s borders.

Figure 2. GDACS 4-day disaster alert map (Source: GDACS)

Historical natural disaster data can provide insights as to which areas are most at risk
with respect to mortality. BestPlaces, a company focused on assessing and ranking metropolitan
areas based on various indexes, published maps detailing which US cities are most at risk to
experience a deadly natural disaster. Specifically, the maps utilized NOAA, USGS, and NCDC
data to determine which US cities had the highest risk based on population and natural disaster
frequency over the last 30 years (BestPlaces 2011). After analyzing 379 metropolitan areas, the
study found that southern US cities were most at risk to experience natural disasters (Figure 3).
Dallas, Texas was the city at highest risk, followed by Jonesboro, Arkansas, Corpus Christi,
Texas, and Houston, Texas. While this type of map may be useful for citizens of the US, there
are regions of the world at much higher risk due to population density and socioeconomic
factors.

15

Figure 3. BestPlaces map showing US cities with highest risk to experience natural disasters

From a global perspective, IDMC data shows that average annual displacement – the
average number of people expected to be displaced each year – is highest in South and East Asia,
as well as the Pacific island countries such as Indonesia and the Philippines. According to the
IDMC, these three regions account for two thirds of the world’s population displacement risk
(IDMC 2019, 85). Furthermore, an IDMC report finds that displacement risk associated with
natural disasters will primarily affect developing countries. This means that countries with
greater socioeconomic vulnerability are disproportionately at risk to experience population
displacement compared to countries that, while highly vulnerable to natural disasters, do not
have as vulnerable of a low-income socioeconomic population. As a result, many island
countries such as the Bahamas, Antigua and Barbuda, and Dominica are disproportionately at
risk compared to the rest of the world (Anzellini et al. 2017, 12). These findings demonstrate the
need for a global perspective when measuring and analyzing natural disaster events and risk.

16
Thus, the web map presented in this thesis contains a global spatial extent that visualizes natural
disaster type, frequency, and IDP as a result of each event.
2.4 The Importance of Preparing for Displacements
With the impending displacement of millions of people from various regions across the
world due to climate change, there exists a need to protect the livelihood of the climate migrants
while simultaneously strengthening the physical and societal infrastructure of the countries in
which the migrants are relocating (Global Compact for Migration 2018). Climate migrants, and
therefore IDP, should be legally protected – however, legal protection is currently a difficult and
complicated work in progress, as there is not yet an official definition of what it means to be a
climate migrant (Mateo 2019). As there is not yet a legal framework in place to protect climate
migrants and IDP, the responsibility of accommodating individuals affected by natural disasters
falls largely on the governments of countries with IDP. Dina Ionesco, the Head of the Migration
Environment and Climate Change Division at the IOM, notes that climate migration is mostly
internal anyway – that is, individuals displaced by natural disasters are usually moving within
their home country’s borders and are not seeking legal protection at an international level.
Instead, the burden to ensure adequate accommodation for IDP falls on the IDP’s home country.
This thesis presents a GIS-based application that demonstrates the scale and location of each
country’s yearly IDP totals, as well as the natural disaster event associated with each IDP. As
such, state governments can visualize their IDP and disaster totals between 2008-2018, a
relatively recent timeframe, in order to prepare assistance for IDP within their own borders. If a
country’s IDP and natural disaster totals are increasing each year, its government should ideally
organize disaster relief infrastructure and resources accordingly. The internal migration of

17
potentially millions of people would otherwise result in various socioeconomic burdens within
the country’s borders.
The Nansen Initiative is a Swiss organization aimed at unifying countries in favor of
protecting IDP in the context of natural disasters, including events spurred by climate change
(Kälin 2015). One of the key conclusions detailed in the organization’s Protection Agenda was
that IDP are likely to continue increasing in the future, and that while many IDP are able to
return to their original homes, tens of millions of IDP will need protection and assistance in order
to successfully relocate. In 2015, the group published a consultation outlining the magnitude of
issues IDP face, and how countries have an obligation to protect all those that are affected by
natural disaster displacement. The authors of the Protection Agenda suggest that countries
neighboring the regions heavily affected by climate change must create treaties and agreements
to help absorb the millions of climate migrants and IDP that will otherwise place a strain on
resources, jobs, and culture. Furthermore, the publication notes that between 2008 and 2013, an
estimated 27.5 million people were displaced from their native country due to sudden-onset
disasters annually (The Nansen Initiative 2015, 1). While many people are able to return to their
original home, millions are forced to uproot and begin new lives in different parts of their home
country. The Nansen Initiative strives to create an international support system for the latter
group, especially given that climate change will likely drive these numbers up in the future. In
their agenda, the group details a need for more scientific studies tracking IDP so that predictive
models can created and set in place for countries with high amounts of IDP (The Nansen
Initiative 2015, 3). While tracking IDP poses itself as a new and difficult problem, it is
nonetheless possible to identify areas that are high-risk for natural disasters to occur, which can
therefore identify populations at risk of displacement. Identifying these at-risk areas throughout

18
the world aligns with one of the primary goals of this thesis. The web application outlined in this
thesis can offer government organizations a perceptive method to discover trends and insights as
to which countries are experiencing disproportionate numbers of climate-related disasters and the
number of associated IDP. While the web application is more of a visualization tool than a
decision-making tool, the web map is able to organize and present extensive IDP and natural
disaster data in a more meaningful format, which in turn can ideally lead to more informed
decision-making.
Climate-related migration will continue to increase globally unless the effects of climate
change can be curbed. The National Resources Defense Council (NRDC) published an article
summarizing sources that detail just how severely climate change is impacting global migration
(Turrentine 2019). Scientists estimate that the three most densely populated developing areas of
the world – Latin America, South Asia, and sub-Saharan Africa – will potentially see the internal
migration of more than 140 million people before 2050 (Turrentine 2019). This means there is a
need to focus on internal migration within largely populated continents such as Africa or Asia,
rather than focusing only on international, multi-continental migration. The economic and social
implications of 140 million people migrating within these continents are massive; if countries
and cities do not have any predictive tools or an understanding of natural disaster and migration
trends, their infrastructure is at risk to be unprepared to handle the influx of IDP. Countries that
are already densely populated, or have existing socioeconomic issues of their own, will be on the
brink of conflict should they experience sudden population surges within their cities.
Specifically, population surges (especially in countries with existing resource scarcity and poor
government management) can exacerbate the demand for natural resources and housing
infrastructure. Countries and cities that are unequipped to handle such rapid population shifts are

19
therefore risking conflict and violence within their borders (Acemonglu, Fergusson, and Johnson
2019). Thus, there exists a need for tools similar to the web application outlined in this thesis
that examine which countries are experiencing rapid population shifts. GIS-based solutions can
provide government leaders a visual and temporal blueprint for global weather-related disasters
and migration patterns.

20
Chapter 3 Methods
The goal of this thesis was to create a web application that visualizes IDPs due to weather-related
natural disasters over a ten-year period (2008-2018). The foundational dataset used in the web
application, created by the IDMC, contains over 6,000 records of natural disaster events and the
number of individuals displaced as a result; the spatial extent of these records is global – each
country the IDMC collected data for is included in the dataset, and therefore the web application.
This vast dataset contains useful information that, when presented spatially, can offer a better
understanding of weather-related trends and their effects on human movement.
This chapter outlines the steps taken in order to organize the foundational dataset – as
well as the sub-datasets created by breaking down the foundational dataset into three separate
categories – and create a customizable web application that visualizes each of the attributes
associated with the data. The following subsections – Data Cleanup, Data Parsing, Data
Mapping, and Web Application & Story Map Creation – outline each stage of creating the web
application. Each subsection will also discuss the rationale for the methodology used, including
any potential benefits and shortcomings.
3.1 Data Cleanup
The web application was built upon a dataset created by the IDMC. The dataset contains
detailed attribute data in addition to displacement numbers and natural disaster event categories;
a snapshot of the master dataset can be found in Figure 4 below. With 6,432 rows of data, there
was first a need to ensure the data quality was sound, with no gaps, duplicates, or other
miscellaneous errors that would prevent an accurate spatio-temporal presentation of the data.

21

Figure 4. A snapshot of the master dataset compiled by the IDMC
As seen in Figure 4, the row dedicated to each country contains natural disaster attribute
data associated with a specific year. Each disaster event was documented via an individual row,
so if a country had multiple disasters it occupies multiple rows of data. For example, China
contains 53 rows of information for the year of 2016 alone, meaning there were 53 natural
disaster events in China during that specific year. If a country did not experience any natural
disasters in a given year, it was not listed in the dataset. For example, Barbados is only listed in
the master dataset twice, in 2010 and in 2016, corresponding with its natural disaster events in
those years. A description of each field in the IDMC’s dataset can be found in Table 1 below.
ISO3 Name Year Start
Date
Event
Name
Hazard
Category
Hazard
Type
New
Displacements
Country
code.
The
name of
the
country.
The year in
which the
natural disaster
event and
displacements
occurred,
ranging
between 2008-
2018.
The
date the
specific
natural
disaster
event
started.
The name
and/or
description
of each
natural
disaster
event.
Examples:
Tropical
Cyclone
Zena, or
Flash flood
There are
two
categories in
which a
natural
disaster can
fall into:
weather
related or
geophysical.
The specific
type of each
natural
disaster
event. For
example:
flood,
drought,
earthquake.
The amount of
IDPs produced as
a result of each
individual natural
disaster event.

22
in the
French
Riviera.
Table 1. A description of each field within the master IDMC dataset

The IDMC dataset is comprehensive but not perfect, and thus needed to be cleaned
before use. For instance, duplicate countries were found within the master dataset. “Viet Nam”
contained natural disaster events for a given year, as did “Vietnam”. Some natural disasters that
occurred in Russia were named as occurring in the country called “Russian Federation” – these
types of errors needed to be fixed manually in order to create sub-datasets that contained
accurate country names. In other cases, hazard events were removed because they were not
considered natural disasters, and therefore did not fit into one of the 11 Natural Disaster Type
categories. As an example, there was an oil spill listed in Colombia in 2018, and no Hazard Type
was assigned to the event. The event was therefore deleted from the master dataset to ensure it
was not included in any of the sub-datasets that would eventually be visualized on the web
application.
3.2 R Script
In order to accurately and efficiently reorganize the master dataset and break apart the
data into the aforementioned categories, an R script was created. Using R allowed for the sub-
datasets to be created quickly and efficiently, rather than having to create new fields manually,
enter the data row by row, and repeat the process for each year.
The R script takes the master dataset as an input file. To obtain yearly “slices” of the data
from the master dataset, the first pre-processing step was to remove any redundant data, e.g.
Start Date, Event Name and Hazard Category. Subsequently, the data was filtered from the
master dataset for each particular year within the given 10-year timeframe (ensuring any empty

23
cell has the value zero assigned to it). At this point, a data frame in the same format as the master
dataset was created – however, only comprised of data for the year in consideration.
The next task of the R script was to create two additional data frames. The first data
frame maintains the Total Natural Disaster category for each country for a given year, while the
second data frame maintains the Total IDP per Natural Disaster Event category for each country
for a given year. Once these sub-data frames were created, they were merged while maintaining
the name of the country, the natural disaster types occurring for the country for a given year, the
frequency of the hazard types for the country for the given year, and the total IDP produced by
the natural disaster type for a country for a given year. After the merge, there were then several
rows of data per country – each row type unique to a natural disaster type. To collapse the data
even further, the R script needed to allocate one row per country per year. Therefore, a new data
frame was created by adding a column each for a particular natural disaster type and the number
of IDP displaced as a result. The initial values in the columns were then set to zero. From the
previously merged data frame, the frequency of each natural disaster type and the number of
IDPs as a result was recorded, and the values in the corresponding cells were updated in the new
data frame. Thus, the data was successfully made available in separate Excel files. This process
was repeated for all remaining years in the master dataset. Once the process was repeated, each
sub-dataset for its corresponding category was created.
In summation, there were four categories created by the R script: IDP Total per Country
(per year), IDP per Natural Disaster (per year), Natural Disaster Total per Country (per year) and
Natural Disaster Event Count (per year). Section 3.3 summarizes how each resulting sub-dataset
was organized within Excel, and eventually ArcGIS Pro.
3.3 Data Parsing

24
In order to decipher the thousands of rows contained in the master dataset’s spreadsheet
and to visualize the information in a valuable, distinct manner, the master dataset was re-
categorized into sub-datasets, with the ultimate goal of symbolizing the resulting sub-datasets by
category. Each of the four categories contains 11 sub-datasets corresponding with the ten-year
timeframe of the master dataset. For example, there is a sub-dataset for each year between 2008-
2018 that contains the total IDP per country for each specific year. Thus, there are 44 total sub-
datasets used to create the web application – 11 for each of the four categories. Of the 44 sub-
datasets, 22 were uploaded to ArcGIS Online for symbolization.
The figures below display the new fields created for each category’s sub-datasets.
Compared to the master dataset’s fields in Table 1, the new sub-datasets allow for the data to be
presented, and subsequently visualized, in a new and more meaningful way.

Figure 5. Snapshot of the 2008 sub-dataset for the total IDP per country category
A primary goal of the web application was to symbolize the total number of IDP per
country per year as a result of the IDMC’s documented natural disaster events. The creation of
this category requires separating the countries by each year between 2008-2018. Figure 5
displays the newly created sub-dataset that contains the total number of IDP in a given year per
country. Note how countries with no displacements in a given year are left intentionally blank
instead of a ‘0’. As such, these countries will not show up on the map for that specific year,
which reduces visual clutter on the map. The R script was built to include this data category for
the remaining years 2009-2018.

25
The next category contains sub-datasets detailing the number of IDP produced as a result
of each natural disaster event type. These sub-datasets take the total IDP per country information
in the first category (Figure 5) and break down the data by the specific natural disaster event that
caused the displacements. This category offers a more detailed perspective on which of the 11
types of natural disasters found in the IDMC’s dataset are affecting a country’s population
displacement. The column names in this category are as follows: Drought Displacements, Dry
Mass Movement Displacements, Earthquake Displacements, Extreme Temperature
Displacements, Flood Displacements, Mass Movement Displacements, Severe Winter Storms
Displacements, Storm Displacements, Volcanic Displacements, Wet Mass Movement
Displacements, and Wildfire Displacements. Each row represents a different country, and IDP
are attributed to each natural disaster type column. The sum of each row will therefore equal the
country’s total IDP number as seen in Figure 5.
In this category’s case, natural disaster types that did not produce displacements are
represented with ‘0’ rather than left intentionally blank. This is because the category is visualized
on the web map using a pie chart, so whether a 0 or null value is in a certain field, it still will not
be displayed in the pie chart. An important step taken with this category’s sub-datasets was to
merge each sub-dataset with the sub-datasets from the IDP total category as outlined in Figure 5
above. The resulting spreadsheets contained both categories’ information: a country’s total IDP
for a specific year, and a breakdown of the IDP produced by each natural disaster. The categories
were merged in order to symbolize both types of information in a single layer, which was
achieved using pop-ups in ArcGIS Online (this process will be detailed in Section 3.3). Finally,
the R code created each disaster type field and filled in the data associated with the specific years
between 2008-2018.

26

Figure 6. Snapshot of the 2008 sub-dataset for the total natural disaster event category
The next category, as seen in Figure 6 above, contains the total number of natural disaster
occurrences in a country each year. As was the case for the total IDP per country category
detailed in Figure 5, countries that did not experience any natural disasters in a given year
contain a blank field rather than a ‘0’. This category is visualized with graduated symbols and a
time-slider similar to the total IDP category – countries with 0 natural disasters are therefore not
relevant to the map and are not displayed as a result.
Finally, the last category is composed of sub-datasets detailing the amount of natural
disaster types a country experienced in a given year. This category breaks down the total natural
disaster category as seen in Figure 6 and provides a more detailed look at the types of natural
disasters that affected a country in a given year. Each column name in these sub-datasets
represents one of the 11 types of natural disasters included in the IDMC dataset. Each row
represents a country, and the sum of each row will equal the total amount of natural disasters in a
given year for that country. Similar to the second category (number of IDP produced per natural
disaster), this category is displayed as a pop-up when a user clicks on a country’s total natural
disaster symbol on the web map. In order to achieve the pop-up functionality, this category’s
sub-datasets were also merged with another’s – this time, with the natural disaster total category
detailed in Figure 6. Both merges occurred in Excel before the sub-datasets were uploaded into
ArcGIS Online. The symbolization of each subcategory will be discussed in more detail in
Section 3.4.

27
In summation, as seen in the column names in Table 1, the original IDMC dataset
contains detailed information regarding event dates, event names, hazard category, and hazard
type. In order to present this information in a way that is both insightful and accessible, while
still retaining as much of the original detail as possible, the dataset needed to be broken up into
separate categories using R. The R script helped neatly organize the wealth of information into 4
distinct groups. As a result, trends and patterns that are otherwise not easily discernible within
the master dataset’s numerous rows were brought to the forefront of the web map. Section 3.4
delves deeper into the logic and methodology behind the symbolization of each categories’ sub-
datasets.
3.4 Data Mapping
The web application aims to provide users an interactive, visual, and seamless tool in
which they can manipulate the extensive master dataset. In order to achieve this goal, the ESRI
software suite was utilized. Specifically, the software utilized to build the final web application
included ArcGIS Pro, ArcGIS Online, Web AppBuilder, and Story Map. After the total IDP
category was merged with the IDP per event category, the resulting 11 sub-datasets were
uploaded to ArcGIS Pro. The Total Natural Disaster category was merged with the Natural
Disaster Event Count category, and the 11 merged sub-datasets were also uploaded to ArcGIS
Pro. Thus, 22 sub-datasets representing the four categories were uploaded to ArcGIS Pro to
begin the symbolization process. The sub-datasets were imported into ArcGIS Pro as a .csv file,
and subsequently visualized using the Geocode Table tool. The Geocode Table tool produced
point data associated with each countries’ location based on the “Name” field in the .csv file. The
“Name” column included every single country documented in the IDMC’s master dataset,
including countries that may not have experienced any natural disaster events. ArcGIS Pro was

28
utilized for the quickness and accuracy of the Geocode Table tool, as well as the ability to easily
add a temporal component to the dataset by adding a time aspect within each layers’ properties
later on in the symbolization process. After a sub-dataset was geocoded, the results were
reviewed to ensure accuracy. Ten countries required a manual edit to confirm the country’s
accurate location on the map. Figure 7 is a snippet of the ArcGIS Pro basemap displaying the
point data that was produced after using the Geocode Table tool.

Figure 7. An example of the point data produced from geocoding the sub-datasets
Once the sub-datasets were geocoded, each individual layer was assigned a time property
based on the year of the data. No further symbolization was completed in ArcGIS Pro, because
ArcGIS Pro does not allow layers to be shared to ArcGIS Online if they contain any chart
symbolization, including pop-ups. Therefore, once each layer was geocoded and given a time
property, they were immediately shared to ArcGIS Online to complete the majority of the
symbolization.
The web map in ArcGIS Online contains 22 layers: 11 for IDP Totals and 11 for Natural
Disaster Totals. The IDP Total layers were symbolized with six classes of red graduated

29
symbols, ranging from £ 500 IDP to £ 20,000,000 IDP. Because the IDMC’s master dataset
contains such a large amount of detailed information, it is nearly impossible to display all the
information at once on the web application. If all four categories were displayed on the web map,
the result would be difficult to interpret, and the user would be tasked with interacting with 44
different layers. Therefore, in order to ensure all of the IDMC data is displayed as neatly as
possible, pop-ups were configured in order to display the IDP per Natural Disaster and Natural
Disaster Event Count categories. The pop-up feature eliminates the need for these categories to
be displayed as their own separate layers, effectively reducing the overall clutter of the web map.
Thus, each of the IDP Total symbols contain a pop-up displaying a pie chart of the IDP per
Natural Disaster category’s data. The pop-up was configured by using the attribute data from the
IDP per Natural Disaster sub-datasets that were previously merged with the IDP Total sub-
datasets. Figure 8 provides an example of the IDP Total symbology displaying the IDP per
Natural Disaster data in its pop-up.

30

Figure 8. Pop-up for Guatemala’s IDP Total symbol for the year 2008
When a user hovers the mouse over a portion of the pie chart, the type of natural disaster and the
exact number of IDP associated with the disaster are displayed. In Figure 8, the pointer was
hovering over the purple portion of the pie chart, which represents the flood displacement
attribute. Figure 9 outlines the configuration of the pop-up title, which displays the exact number
of total IDP for the country selected. Displaying the exact total provides users a more detailed
look at the data, rather than relying on the graduated symbols. Instead, the graduated symbols are
utilized to provide a more general view of the data when the time-slider is activated.

Figure 9. The configuration of the pop-up title for the IDP Total layers

31
Next, the Natural Disaster Total category’s layers were symbolized. By separating the
natural disaster data from the displacement data, the web map is able to neatly display two
different, yet equally in-depth data categories from the master dataset. Without separating the
sub-datasets, the resulting web map would be entirely too cluttered, difficult to read, and nearly
impossible to glean any useful information. In order to avoid a map with too much data and
various different types of symbolization, the Natural Disaster Total layers were symbolized using
the same graduated symbols as the IDP Total layers, providing the map with a consistent
visualization of the categories. Many countries experienced less than five recorded natural
disasters, according to the IDMC’s data – several countries had only one instance in a given year.
Conversely, a handful of countries experienced over 150 natural disasters in a single year, such
as Indonesia or the United States in 2018. Six classes were created ranging from 1 - 50 and
above to account for the large variance of natural disaster occurrences. Next, a pop-up was
configured that broke down the number and type of a country’s total recorded natural disasters
by using the Natural Disaster Event Count category’s data found in the attribute data of the
Natural Disaster Total layers. Figure 10 details an example of the Natural Disaster Total’s pop-
up information.

32

Figure 10. Pop-up for the United States Natural Disaster Total symbol for the year 2008
In this category’s case, the information was displayed as a column chart instead of a pie chart.
This is because the data for the Natural Disaster Event Count category contains much lower
numbers than the IDP per Natural Disaster category, and the disparity between each number is
also much lower than the IDP per Natural Disaster numbers. For example, if the IDP per Natural
Disaster data was displayed with a column chart, countries with over 1,000,000 IDP as a result of
floods would sky-rocket the y-axis – if that same country had 100 IDP as a result of wildfires,
the column would be barely visible. The Natural Disaster Event Count data is dispersed more
evenly, and contains lower numbers overall, which make the category a sensible candidate to be
displayed with a column chart. The title of the pop-up was configured the same way as the title
for the IDP Total pop-up – the only difference is the total number was pulled from the natural

33
disaster total column in the Total Natural Disasters layer’s attribute table, using
“{USER_Total_Natural_Disasters}”.
Once the symbology and pop-ups for each of the 22 layers were configured, the web map
was complete. The web map and each of its layers’ sharing level was set to “Everyone (public)”,
which allowed for users to access the web application without requiring ESRI credentials. The
last step to complete the web application at this stage was to import the web map into ESRI’s
Web AppBuilder software. Section 3.5 details the creation of the web application itself using this
tool.
3.5 Web Application & Story Map Creation
The steps taken leading into the web application and Story Map creation stage are as
follows: each sub-dataset was created using the R script, then uploaded to ArcGIS Pro where
they were geocoded. Once the sub-datasets were geocoded, they were published and shared to
ArcGIS Online. Each individual layer was imported into a blank web map on ArcGIS Online,
and symbolized accordingly. The final step involved creating the interface for the user, which
was built using ESRI’s Web AppBuilder. Web AppBuilder was selected because it provides a
user interface that is clean and easy to use with little to no prior GIS experience. Web
AppBuilder also offers users the option to use widgets that further filter the data to match their
preference, rather than a static map. Lastly, the finalized web application was then embedded in
ESRI’s Story Map tool, which provided context for the web map’s theme as well as basic user
instructions on how to interact with the data.
First, the web map built in ArcGIS Online was imported into Web AppBuilder. After the
web map was imported into Web AppBuilder, various different widgets were configured and
added to the web application, giving the application an extra layer of customization for the users

34
that would otherwise be missing from a static map created in ArcGIS Pro. Widgets can be
powerful tools that can perform mathematical functions on layers’ attributes and return values
that are otherwise not listed in the layers. The theme chosen for the web application was the
Foldable Theme, due to the fact that it allows for an unlimited number of widgets while still
offering users a simple and clean layout (Figure 11).

Figure 11. The Web AppBuilder displaying the theme, web map, and widget options

The first widget added to the map was the Time Slider widget. Because the time
properties were already configured on the layers in ArcGIS Pro, there was little need for any
customization of the Time Slider widget. The time parameters of each layer were already
configured before they were imported to Web AppBuilder, so when the Time Slider is activated,
the layers respond accordingly.
The second widget added to the application was the Layer List widget. The Layer List
widget is used as an organizational tool for users to easily toggle layers’ visibility on the map.
There are two Layer List widgets on the web application: one for the IDP Total category’s

35
layers, and another for the Natural Disaster Total category’s layers. The IDP Total Layer List
widget contains the 11 layers for the IDP Total category. Users can turn them on or off to match
their preference, and can also turn them all on at once, click the Time Slider widget, and view the
data change over time. This functionality is the exact same for the Natural Disaster Total
category as well. Users can toggle the 11 Natural Disaster Total layers and view each year
individually if they choose.
The third widget added to the web application is the Summary widget. There are 22
instances of the Summary widget on the web application, as each instance corresponds with a
specific layer’s year. The Summary widget presents the sum of each attribute of the
corresponding layer depending on the web map’s extent. To ensure this tool’s functionality is
made clear to the user, the Summary tool’s icon is the year and color of the layer it summarizes
(example: the Summary widget for IDP Totals in 2015 is a red box with a “15” inside the box, as
the IDP Totals layer is symbolized using the color red). The Results section provides an in-depth
examination of the web application and delves into the details of each aforementioned widget.

36
Chapter 4 Results
This chapter explores the web application and its functionality in detail, as well as the
development process leading up to the final application. Additionally, Figure 19 displays the
Story Map template, which was added to provide users additional context regarding climate
change, natural disasters, and IDP. The Story Map also details instructions for users on how to
interact with the web application and outlines the use and purpose of the various widgets.
The web application’s foundation is built upon the 22 layers imported from the map built
on ArcGIS Online. One of the challenges facing an application containing 22 layers is ensuring
the layers are organized and displayed in a manner that does not distract the user or become too
complicated for interaction. If a user wishes to see all layers in a single list, they have the option
to do so by utilizing a Layer List widget in the top right tool bar. However, in order to avoid a
cluttered map with overlapping layers and easier navigation between layers, a Layer List widget
was created for both data categories (Figure 12). The Layer List widget separates the categories
into two different widgets, allowing for users to toggle and interact with each group of layers
separately.

37

Figure 12. View of the web application with the IDP Total Layer List widget activated
Separating the two categories allows for users to interact with each category seamlessly, and the
connection between the layers and the Summary widget is made clearer using color coding. As
seen in Figure 12, users have the ability to turn all layers on or off at the same time. If users turn
all layers on, they are then able to utilize the Time Slider widget to the left of the Layer List
widgets.
The Time Slider widget is designed to work when all layers of a single category are
turned on. After a user turns on all the layers of a category, they can click the Time Slider to
view the individual data totals change year over year (Figure 13). Users can also take advantage
of the search bar located above the Time Slider widget to view a specific country’s totals over
time, and pause the Time Slider tool at the bottom of the application in order to focus on a single
year. When a single layer is activated, users can click on the layer’s symbology to view a graph
of the layer’s attributes that are otherwise not shown on the web map. This functionality was

38
outlined in Chapter 3.3 – Figures 8 and 10 demonstrate the pop-ups that display whenever a
graduated symbol is selected by a user.

Figure 13. The web application’s time slider activated for Total Natural Disasters layer

The Summary widget provides a deeper look into the attribute data for both IDP and
Natural Disaster categories. When a data layer’s year is turned on in the Layer List, users can
select the corresponding Summary widget in the widget controller in the top right portion of the
application. The Summary widgets are color-coded and labeled with a year to correspond with
their respective category’s layer. For the IDP Total category, users can zoom to an extent of the
web map to view the total number of displacements per natural disaster type in that specific
extent of the map (Figure 14).

39

Figure 14. View of the Caribbean islands’ total IDP per natural disaster type in 2018

Figure 14 displays the web map zoomed to the extent of the Caribbean islands with the
2018 Summary widget activated, as well as the Total IDP in 2018 layer turned on. The result is a
sum of the IDP produced by each natural disaster type in the region. The three dots at the bottom
of the tool allow for users to toggle through each page of the Summary widget to show all of the
natural disaster types. Note that the country’s symbology must be displayed on the web map’s
extent in order for the country to be included in the Summary widget’s count. Furthermore, a
filter feature is built-in to the Summary widget. Figure 15 is an example of the filter feature’s
utilization. Users can click the “All” drop-down arrow to select a specific country. In Figure 15,
Sri Lanka was selected. The web map responds by zooming to the extent of Sri Lanka and
displays the IDP per Natural Disaster totals. However, this information can also be found by
clicking the symbology of a country, so the Summary tool is best utilized for multiple countries.
Nonetheless, the filter feature is an easy shortcut to move from country to country without
having to scroll and zoom throughout the map.

40

Figure 15. View of the Summary widget while utilizing the filter feature
The Summary widget works similarly when activating the Natural Disaster Total layers.
Only 13 widgets can be displayed in the top right widget controller, however, so the rest of the
green-labeled Natural Disaster Total Summary widgets are housed in the “more” tab at the end
of the widget controller (Figure 16).

41

Figure 16. The Natural Disaster Total Summary widgets found in the “more” tab

When the Summary widget is activated for a Natural Disaster Total layer, the widget
displays the sum of all the types of natural disaster occurrences within the map’s extent. Figure
17 is an example of the Summary widget focused on a group of South East Asian islands.
According to the Summary widget, 159 floods occurred in that area in 2018. Note that the year
of the Summary widget will always be highlighted in the top right widget controller, as seen in
Figure 17.

42

Figure 17. Summary widget used in Pacific Island countries for the 2018 dataset

Finally, if a user has been zooming and scrolling throughout the map and wants to return
to the original extent of the web map, a Bookmark widget was added next to the Time Slider
widget. This simple tool allows for users to quickly zoom back to the original extent of the map
with just two clicks (Figure 18). The widget also allows for users to create their own bookmark’s
if they are focusing on a specific area of the map.

43

Figure 18. View of the Bookmark widget and the original extent shortcut
The web application’s purpose was to provide users an intuitive tool to view IDP and
natural disaster data over a ten-year time span, and on a global scale. The widgets and their
functionalities outlined in this chapter helped achieve this purpose, while still providing a clean
and practical interface. While 22 layers may initially be difficult to organize, the widgets offered
in ESRI’s Web AppBuilder were able to manage them as seamlessly as possible.

44

Figure 19. The landing page for the web application on ESRI’s Story Map
Finally, ESRI’s Story Map tool was utilized as a landing page for the web application
(Figure 19). In embedding the web application into the Story Map website, the site also provides
users a summary of the web application’s theme and an explanation of IDP in relation to natural
disasters. This added context will allow for users to better understand the data with which they
are interacting. Instructions on how to navigate and interact with the web application, as well as a
description of the Summary widgets, are also included above the web application.

45
Chapter 5 Conclusion
This chapter discusses the web application as a whole, including a summary of the web map
design and application creation process, difficulties that arose throughout the data parsing and
web map development process, application limitations, and finally, the opportunity for future
development, improvement, user testing, and expansions.
5.1 Summary of Web Application Development
The web application’s foundation is built entirely on the single master dataset compiled
by the IDMC. The various widgets, layers, and symbolization would not have been possible
without the R code created to count various attributes found throughout the master dataset, such
as natural disaster frequency per country and IDP produced per natural disaster. The R code
successfully created four separate categories, which were then combined into the two main
categories displayed on the final web map: Total IDP and Total Natural Disasters.
ArcGIS Pro served as the foundational application used to visualize the sub-datasets.
Once the sub-datasets were accurately geocoded and visualized on a static map, they were given
a time property in each layer’s properties. After each sub-dataset’s attributes contained an
accurate location and time property, they were shared to ArcGIS Online. The bulk of the
symbolization, including the creation of the graduated symbols for both categories and the
configuration of their respective pop-ups, took place in ArcGIS Online. After the web map was
finalized, it was imported into ESRI’s Web AppBuilder. Web AppBuilder bolstered the
functionality of the web map with the addition of its built-in widgets, including the Layer List,
Time Slider, Bookmark, and Summary widgets. Lastly, the finalized web application was
imported into ESRI’s Story Map application. The Story Map provided context for the web
application’s theme by including information about climate change, natural disasters, and an

46
overview of the definition of IDP. Additionally, the Story Map contained a set of instructions for
users regarding widget functionality, and information about the layer categories. Overall, the
utilization of the Story Map helped prepare users to better comprehend the data with which they
are interacting, and provided additional information and instruction regarding the functionality
and tools built into the application.
5.2 Difficulties Encountered During Development
Once the web map was imported into Web AppBuilder, there were few difficulties
encountered aside from briefly troubleshooting the sums found in the Summary widget. This was
quickly resolved by deleting and re-adding the widget, then refreshing the application, which
likely means the issue was a bug in the widget.
Most of the issues encountered throughout the project occurred in the early stages of
development. The R code needed to be re-ran multiple times on the master dataset to eliminate
random errors or unnecessary information that existed in the IDMC’s original data. For instance,
when the R code was running a count on each natural disaster type, the IDMC data contained an
“oil spill” midway through their spreadsheet. This caused the R script to stop counting all natural
disaster types after the oil spill record, resulting in the latter half of the Natural Disaster Event
Count sub-dataset for that specific year to have 0’s in every single row. Finding the error meant
manually checking each record in the IDMC dataset until the anomaly record was found. Once
the record was deleted from the IDMC dataset, the R script ran as initially planned.
There were also numerous errors when attempting to share the sub-datasets’ layers from
ArcGIS Pro to ArcGIS Online. If any chart symbology was applied to any of the layers in
ArcGIS Pro, they could not be shared to ArcGIS Online – ultimately, layers with certain types of
symbology cannot be shared. As a result, the layers shared from ArcGIS Pro to ArcGIS Online

47
were extremely rudimentary – each layer contained simple location point data, which is why the
majority of symbolization and pop-up configuration was completed in ArcGIS Online.
Furthermore, when attempting to merge the 4 categories’ sub-datasets into 2, the resulting layers
also could not be shared. This is because layers that underwent a join are unable to be shared to
ArcGIS Online. Therefore, the merging of the Natural Disaster Event Count category with the
Natural Disaster Total category, as well as the merging of the Total IDP category with the IDP
per Natural Disaster category, was done manually in Excel. After these steps were completed,
each layer was successfully shared to ArcGIS Online.
Lastly, layers cannot be grouped in ArcGIS Online, which is why each layer needs to be
turned on and off individually in the Web AppBuilder. The layers could have been grouped in
ArcGIS Pro – however, once layers are grouped in ArcGIS Pro and published to ArcGIS Online,
the editing and symbolization is far more limited and would have inhibited the functionality of
the web application.
5.3 Future Development
Currently as it stands, the web application is hosted on USC SSCI servers. In order for
the web application to continue existing long-term, either the Dornsife Spatial Sciences Institute
would agree to continue hosting the application, or the application would need to be recreated on
a personal or different organization’s server that has access to each of the ESRI software
programs used to create the application.
One long-term goal of the application is to be able to implement future IDMC data. Just
recently, the IDMC published the data for the year 2019. A workflow needs to be created so that
new data for future years can be seamlessly added to the web application. Once the workflow is
established, the goal would be for future users to be able to quickly add new IDMC data to the

48
map. For example, the IDMC compiles displacement data for conflict and violence formatted
similarly to their natural disaster dataset. It is entirely possible to add this non-weather related
data to the web application, but would require various changes to the web application’s widgets
depending on the attribute data found in the conflict and violence dataset. However, the scope of
this web application focused solely on weather-related displacements.
Furthermore, the web application’s theme can be enriched with the addition of
supplemental data for each country. For instance, each country’s total population can be added to
the web map to provide context for the country’s IDP total. The visualization of countries with
smaller populations that contain a large amount of IDP in a given year can demonstrate the scale
of natural disaster displacement. Population data would be able to determine the ratio of IDP per
population in each country. Other supplemental data types that would further enrich the
application include a country’s wealth and poverty statistics, urbanization, cultural diversity, or
land use. Such information would help determine whether poorer countries are
disproportionately affected by natural disasters compared to wealthier countries with stronger
emergency infrastructure in place, for example. Demographic data would provide a glimpse into
which groups are affected as well – however, the IDMC does not include the demographics of
each IDP. Rather, demographic data would instead provide more detailed context for each
country affected by natural disasters.
The web application could also benefit from user testing. User testing would allow for
multiple groups of individuals from various different backgrounds and GIS experience to interact
with the application, and ultimately provide feedback regarding the user interface and
functionality. While the user interface of the web application was designed for both technical and
non-technical users, user feedback would help determine whether the application’s interface

49
meets its design goal. Feedback from technical users with GIS experience would also be
beneficial, as these types of users could provide recommendations regarding widget
functionality, symbolization, and the potential addition of new widgets, features and
supplemental data. The feedback garnered from user testing would ideally be compiled and
evaluated, with the end goal of implementing relevant suggestions to the web application.
Overall, the primary goal of the project was met, which was to provide users with a
dynamic user interface that is simple enough to use with no prior GIS experience, yet powerful
enough to transform a large spreadsheet into an interactive, visual tool that can provide new
trends and information to the user. The web application serves as a data visualization tool that
can be utilized by a wide variety of users. It can be applied as an educational tool in a classroom,
journalists can use it to present large amounts of data to their audience, and government
organizations and disaster relief agencies can discover trends in the data to make more informed
decisions to protect IDP. Rather than depending on headlines presented by the news, attempting
to decipher information on a spreadsheet with over 6,000 rows, users instead have the ability to
directly access, customize, and filter natural disaster and IDP data to match their preference in a
dynamic, visual web application.

50
References
Acemonglu, D., Fergusson, L., and Johnson, S. “Does population growth cause conflict?”
VoxDev. November 22, 2019. https://voxdev.org/topic/health-education/does-population-
growth-cause-conflict. (accessed August 25, 2020).
Anzellini, V., Desai, B., Fung, V., Ginnetti, J., Milano, L., Montandon, R. and Ponserre, S.,
2017. “Global Disaster Displacement Risk.” Internal Displacement Monitoring Centre.

Byravan, Sujatha, and Sudhir Chella Rajan. “Taking Lessons from Refugees in Europe to
Prepare for Climate Migrants and Exiles.” Environmental Justice 10, no. 4 (August 1,
2017): 108–11. https://doi.org/10.1089/env.2016.0026.

“Climate Migrants.” storymaps.esri.com. ESRI. http://storymaps.esri.com/stories/2017/climate-
migrants/index.html. (accessed February 23, 2020.)

Coronese, Matteo, Francesco Lamperti, Klaus Keller, Francesca Chiaromonte, and Andrea
Roventini. "Evidence for sharp increase in the economic damages of extreme natural
disasters." Proceedings of the National Academy of Sciences 116, no. 43 (2019): 21450-
21455.

Cybulski, Paweł, and Tymoteusz Horbiński. "User experience in using graphical user interfaces
of web maps." ISPRS International Journal of Geo-Information 9, no. 7 (2020): 412.

“DAT: The International Disasters Database.” The International Disaster Database. Centre for
Research on the Epidemiology of Disasters. https://www.emdat.be/. (accessed April 13,
2020.)

“DATA TRANSFORMATION STRATEGY 2020-2025 Supporting Protection and Solutions.”
UNHCR. The UN Refugee Agency, September 2019. https://www.unhcr.org/en-us/data.html.

“Disasters and Climate Change.” IDMC. Accessed April 9, 2020. https://www.internal-
displacement.org/disasters-and-climate-change.

“Environmental Migration.” Migration data portal. https://migrationdataportal.org/themes/
environmental_migration. (accessed February 5, 2020.)

“GLOBAL COMPACT FOR SAFE, ORDERLY AND REGULAR MIGRATION”. Global
Compact for Migration. July 13, 2018.

“Global Disaster Alerting Coordination System.” GDACS. https://www.gdacs.org/.
(accessed April 15, 2020.)

“Global Internal Displacement Database.” Internal Displacement Monitoring Centre,
2019. https://www.internal-displacement.org/database/displacement-data-exploration-
tool.

51

“Global Migration Indicators 2018.” Global Migration Indicators 2018 | IOM's Global Migration
Data Analysis Centre. https://gmdac.iom.int/global-migration-indicators-2018-report. (accessed
February 21, 2020.

“Global Report on Internal Displacement.” Internal Displacement Monitoring Centre, 2019.
https://www.internal-displacement.org/global-report/grid2019/. (accessed February 13,
2020.)

Groen, Derek, and Diana Suleimenova. “Climate Refugees: Why We Can't Yet Predict Where
Millions of Displaced People Will Go.” The Conversation, November 28, 2019.
http://theconversation.com/climate-refugees-why-we-cant-yet-predict-where-millions-of-
displaced-people-will-go-119414.

Ionesco, Dina. “Let’s Talk About Climate Migrants, Not Climate Refugees,” United Nations
Sustainable Development, June 6, 2019.
https://www.un.org/sustainabledevelopment/blog/2019/06/lets-talk-about-climate-
migrants-not-climate-refugees/.

Kälin, Walter. 2015. “The Nansen Initiative: Building Consensus on Displacement in Disaster
Contexts.” Forced Migration Review, no. 49: 5–7.

Mateo, Brian. “A Case For Giving Climate Migrants Protected Legal Status”. Carnegie Council
For Ethics in International Affairs, September 20,

2019.
https://www.carnegiecouncil.org/publications/articles_papers_reports/case-giving-
climate-migrants-protected-legal-status.

“Methodology Annex.” Internal Displacement Monitoring Centre, 2019. https://www.internal-
displacement.org/database/methodology.

“Safest Places from Natural Disasters.” Safest Places from Natural Disasters. BestPlaces.
https://www.bestplaces.net/docs/studies/safest_places_from_natural_disasters.aspx.
(accessed April 18, 2020.)

Smith, Adam B. The National Oceanic and Atmospheric Administration, January 8, 2020.
https://www.climate.gov/news-features/blogs/beyond-data/2010-2019-landmark-decade-
us-billion-dollar-weather-and-climate.

Smith, Adam B., and Richard W. Katz. “US Billion-Dollar Weather and Climate Disasters: Data
Sources, Trends, Accuracy and Biases.” Natural Hazards 67, no. 2 (March 2013): 387–
410. https://doi.org/10.1007/s11069-013-0566-5

The Nansen Initiative, April 8, 2015. https://www.nanseninitiative.org/wp-
content/uploads/2015/02/DRAFT-Nansen-Initiative-Protection-Agenda-for-Consultation-
08042015.pdf.

52
Turrentine, Jeff. “Climate Change Is Already Driving Mass Migration Around the Globe.”
NRDC. Natural Resources Defense Council, October 4, 2019.
https://www.nrdc.org/onearth/climate-change-already-driving-mass-migration-around-
globe.

United Nations, Department of Economic and Social Affairs, Population Division, December
2015. Trends in International Migrant Stock: The 2015 Revision. (United Nations
database, POP/DB/MIG/Stock/Rev.2015).

United Nations. “Figures at a Glance.” UNHCR. https://www.unhcr.org/en-us/figures-at-a-
glance.html. (accessed February 2, 2020.)

“US Billion-Dollar Weather and Climate Disasters (2020)”. NOAA National Centers for
Environmental Information (NCEI). https://www.ncdc.noaa.gov/billions/.

You, Manlai, Chun-wen Chen, Hantsai Liu, and Hsuan Lin. "A usability evaluation of web map
zoom and pan functions." International Journal of Design 1, no. 1 (2007).

Zagorsky, J. L. "Are catastrophic disasters striking more often?" The Conversation. 2017.

Abstract (if available)

Abstract Natural disasters have always influenced migration, whether international or within one country’s borders. However, as the effects of climate change continue to cause irregular weather patterns and stronger, more frequent natural disasters, the number of individuals at risk of being displaced from their home due to natural disasters is poised to substantially increase. Given the millions of people on the brink of needing to relocate due to natural disasters, as well as the potential billions of dollars needed to repair the resulting damages, there exists a need to better understand trends in weather and migration patterns. Such an understanding would allow for governments and emergency response teams to be more prepared to face sudden onset disasters. The Internal Displacement Monitoring Centre (IDMC) published a dataset detailing the number of internally displaced people (IDP) per country per year between 2008-2018, and the specific natural disaster event associated with each IDP. This project utilized the IDMC dataset to create a web map application using ArcGIS Online that will organizes and visualizes the data in a spatial context. The original IDMC dataset was broken down into smaller thematic datasets using the R programming language, which were subsequently visualized using the ESRI products ArcGIS Pro and ArcGIS Online. The application was designed for ease of use, thus allowing for new trends and potential patterns to be discovered far more easily. The resulting web application includes widgets and tools that allow users to manipulate the dataset in meaningful ways unique to their needs.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Creating a Web GIS to support field operations and enhance data collection for the Animal and Plant Health Inspection Service (APHIS)

PDF

The movement of Mexican migration and its impact based on a GIS geospatial database

PDF

A Web GIS application for airport pavement management

PDF

Building a spatial database of biochar research and practice with Web-GIS

PDF

Tracking trends in earthquakes and tropical storms: a web GIS application

PDF

Comparative 3D geographic web server development: visualizing point clouds in the web

PDF

Geospatial web application development to access irrigation asset data: Veterans Affairs Palo Alto Health Care System

PDF

Building a geodatabase design for American Pika presence and absence data

PDF

Cal ToxTrack: a full stack Web GIS for mapping pollution in California

PDF

Implementing spatial thinking with Web GIS in the non-profit sector: a case study of ArcGIS Online in the Pacific Symphony

PDF

Cartographic design and interaction: An integrated user-centered agile software development framework for Web GIS applications

PDF

Assessing the impact of a web-based GIS application to promote earthquake preparation on the University of Southern California University Park Campus

PDF

Designing an earthquake preparedness web mapping application for the older adult population of Los Angeles, California

PDF

Analysis of park accessibility in Redan, Georgia Web GIS application

PDF

Student engagement capabilities in mobile GIS: A framework for mobile GIS education

PDF

California ballot results viewer, 2008-2018: a Web GIS application for viewing ballot proposition results in California

PDF

Using GIS to predict human movement patterns in complex humanitarian emergencies: a test case of the Syrian Conflict

PDF

Using GIS and asset management to understand hydrant damages and required maintenance

PDF

Recreational Off-road Adventure Motorcycle mapping System (ROAMS): a web application facilitating adventure motorcycling in Idaho public lands

PDF

Designing an early warning system web mapping application for the Atlanta Metropolitan Area before a flooding event

Asset Metadata

Creator Forberg, Declan (author)

Core Title Exploring global natural disaster and climate migration data: a Web GIS application

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 10/26/2020

Defense Date 08/25/2020

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

Tag climate change,climate migration,GIS,migration,natural disasters,OAI-PMH Harvest,web application,Web GIS

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bernstein, Jennifer (committee chair), Ruddell, Darren (committee member), Vos, Robert (committee member)

Creator Email declan.forberg@gmail.com,forberg@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-384614

Unique identifier UC11666366

Identifier etd-ForbergDec-9068.pdf (filename),usctheses-c89-384614 (legacy record id)

Legacy Identifier etd-ForbergDec-9068.pdf

Dmrecord 384614

Document Type Thesis

Rights Forberg, Declan

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