University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Using GIS to Predict Human Movement Patterns in Complex Humanitarian Emergencies: A Test Case of
the Syrian Conflict
by

Colleen Campbell

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

May 2019

Copyright © 2018 by Colleen Campbell

To Michael Campbell and Topher Campbell,
thanks for sticking with me through this process.

iii
Table of Contents
Acknowledgments........................................................................................................................... v
List of Figures ................................................................................................................................ vi
List of Abbreviations .................................................................................................................... vii
Abstract ........................................................................................................................................ viii
Chapter 1 Introduction .................................................................................................................... 1
1.1. The Syrian War as a Test Case ...........................................................................................2
1.2. Motivation ...........................................................................................................................5
Chapter 2 Literature Review ........................................................................................................... 7
2.1. CHE Planning and Response ..............................................................................................7
2.2. Collaboration.....................................................................................................................10
2.2.1. History of Collaborative Software .............................................................................. 10
2.2.2. Collaboration within GIS ............................................................................................ 12
2.2.3. Challenges in Collaboration ........................................................................................ 13
2.3. Networks and Agent-Based Modelling .............................................................................14
2.4. Current Response Methods and Collaboration .................................................................15
Chapter 3 Methodology ................................................................................................................ 17
3.1. Research Design................................................................................................................17
3.2. Platform Selection .............................................................................................................17
3.3. Data Sources .....................................................................................................................20
3.4. Methods.............................................................................................................................20
3.4.1. Travel Method ............................................................................................................. 20
3.4.2. Justification for ArcPy scripting ................................................................................. 21
3.4.3. Preprocessing .............................................................................................................. 21
3.4.4. Python Script Process ................................................................................................. 25

iv

Chapter 4 Results .......................................................................................................................... 28
Chapter 5 Conclusions and Further Research ............................................................................... 35
References ..................................................................................................................................... 38
Code ...................................................................................................................................... 42

v
Acknowledgments
I want to thank the USC Spatial Sciences Department for all the help and guidance given
throughout this program. In particular, I would like to thank Dr. Steven Fleming, Dr. Jennifer
Bernstein, and Dr. Andrew Marx for their guidance in this thesis.

vi
List of Figures
Figure 1 Refugees on the streets of Paris. ....................................................................................... 3
Figure 2 Refugees walking from France to Croatia ........................................................................ 4
Figure 3 Time and location classifications. .................................................................................. 11
Figure 4 Command Flow Chart .................................................................................................... 16
Figure 5 Location of refugee camps and IDP camps .................................................................... 23
Figure 6 Distance to the camps, weighted by linear distance.. ..................................................... 24
Figure 7 Irmischer travel mode applied to the Syria and surrounding areas ................................ 25
Figure 8 Predictive models for days 1-3 and a locator map of Aleppo. ....................................... 28
Figure 9 Predictive models for Aleppo days 4-7 .......................................................................... 29
Figure 10 Predictive models for days 1-3 and locator map of Harasta. ........................................ 30
Figure 11 Predictive models of Harasta days 4-7 ......................................................................... 31
Figure 12 Predictive models for days 1-3 and locator map of Flaita. ........................................... 32
Figure 13 Predictive models of Flaita days 4-7 ............................................................................ 33
Figure 14 Sample StoryMap of Flaita. .......................................................................................... 34

vii
List of Abbreviations
GIS Geographic information system
GISci Geographic information science
SSI Spatial Sciences Institute
USC University of Southern California
CHE Complex Humanitarian Emergency
NGZ No Go Zone
HRC Human Relief Chains
WAB Web App Builder
PLO Palestinian Liberation Organization
IDP Internally Displaced Persons
viii

Abstract
There are Complex Humanitarian Emergencies (CHE) happening worldwide. Currently, 68.5
million displaced persons exist around the world resulting from these CHEs. This project seeks
to develop an Arcpy script for predicting movements within these crises to help government and
non-government agencies manage resources. This project is intended to predict large-scale flows
of people to allow agencies to prepare for an influx of displaced persons and organize shelters,
food, clothing, medical personnel, and other necessities. ArcPy within ArcGIS Pro will give non-
trained users an easy format to view spatially relevant processes and information about the CHE
that they may be unable to share with the public. Additionally, this project looks at theoretical
models of human movement that are most relevant to the circumstances of the CHE and should
help speed communication between GIS and non-GIS professionals. This project uses the Syrian
refugee crisis as a test case because of the large-scale movement and long duration involved.

1

Chapter 1 Introduction
A Complex Humanitarian Event (CHE) is an event, such as a natural disaster, war, or man-made
disaster, that leaves people in crisis and in need of humanitarian efforts and relief (Pakes, 2008).
Disasters are becoming more frequent and stronger, wars are global and are lasting longer, and
droughts and economic instability are causing harsh conditions (Coleman, 2006). These
increasingly severe events are extending the impacts of CHEs to an increasing number of people
as the world’s population continues to grow (Hall, 2017). Many of these events require that
impacted individual’s basic needs are met. A few examples of such needs include safety, medical
services, food, shelters, clothing, and clean water. As the spatial and temporal scale of a CHE
changes, so too do the needs of those impacted. Someone who is affected by a hurricane may
need a week’s worth of supplies, while those leaving during a regime change may need supplies
and support for several years.
There are 68.5 million forcibly displaced persons (FDP) worldwide; this includes 40
million internally displaced persons (IDP), 25.4 million refugees, and 3.1 million asylum seekers
(UNHCR, 2017). These numbers do not include the approximately 10 million stateless people
who have been denied any nationality with associated rights and protections. Over half of FDP
are children. A recent report stated that one person is displaced every two seconds (UNHCR,
2017). The FDP crisis is not a small problem. Often it is assumed that refugees leave their homes
for a short time and then return to them. That is not always the case, as the average time a
refugee is displaced is 19 years (Xu, 2015). Many refugees are forced to leave their home
country and live permanently in other countries, while others return to their country of origin but
are not able to return their city of origin. Some organizations feel that if a refugee does not return
to their original home, that it is a failure (UNHCR, 2008).
2

It is essential, however, to consider that many people do not live in their original homes
their entire lives. Therefore, it would be more appropriate to determine a situation a success if the
life of refugees can be or has been improved. This project serves as the first step in developing a
set of tools to aid in the return or resettlement of FDP and IDP citizens by providing support
information for decision-makers. This project does not intend to solve the entirety of the
problem, but to provide aid to those who are working on the problem.
1.1. The Syrian War as a Test Case
In 2011 violence in Syria began; it is a continuation of the protests associated with the Arab
Spring and the start of a civil war within the country (Britannica, 2017 ). The longevity,
visibility, and the number of people affected are the reasons this CHE was chosen as a case study
for the project. This CHE also involves a large number of refugees who are both internally and
externally displaced. When this conflict began in 2011, people moved away from their homes in
response. At first, they traveled short distances to neighboring towns or to stay with family, but
soon they were traveling further and leaving the country. There was and continues to be bombs
and guns; even hospitals are targets (Shaheen, 2017). Situations like the Syrian CHE leave
decision-makers and refugees with many questions that need to be addressed. Where will these
people go? What will happen to them once they arrive? Will there be food, clothing, toiletries, a
place to sleep? Will they find work as doctors or engineers? Can their children finish their
university studies? Figures 1 and 2 show photos taken four years after the violence started in
Syria. These people have left with what they can carry and made their way to France. However,
France was unprepared for the influx of refugees and asylum seekers (Holmes, 2016), despite
being a first world country with many resources. Of course, there are other things at play such as
3

politics, but if France had been prepared, it is possible that the policies would have been very
different.

Figure 1 Refugees on the streets of Paris. (AFP Photo/Nimani, 2015)
The recent response to refugees by many countries legitimizes these questions. Many
times, refugees are not welcome in both their home countries or new countries where they seek
refuge. Why? Simply put, refugees are challenging, because they have needs like food, shelter,
sanitation, and medical attention. In addition, they are perceived to have no positive impact on
the host countries.
4

Figure 2 Refugees walking from France to Croatia (AFP Photo/Francois Guillot, 2015 )
Preparedness could have been the difference between French citizens gaining a few
neighbors and tent cities developing along major streets. Figure 2 is a photo of people leaving
France attempting to find another country to provide assistance. This departure from France is a
further loss for them after losing everything to the ravages of war. Violence and natural disasters
continue to increase in severity and frequency, resulting in a world now grappling with a crisis of
how to better respond to the vast humanitarian need.
5

1.2. Motivation
This problem affects people. It is fundamental to recognize the needs of these people and how to
provide for them. Many times, women and children make up the majority of those who are
migrating, and they are the most vulnerable group (Jahngiri, 2012). Unfortunately, some
governments have not prepared for women because they do not want the experience to be “too
comfortable” with the risk of the women choosing to remain in the host country instead of
returning to their home country. These “comforts” include avoiding sex trafficking, rape,
prostitution, successive pregnancies, unwanted pregnancies, STD’s, and many other atrocities
(Gasser 2004). It is not only imperative to have food and water available for evacuees and
refugees but also to provide safe places. Medical teams who can assess the needs of women who
are often pregnant or with small children are needed in these areas, but the medical expertise is
not always accessible. CHEs are traumatizing experiences for everyone involved. It is an
important part of the assistance effort that people trained to respond to traumas are available
(Newnman 2016).
When a CHE happens, whether it is long-term like a war or short-term like an
earthquake, there will be an economic impact. An estimated 2.5 million jobs were lost due to
9/11 around the United States as a whole, not just in New York. (Sullivan, 2003). After a recent
earthquake in China, people in affected impoverished areas took jobs in other parts of the
country unaffected by the earthquake. While some of these people commuted, others moved and
changed to the new locations. This earthquake had an economic impact for at least two years past
the initial CHE (Yuan, 2016).
From all of this, the primary motivation for this project is to assist in reducing the
suffering of these people. Many of them, especially children, are vulnerable. This project, and
6

future ones like it, are intended to aid those on the ground and those in government positions in
making decisions that will be beneficial and compassionate.

7

Chapter 2 Literature Review
2.1. CHE Planning and Response
There are three main phases to a CHE; (1) pre, (2) during, and (3) post. The timing of these
stages can range from predictable to spontaneous and from a few days to years (Williams, 2017;
Hu, 2016). Because of these differences in the timing of stages, responses to CHEs vary as well.
In some areas, like the southeastern United States, hurricane season is an annual event, in which
CHE of this type has a warning period. The consistency of the CHE allows planners to prepare
for these events year-round and focus on how, when, and why people evacuate (Yin, 2014).
Agencies involved in hurricane preparation know that not everyone will evacuate, so they must
prepare to respond during the incident. One method of preparation is to place supply depots
along possible evacuation routes in case people essential supplies while evacuating both during
and after the storm (Lodree, 2009). When planning for these stations, it is crucial to know which
routes for people will remain accessible and what locations along these routes are the best places
to establish supply depots. GIS has been used to create predictive models that include areas
where streets are no longer accessible (Horner 2011). The 2018 hurricane season for North
Carolina was particularly rough. Two hurricanes, Florence and Michael, affected the state in
succession, an example of how subsequent events can have compounding effects on a CHE.
There are CHEs other than natural disasters that can be anticipated, allowing for
necessary planning and preparations to be made. The European Union (EU) foresaw the current
Syrian refugee crisis that started in 2014 and preemptively established a fund for 2014-2020 to
help resettle refugees within Europe. These resources, along with domestic aid, were meant to
offset costs associated with the predicted influx of refugees. The money was partially successful,
but significant border closures limited the ability to apply the funding in a meaningful way.
8

(Newnham 2016). The timeframe for the fund is not complete, so it will be a few years before
analysis will allow for an actual discussion of its effectiveness. Unfortunately, even when there
are new settlements and people can relocate there are many other stressors involved. Many
people in these situations do not feel that their lives have any worth other than contributing to the
success of future generations (Yako, 2014). In Canada, the city of Toronto prepared for refugees
by using GIS-based site suitability to find the best areas to resettle refugees. The criteria used
included access to affordable rent, language classes, employment opportunities, and schools. In
this example, decision-makers chose outcomes that benefited both the refugees and established
citizens of Toronto (Vaz, 2016).
In contrast, Lebanon was opposed to refugee camps from the Syrian crisis within its
borders due to recent historical events. Lebanon has become the home to many Palestinian
refugees who live in camps. These camps are now mostly controlled by the Palestine Liberation
Organization (PLO). This has created an environment where Lebanese police cannot safely enter
(Dunne, 2017). The loss of control is in part due to the continuation of the camps over an
extended period and no attempt by the government to integrate Palestinian refugees into
Lebanese society. As a result of this experience, Lebanon has officially closed and restricted its
border with Syria. However, people are still fleeing into the country, and the lack of preparation
prevents both illegal and legal immigrants from finding work or housing. One result of this lack
of preparation and development is unsanctioned tent cities in fields or along roads. In addition,
this has resulted in a non-working population who are draining the resources of Lebanese
citizens (Sanyal, 2017).
While it is helpful to predict and prepare for a possible event, during a CHE it is essential
to know how to aid people who have immediate needs. In the past, it was almost impossible to
9

have instant information about what was happening and where. Social media has provided
unparalleled access to near real-time information. Both Twitter and Flickr are a great way for
emergency professionals to find out what is happening, where it is occurring, and even how
severe the situation may be. Twitter was especially useful in response and circulation of essential
information during Hurricane Sandy. This hurricane was one of the largest on record when it
struck the US northeast, including areas, like New York City, where there are large numbers of
Twitter users (Yoo, 2016). Semantic searches can insert results into a GIS-linked database
(Zhang, 2017), further connecting the nearly instant information from social media to the
necessary response professionals.
The post-CHE phase is a complicated time. In some cases, such as a storm or earthquake,
the first two stages can last for a short time, but the post period can last much longer. The first 72
hours are critical, and humanitarian relief chains provide needed food, clothing, diapers, formula,
and other supplies (Tofighi, 2016). Medical care, with a particular emphasis on psychological
and trauma care, are among the necessities to assimilate into a new life or return home
(Newnman 2016).
Because these situations are complex; the pre, during, and post needs and methods are
sometimes mixed. The Toronto relocation study happened during the pre-phase, before the
arrival of refugees, and the actual relocation occurs post arrival. In the during and post CHE
phases it is important to respond with sufficient supplies. Adding to the complexity, sometimes
there are waves within a CHE, especially in the case of war. There may be days with no fighting,
in which there will briefly be a post phase as people leave and a pre-phase as organizations plan
for the next influx. Due to the ever-changing conditions of these situations, it is vital for experts
to have information readily available during all phases of the CHE.
10

2.2. Collaboration
Both synchronous and asynchronous communication have become part of everyday interactions.
Blogs, forums, Skype, Twitter, online education, and other media allows communication
regardless of geographic location and time. A person can comment on a blog written five years
ago or talk “face-to-face” with a coworker in a different time zone. Many types of collaboration
are essential to this project because it is rare that all stakeholders and parties can be in the same
location. There are generally several stakeholders addressing a CHE, and in order for them to
collaborate and efficiently respond, they need to be able to communicate quickly and precisely
from various locations and settings.
2.2.1. History of Collaborative Software
Research and companies have made a variety of shareware and groupware also called Computer
Supported Cooperative Work (CSCW) systems products to facilitate collaboration, for both
synchronous and asynchronous communication. One of the goals of CSCW is to allow the
computer not only to solve problems, but aid in human interaction to solve those problems (Ellis,
1991). Types of collaboration can be sorted into four main categories as shown in Figure 3.
11

Figure 3 Time and location classifications (Armstrong, 1993; Sun, 2016)

One of the problems with early collaborative software was turn-taking. Turn-taking is
when the file is accessible to multiple users, but only one active user is permitted to edit the file
at a time (Sarin,1988; Greenberg, 1990). In an attempt to solve that problem, “locking” was used
(Stefik, 1987). “Locking” allows users to “lock” only the sections they were actively working on
while letting other users edit within the same file. The “lock” allowed multiple users to work on
the same file, but not in the same sections of that file. Operational Transformation (OT) allowed
users to each have a copy of the file and then send updates to other users’ files. The
establishment OT allowed users to work on the “same” file simultaneously and give updates
periodically (Ellis, 1989). In the mid-2000s Google introduced Google Docs which allows users
to edit and comment on the same document simultaneously. Since then, there has been an
expansion in CSCW programs that allow for simultaneous, document sharing, and manipulation
as well as video conferencing.
12

2.2.2. Collaboration within GIS
At the start of GIS collaboration attempts, many GIS were added into current CSCW software or
used in conjunction with that software. An example is GroupArc (Churcher, 1999) which used
GroupKit, an existing CSCW and Arc/Info to allow collaborators to visualize spatial topics while
discussing and collaborating in real time. A problem with this approach was that every
collaborator needed to have a copy of the software downloaded to their computers. Having
multiple copies is problematic because it can be cost prohibitive as well as challenging to
maintain updates. Another project in this era was Spatial Group Choice (Jankowski, 1997) which
used the CSCW version of Microsoft Windows for Groups and Group Choice with ArcView 2.
The limitations here are similar in that each collaborator needed to have access to all of the
software and operating systems in order to share information.
As the internet advanced, so did GIS collaborative platforms. Researchers not only
wanted to have collaborations on the same documents, but they also wanted it to be synchronous.
A primary goal of future programs was to facilitate a way for all collaborators to see what the
other collaborators were seeing. One of these platforms, called GCC_CONNECT, was explicitly
created to aid in crisis management or CHEs (Cai, 2005). This platform had a similar goal to this
thesis, allowing collaborators in different locations and capacities to be able to make decisions
using available the data quickly. However, the majority of the work presented for this software
was focused on developing a multi-user, shareable architecture with the ability to share
geographic analysis. There are also other challenges such as bandwidth availability and
connections which may limit the number of people available to participate in the collaboration.
GIS is currently addressing the lack of a collaborative medium with the use of changing
technologies. Desktop GIS applications are still needed and very useful, but GIS has branched
out into both internet and mobile apps (Maguire, 2006). However, many of the desktop GIS
13

functions allow for rapid prototyping of new applications, which makes developing new
collaboration methods a multi-stage process involving desktop, online, and mobile GIS.
Currently, interactive GIS-based products are frequently used, finding location-based weather,
directions, traffic information, travel plans, dinner options.
2.2.3. Challenges in Collaboration
There are currently two main collaborative challenges in GIS as a field: (1) expertise exchange
between fields of study and (2) the lack of collaborative mediums (Schafer, 2005; MacEachren,
2005). GIS exists as a field in its own right; however, many projects within the field pair with
other areas of research. Often within collaborations, participants share some degree of
knowledge between the involved fields without being an expert (McGee, 2007).
Other inter-field collaboration issues include differences in semantics, metadata, and
methodologies. GIS experts and hospital workers sometimes have problems relaying information
due to a lack of uniformity. Granell discusses that during a health outbreak, even within the same
type of organization, such as hospital to hospital or county to county, there are often different
methodological approaches (2014). There are different organization levels, EMT, hospitals,
health agencies, and citizens working to create the best outcome. Minimizing confusion is an
essential aspect of collaboration (Granell, 2014). As large amounts of accessible data held in
geoportals and geodatabases data become available, the metadata and the keywords (semantic
search) should be standardized to avoid difficulties when searching for data. Agencies should
create a standard for metadata that is required. Enforcing data consistency reduces data
redundancies that occur when users duplicate datasets that already exist, but cannot be easily
found (Hu 2015).
14

One of the primary purposes of this project is to give non-experts within the GIS field a
tool that they can efficiently use. Although Google Maps uses GIS elements at its core, like
digitization, photogrammetry, and satellite data, few untrained people recognize this component.
Instead, they merely type in their destination and wait for directions, utilizing a GIS without
noticing. Ideally, this system will allow for seamless collaborations in which the user does not
need to understand the background GIS processes in order to use the system effectively.
2.3. Networks and Agent-Based Modelling
Where, why, and how people will move is a question that geographers have tried to answer for
years. Tobler’s First Law of Geography states that things that are closer are more related than
those that are farther apart (Tobler, 1970). The friction of distance states that the further
something is, the less likely someone is going to go there (Mayhew, 2009). For instance, a
person is more likely to shop at their neighborhood grocery store than at a grocery store hours
away.
Within the CHE, users who help people, whether in the pre, -during, or post time periods,
want to make sure during all stages of the event that the most amount of people are safe and that
supplies are available quickly. One of the ways to ensure this is modeling how, why, and when
people will either evacuate, migrate, or shelter in place during a situation. Agent-based modeling
is one way to study active systems and the individuals within the system (Kelly, 2003). Some of
the factors that determine behavior are family size, economics, transportation availability, and
education levels (Yin 2014). For instance, a wealthier family is more likely to evacuate early and
stay at a hotel outside of the range of the hurricane, whereas poor minorities are likely to go to a
shelter or stay in their homes. However, knowing where people will go isn’t the only reason for
15

agent-based models. Disease is often found at refugee camps and the ability to predict an influx
of disease can aid in prevention and management (Crooks, 2014).

2.4. Current Response Methods and Collaboration
For CHE response teams, both civilian and military, the term “field” means anywhere the
disaster is happening, this can vary in size depending on the size of the CHE. The field is often
defined by where the CHE is not, or where life starts functioning normally (Arziman, 2015).
Forward teams go ahead of the main group to asses medical and supply needs. They also look for
ways to mitigate further adverse health effects. Usually, these teams are small, the Australian
Disaster Response forward teams (AUSMAT) consist of 1-4 civilians (Robertson, 2011). These
forward teams then report back to their organizations, NGOs or government, to report what they
have found so other teams can be organized and mobilized. Besides forward teams, other
response teams are found in the organizations, whether NGO or government, and these teams are
grouped into clusters. These clusters function to ease communication roles between members of
teams. Inter-organization coordination and communication is essential but can be complicated
when there are many people involved. Lack of communication can result in loss of life and
resources (Kaynak 2014). The US has adopted the Incident Command System in 1970 for all
disasters. This system relies on the deployment of intervention teams, which consist of planning,
logistics, security, and finance persons. Each team uses a single command structure with a
limited, manageable scope of control. For instance, the security person oversees security, and
they are the one making all the security decisions (Arziman, 2015). Figure 4 gives an overview
16

of the command structure.

Figure 4 Command Flow Chart (Arziman 2015).
This project works well for these small groups as well as larger ones that follow. The
small groups can benefit from having a map of predicted areas of interest. They can go to those
locations and then asses the needs of future larger endeavors. They can also relay information
back about more possible locations that can be run through the simulation.

17

Chapter 3 Methodology
3.1. Research Design
This project is intended for use at the organization level, rather than providing a route for
individuals trying to leave a situation. The targeted organizations for this application include
both government agencies, such as FEMA, departments of health and human services, and non-
government /humanitarian organizations like the UN, UNICEF, and Red Cross/Crescent. The
primary purpose for this project is modeling movements of people as groups for agency
response, not for individuals to map a safe route.
3.2. Platform Selection
The selected platform for this project has a list of four criteria that need to be met in order to be
useful to the targeted organizations. The first criterion for a successful implementation of this
project is that the platform needs to be consumed by an audience that has a wide range of GIS
expertise. Any outputs from the project need to be accessible not just to the analyst producing the
results, but to a broader audience involved with planning and preparation of sites for incoming
people. To accomplish this the resulting analysis needs to be available as a web-based display.
The second criterion is that the platform needs to have enough processing power to run multiple
spatial calculations and maintain the integrity of the resulting datasets. The third criterion is that
the cost of modeling movement must be low. Since the events can happen quickly and affect a
large area and number of people, it is important that the modeling and any possible
reassessments can happen at low cost in both time and money. The final criterion is that the
platform should be robust, with current support systems in place and future support expected for
the foreseeable future. Four different platforms, Google Maps/Earth, ArcGIS WebAppBuilder
(WAB), ArcGIS GeoEvent Server, and ArcGIS Pro were examined using these four criteria.
18

Google Maps/Earth were considered together as a possible platform for the project. The
main appeal for using Google Maps/Earth for this project is their common usage among the
general population. The outputs from either part of the Google platform are formats that are
easily shared with many collaborators and allows for an integration of data from multiple users
as either inputs or outputs. However, the Google platforms fail on the second criterion in that
they do not have enough spatial processing power to run the analysis in a way that can
incorporate the individual data sets into a single output. Google does pass on the remaining two
criteria, as many of the functionalities are free and support for both Google Maps and Google
Earth is expected to continue for some time.
The next platform that was considered for this project was WAB. Since this platform is a
web-based, highly flexible and customizable tool; it had much initial appeal as the primary
platform for the project. It passes the first criterion in that it can be made available to a broad
audience and has a customizable interface that allows non-GIS users to interact with the model in
a natural way. WAB almost passes the second criterion. While WAB can easily handle each
individual calculation, at the time of its testing, it had trouble combining multiple tools or
widgets. In order to run the project through WAB both the buffer and overlay widget needed to
be run at the same time, which was not possible during the initial testing. WAB does not pass the
third criterion of cost. Many of the widgets and performing spatial analysis using WAB
consumes credits. While some credits are included as a part of annual licensing of Esri software
some of the processes that were run started to increase as the project expanded in scale greatly.
Since to goal of this project is to provide a tool the users could run over an unknown number of
days/weeks/years any credit cost would limit the usage of the project. This may be resolved with
Esri’s program for non-profits which offers cost-saving measures (Esri, 2018). WAB has a lot of
19

support and development currently occurring and so passes the fourth criterion, and future
versions of this model may be converted to WAB as processing costs decrease and functionality
increases.
The third platform, GeoEvent Server appears to meet all of the criteria listed above,
however, due to personal logistics and local internet firewalls, this platform could not be
operated remotely. I did not have access to a local version of GeoEvent Server, but it is
anticipated that this platform will be a viable option for future projects.
ArcGIS Pro was selected as the platform of choice because it manages to meet all four of
the criteria needed for successful implementation. ArcGIS Pro allows for direct export of maps
to ArcGIS Online, so any results from analysis can be directly shared with collaborators. The
maps from ArcGIS Online can even be embedded into HTML code for access across any type of
webpage. ArcGIS Pro has extremely flexible spatial analysis capabilities through the use of
ArcPy Python scripting. ArcPy allows for almost every functionality within ArcGIS Pro to be
coded into Python scripts, which can be run either within ArcGIS Pro itself or through a separate
Python interpreter. ArcGIS Pro also passes the third criterion of cost as many of the target
organizations for the project either already have ArcGIS Pro available or can get the software for
government or non-profit pricing considerations. ArcGIS Pro is also currently under full support
and development from Esri as the future replacement of the current desktop software and thus
passes the final criterion.

20

3.3. Data Sources
This project is the development of a proof of concept model that in future stages will incorporate
live data streams. In order to show the success of the model, a data set was compiled of recent
real events from Syria from multiple web-based sources. These events range from January 2015
to April 2018 and include over 550 individual events that are potential triggers for forced
migration. Included in this data set is the date and location of the event, if violence or death
occurred during the event, and finally the source of information for the event. Also, weather data
from the National Oceanic and Atmospheric Administration was compiled from the same time
frame for all weather stations in the nearby area. The weather data includes the location of the
station as well as the recorded maximum and minimum temperatures and any precipitation. The
data is available daily for the time frame of the project. The final initial data source was obtained
from ArcGIS online and includes a Digital Elevation Model of the area around Syria and the land
borders of the surrounding countries. This data was gridded to 750-meter resolution and is used
as the primary source of slope and landcover calculations to aid in understanding the overall
topography of the study area.
3.4. Methods
3.4.1. Travel Method
The model examines the likelihood that individuals will walk a set distance from an event
location over the course of a week. Walking was chosen because it is a travel method that is
available to almost everyone and can be used when access to a vehicle is not possible, practical,
or safe. The development of events in Syria has shown that walking is an important travel
method for these types of events, as individuals have walked to refugee camps and cities
throughout Europe. Several drivers to the distance individuals will walk were added in, including
21

the distance to the nearest refugee camp, slope, water, weather, and level or type of violence.
Any of the variables that do not change from a day to day basis were precalculated beforehand as
described in section 3.4.3. Travel likelihoods are calculated on a daily basis for up to seven days
to give planners an idea of how quickly refugees will be arriving in the area.
3.4.2. Justification for ArcPy scripting
One of the greatest strengths of using scripting through ArcPy is the added flexibility to
incorporate the many tools that already exist within ArcGIS toolbox. Initially, the project was
intended as a series of interconnected tools within ArcGIS Model Builder, with a focus on the
Least Cost Path and Weighted Overlay tools. However, these tools can only use raster data, and
each daily iteration must be entered manually. This significantly reduced the efficacy of the tool
for this study. The Raster Calculator only works for iterations within ModelBuilder works if the
file name remains the same and can run with live/ changing data. ArcPy scripting allows for
customization and a combination of tools and outcomes with logical if statements. This was used
to make weights that were then added to the Raster Calculator.
3.4.3. Preprocessing
Using ArcGIS Pro three pre-processing rasters were created. These files are created before
running the Python script because the information from these calculations remains fixed
throughout the study area. The first of these files is a binary file raster file used as a mask to
exclude large bodies of water from the final calculation. The second file is a raster that calculates
the distance to the nearest relocation camp over the entire study area. This raster is then
reclassified to an inverse ordinal ranking system based on days of travel (1-7) to the nearest
relocation camp, to indicate a travel preference toward places of safety. This variable is
considered a “pull,” indicating a preference to travel toward locations of perceived safety. In this
22

case, both internal and external refugee camps were used to create the raster. People are more
likely to go to places they feel are going to be safe and have supplies. Figure 4 shows the
location of the camps, while Figure 5 shows the weighted raster. The final raster is a travel
likelihood raster based on the ease of travel over the terrain based on slope. This calculation uses
Irmischer’s travel model to calculate the travel velocities over the study area. The travel
velocities (given in m/s) are used as a proxy for ease of travel, with the assumption that people
trying to leave in a hurry are more likely to choose terrain that is easier to travel. These values
are then converted to dm/s in order to provide equal mathematical weight with the distance to
camp calculation. Initially, Tobler’s law was used to calculate the terrain’s ease of travel.
However, Irmischer’s travel model (Irmischer 2017) was used instead (Figure 6). Irmischer
compares his results to that of Tobler’s and provides results that are more accurate. This method
also divides the predictions by gender. For this project, the travel model of males over uneven
terrain was used. The project goal is to see when people will start arriving at the relocation
camps. Men are usually faster than women; in addition, women are usually traveling with
children or groups which would be considerably slow their progress. Other groups such as the
elderly and disabled were not accounted for because again, they would be moving much slower.
23

Figure 5 Location of refugee camps and IDP camp
24

Figure 6 Distance to the camps, weighted by linear distance

25

Figure 7 Irmischer travel mode applied to the Syria and surrounding areas

3.4.4. Python Script Process
The Python script has several stages that are linked together to perform the overall model
calculations. The first step is to read in the data sources to define the variables used in the
calculation. The defined variables include two feature classes and three raster datasets. The two
feature classes are point features that define the event and the weather conditions for the day of
the event. In order to simplify the project, these features were separated from their respective
data sets for the full-time period. The three raster datasets include the previously run calculations
26

for land coverage, slope, and distance to camp. In addition, the output data location needs to be
identified before running the script.
The second process defines the travel weights to be used in the model. First, the weather
feature class and event feature class are combined using a Spatial Join of the nearest weather
station to the event. This creates a single file with both the violence level of the event and the
current weather conditions. These inputs are used within a conditional statement to determine
what the daily travel distance will be. The first set of variables considered are violence related. A
binary query is established for violence yes/no and death yes/no. Not all events are violent such
as protests, just as not all violence-related events result in deaths. These queries are ranked to
determine how far people will travel following an incident. Protests, while indicative of social
unrest, don’t warrant mass exodus and thus people who leave are less likely to travel as far.
However, some people may choose to leave at this point. When violence and death occur, more
people are likely to leave and will travel further. This project does not quantify the level of
violence or death only if the event is violent or deadly.
The next variable is the weather. This proved to be the most complex of the variables and
determined considerably the methods used. The parameters for this variable were, precipitation,
heat, and cold. Precipitation included snow, rain, or hail in any recorded amounts. Heat was
determined as anything over 95° F or 35° C, and cold was determined as anything under 32° F or
0° C. These temperature extremes would make travel more difficult. The temperature range
proved to be the most difficult parameter to incorporate into the analysis.

27

The next step is a series of logical queries to determine the distance that will be
traveled each day. The default daily travel distance is set to 16 km based on how far a healthy
person can walk without distance training (Bumgardner, 2018). This default distance is then
modified in the case of violent (18 km) or death (20 km) events, or negatively modified due to a
high maximum temperature (14 km), a low maximum temperature (12 km), or precipitation (10
km). This distance is then set as the buffer distance in the Buffer tool and used in an iterative
loop modeling the next seven days of travel. This buffer is then converted to raster format
(Feature to Raster). Additionally, a fixed value is added to the raster based on if the event is
either non-violent, violent, or if death is involved. This factor will be used as a part of a
predefined colormap to indicate likelihood of people leaving.
The final step of the calculation is to use map algebra to combine the modeled travel
distance with the land classification, travel model, and distance to relocation camp rasters. This
step is contained within the iterative loop and applied individually for each day. This final raster
is then converted to integer values to provide an ordinal classification and to allow for the
addition of a fixed colormap (Add Colormap). This colormap contains a set of pre-coded
classified coloring system to ensure that the data outputs are consistent. The colors chosen are to
represent a scale of threat similar to what is seen at airports and other security areas. The full
code is available in the appendix.
The calculated results are then automatically displayed as a map. All seven days are in
individual rasters. These can then be exported in a variety of ways, however, for this project, a
web map in the form of a StoryMap was chosen. A StoryMap was chosen because it is already in
the format to accept the map and non-expert could easily update it.

28

Chapter 4 Results
Three events were chosen based on location, level of violence, and precipitation levels. Event
one is in the city of Aleppo, Syria and took place on July 28, 2016. This event was non-violent,
and there was no death for this event, but there were tensions and changes in rebel and terrorist
organization structure. There was no maximum temperature recorded that day, and the minimum
temperature was 68 F or 20 C. There was also no precipitation recorded. The green color ramp in
figure 8 reflects the event being of interest, but no violence or death.

Figure 8 Predictive models for days 1-3 and a locator map of Aleppo
29

Figure 9 Predictive models for Aleppo days 4-7

As noted in figures 8 and 9, the darker the color more likely people will move in that
direction. The blue triangles are the camp locations, both internal and external. The darkest areas
are around the camp location bands.
The next event located in Harasta on June 18, 2017. The maximum temperature for that
day was 97 F or 36.1 C, and no minimum temperature or precipitation was recorded. This
situation had violence, rebels fired on aid workers, but no deaths were reported. The orange color
ramp reflects the violence, but lack of deaths reported. The results are displayed in Figures 10
and 11.
30

Figure 10 Predictive models for days 1-3 and locator map of Harasta

In Figures 10 and 11 once again, the darker the color are the areas that people are more
likely to go. The buffer ring is smaller on this set of maps than the other two events. The reason
for this is that the temperature was higher than the determined more manageable temperature.
However, the darker areas indicate that travel is more likely toward the camps, which cannot be
seen until maps from days 4-7.
31

Figure 11 Predictive models of Harasta days 4-7

The third event, pictured in Figures 12 and 13, took place in the city of Flaita on January
2, 2015. This event was both violent and deadly as rebels were engaged with terrorists. The max
temperature was 58 F or 14 C, no minimum temperature or precipitation was reported.

32

Figure 12 Predictive models for days 1-3 and locator map of Flaita.

Again, in Figures 12 and 13 the darker the color, the more likely the people are to go to
that area. The darker color goes again towards the camps which can be first seen on Day 3. This
buffer ring is slightly larger than the other ones. This is because death was involved which
increases the rate at which people would leave.
33

Figure 13 Predictive models of Flaita days 4-7

The maps can then be shared as a webmap and added to any number of platforms.
Chosen for this project was Esri’s StoryMap. This is for ease of adding information by anyone
with editing rights. Also, it already is compatible, and the user can easily select maps to add to
the StoryMap. The following figure 14 is a sample StoryMap.

34

Figure 14 Sample StoryMap of Flaita

35

Chapter 5 Conclusions and Further Research
The primary application of this project is for small forward teams operating at CHE sites and
nearby areas. The output of the project is intended to provide timely information to aid in the
decision making that these teams participate in to assist displaced persons. This project has been
successful because it provided a tool that covers the four criteria mentioned in the methods
section. The first of the four criteria is that it is consumable by a large audience. The maps that
are produced can be displayed on a webpage, pdf, email, or other formats as needed. One of the
aspects that will be addressed in any future expansion of this work would include web-based
interactivity and participatory GIS. The second of the four criteria is Does the method allow for
sufficiently complex calculations to produce a useful output? Arcpy is robust enough to do the
calculations necessary for the complexity of the problem. The third of the four criteria is cost.
The tools used in this script are part of a typical deployment of an ArcGIS account used by large
organizations such as an NGO or government. The cost in time is also low due to the current
version of the script being short. Each iteration took less than 5 minutes to run once all the data
had been entered. The final of the four criteria will support for the platform continue for a
significant amount of time was also met using Arcpy in ArcGIS Pro. ArcGIS Pro has been
proven to be robust in the past, with ESRI support being comprehensive and helpful.
Additionally, there is increasing support for online app development.
Future platform selection could reconsider at Esri’s WAB and Esri’s GeoEvent Server.
Cost for WAB may be prohibitive if used for daily functioning, however, because CHEs are not
daily events it remains a platform that has potential. GeoEvent Sever could be cost prohibitive,
but the fee is annually and not based on usage. Importantly, Esri offers special pricing for NGOs
and other service-based organizations (Esri, 2018). This could lower the cost enough to be a
36

viable option. A future project could compare the platforms for cost, efficacy, and user-
friendliness. Currently, this project could only really be implemented by someone with
knowledge of Modelbuilder and Python, in the future creating something that could be used by
anyone would be the goal. The more people able to use the tool, the more likely it is to be
implemented.
There are new ways of creating models such as looking at cell phone data before, during,
and after a CHE. Participatory data allows planners to see how people respond to an event. A
finding of particular interests is that people tend to leave in waves after the event as jobs and
housing become scarcer (Yuan 2016). This type of collected data falls under the participatory
category, meaning that a user does not only query the computer for calculations, but also adds in
information from cell phones, social media, government records, and other available sources.
When this participatory data is semantically and ontologically consistent, it allows for a better
model response (Goodchild, 2008). Increasingly fine-scaled predictions of human movement are
available with more advanced computational capabilities and the ability to run sophisticated
statistical analysis (Longley & Batty, 2007). However, many of the outputs are not easily
understood or used by non-experts, including those who have little computer knowledge outside
of using a web browser. Unfortunately, the vast majority of first responders and decisions makers
involved in CHEs lack this expertise. This is due to having a focused their expertise in other
fields, further exemplifying the need for collaborative efforts between fields.
Another future consideration is the cell size of the raster. The amount of detail for the
elevation data for Syria is not as fine-scaled as other places in the world. Scale for the rasters is
something that will be a consideration for other projects.
37

Adding other parameters such as violence happening in other locations, not just in the
city itself, would be useful as people travel away from violence. Changing the travel parameters
for women, children, large groups, elderly and disabled would be beneficial as well. Then a
prediction could be made not only about the first wave of people entering areas, but when the
next wave of people would enter.
Unfortunately, a CHE where this could be implemented would be ideal for further
research and field testing of the application’s efficiency. Until it is used and tested in an actual
event by field teams, there is no way to measure its efficacy. Once this is field tested possibilities
for other parameters besides weather and terrain can be entered giving more information to
stakeholders. Organizations have been contacted determine to see if possible in the future.
When groups of people are displaced by a CHE, of any sort, there can be preparations
made that will lessen the suffering of those who are displaced. This project and others like it can
help government and NGOs know where refugees and IDP will be fleeing. This can help them
prepare with resources and personnel there ready to help. When used in conjunction with already
available good practices for providing aid in a CHE, it is hoped that this project can reduce the
overall feelings of loss and helplessness that displaced persons exhibit in the course of their
crisis.

38

References
Al Gasseer, N., E. Dresden, G. B. Keeney, and N. Warren. "Status of Women and Infants in
Complex Humanitarian Emergencies." J Midwifery Womens Health 49, no. 4 Suppl 1
(Jul-Aug 2004): 7-13.
Armstrong, M. P. "Perspectives on the Development of Group Decision Support Systems for
Locational Problem-Solving." Geographic Systems 1, no. 1 (1993): 69-81.
Arziman, I. "Field Organization and Disaster Medical Assistance Teams." Turk J Emerg Med 15,
no. Suppl 1 (Oct 2015): 11-9.
Bandana Kar, Michael Hodgson. "A Gis-Based Model to Determine Site Suitability of
Emergency Evacuation Shelters in a Disaster." Transactions in GIS 12, no. 2 (2008):
227-48.
Bumgardner, Wendy. "How Far Can a Healthy Person Walk?" Verywellfit.com,
https://www.verywellfit.com/how-far-can-a-healthy-person-walk-3975556.
Cai, Guoray. "Extending Distributed Gis to Support Geo-Collaborative Crisis Management."
Annals of GIS 11, no. 1 (2005): 4-14.
Churcher, Clare, and Neville Churcher. "Realtime Conferencing in Gis." Transactions in GIS 3,
no. 1 (1999): 23-30.
Crooks, A. T., and A. B. Hailegiorgis. "An Agent-Based Modeling Approach Applied to the
Spread of Cholera." [In English]. Environmental Modelling & Software 62 (Dec 2014):
164-77.
Dunne, Máiréad, Naureen Durrani, Kathleen Fincham, and Barbara Crossouard. "Lebanon:
National Imaginaries, State Fragilities and the Shi’a Other." In Troubling Muslim Youth
Identities, edited by Máiréad Dunne, Naureen Durrani, Kathleen Fincham and Barbara
Crossouard, 252-62: Palgrave Macmillan, London, 2017.
Ellis, C. A., and S. J. Gibbs. "Concurrency Control in Groupware Systems." Paper presented at
the International conference on management of data, Portland, Oregon, USA, June 2,
1989.
Ellis, C. A., S. J. Gibbs, and G. Rein. "Groupware: Some Issues and Experiences."
Communications of the ACM, Jan 1991 1991, 39-58.
ESRI. https://www.esri.com/en-us/industries/sustainability/nonprofit-program/overview.
Goodchild, Michael. "Spatial Accuracy 2.0." Paper presented at the Proceedings of the eighth
international symposium on spatial accuracy assessment in natural resources and
environmental sciences, Shanghai June 25-27 2008 2008.
39

Granell, C., O. B. Fernandez, and L. Diaz. "Geospatial Information Infrastructures to Address
Spatial Needs in Health: Collaboration, Challenges and Opportunities." [In English].
Future Generation Computer Systems-the International Journal of Grid Computing and
Escience 31 (Feb 2014): 213-22.
Guillot, Francois. Children play in the street beside tents at unofficial temporary encampment of
Syrian refugees and migrants in the Porte de Saint-Ouen district of Paris on September
19, 2015. France: Agence France Presse, 2015.
Hall, C., T. P. Dawson, J. I. Macdiarmid, R. B. Matthews, and P. Smith. "The Impact of
Population Growth and Climate Change on Food Security in Africa: Looking Ahead to
2050." International Journal of Agricultural Sustainability 15, no. 2 (2017): 124-35.
Holmes, Seth M., and Heide CastaÑEda. "Representing the “European Refugee Crisis” in
Germany and Beyond: Deservingness and Difference, Life and Death." American
Ethnologist 43, no. 1 (2016): 12-24.
Horner, Mark W., and Michael J. Widener. "The Effects of Transportation Network Failure on
People’s Accessibility to Hurricane Disaster Relief Goods: A Modeling Approach and
Application to a Florida Case Study." Natural Hazards 59, no. 3 (2011): 1619-34.
Hu, Y. J., K. Janowicz, S. Prasad, and S. Gao. "Metadata Topic Harmonization and Semantic
Search for Linked-Data-Driven Geoportals: A Case Study Using Arcgis Online." [In
English]. Transactions in Gis 19, no. 3 (Jun 2015): 398-416.
Irmischer, Ian J., and Keith C. Clarke. "Measuring and Modeling the Speed of Human
Navigation." Cartography and Geographic Information Science 45, no. 2 (2017): 177-86.
Jankowski, Piotr, Timothy L. Nyerges, Alan Smith, T. J. Moore, and Emory Horvath. "Spatial
Group Choice: A Sdss Tool for Collaborative Spatial Decisionmaking." International
Journal of Geographical Information Science 11, no. 6 (1997): 577-602.
Katayoun Jahangiri, Yasamin O Izadkhah, Jila. "Women’s Health in Natural Disasters: A
Vulnerability Analysis." In International Disaster and Risk Conference. Davos
Switzerland: Davos Planet at Risk, 2012.
Kaynak, R., and A. T. Tuger. "Coordination and Collaboration Functions of Disaster
Coordination Centers for Humanitarian Logistics." [In English]. 2nd World Conference
on Business, Economics and Management 109 (2014): 432-37.
Kelly, Rebecca A., Anthony J. Jakeman, Olivier Barreteau, Mark E. Borsuk, Sondoss ElSawah,
Serena H. Hamilton, Hans Jørgen Henriksen, et al. "Selecting among Five Common
Modelling Approaches for Integrated Environmental Assessment and Management."
Environmental Modelling & Software 47 (2013): 159-81.
Lodree, Emmett J., and Selda Taskin. "Supply Chain Planning for Hurricane Response with
Wind Speed Information Updates." Computers & Operations Research 36, no. 1 (2009):
2-15.
40

Longley, Paul, and Michael Batty. Advanced Spatial Analysis: The Casa Book of Gis. University
College London Center for Advanced Spatial Analysis: ESRI Inc, 2003.
MacEachren, Alan M. "Cartography and Gis: Facilitating Collaboration." Progress in Human
Geography 24, no. 3 (2000 2000): 445-46.
MacGuire, David. "The Changing Technology of Space and Time." Chap. 1 In Dynamic and
Mobile Gis, edited by Jane Drummond. Innovations in Gis, 3-18: CRC Press, 2006.
A Dictionary of Geography. Oxford University Press, 2009.
McGee, John. "Enhancing Virginia Cooperative Extension's Reach with Gis." In The Virginia
Geospatial Newsletter, edited by Virginia Geospatial Extension Program, 1,7. Virginia:
Virginia Geospatial Extension Program, 2007.
Newnham, D. S. "The Organization of People in Distress: Unprogrammed, Unintended and
Unstructured Consequences for Refugee Women." [In English]. Learning Culture and
Social Interaction 12 (Mar 2017): 100-12.
Nimani, Armend. Migrants walk on a dirt road towards the Serbia-Croatia border, near the
western-Serbia town of Sid, on September 20, 2015. France: Agence France Presse,
2015.
Pakes, Barry. "Complex Humanitarian Emergencies." In Encyclopedia of Global Health, edited
by Yawei Zhang, 438. Thousand Oaks California: Sage Publications Inc., 2008.
Robertson, A. G., E. K. Griffiths, I. Norton, and T. S. Weeramanthri. "Disaster Response from
Australia: What Is the Role of Forward Teams?". Travel Med Infect Dis 9, no. 5 (Sep
2011): 249-52.
Sanyal, R. "A No-Camp Policy: Interrogating Informal Settlements in Lebanon." [In English].
Geoforum 84 (Aug 2017): 117-25.
Sarin, S., and I. Greif. "Computer-Based Real-Time Conferencing Systems." [In English].
Computer 18, no. 10 (1985): 33-45.
Schafer, Wendy A., Craig H. Ganoe, Lu Xiao, Gabriel Coch, and John M. Carroll. "Designing
the Next Generation of Distributed, Geocollaborative Tools." Cartography and
Geographic Information Science 32, no. 2 (2005): 81-100.
Shaheen, Kareem. "Syrian Doctors Decry Fresh Wave of Hospital Attacks." The Guardian,
https://www.theguardian.com/world/2017/sep/29/syria-renewed-bombing-of-hospitals-
sparks-human-rights-protests.
Sullivan, Brian. "Job Losses since 9/11 Attacks Top 2.5 Million
" Computerworld https://www.computerworld.com/article/2581548/it-careers/job-
losses-since-9-11-attacks-top-2-5-million.html.
41

Sun, Y. Q., and S. N. Li. "Real-Time Collaborative Gis: A Technological Review." [In English].
Isprs Journal of Photogrammetry and Remote Sensing 115 (May 2016): 143-52.
"Syrian Civil War." In Encyclopaedia Britannica, edited by The Editors of Encylopaedia
Britannica: Encyclopædia Britannica, inc., 2017.
Tofighi, S., S. A. Torabi, and S. A. Mansouri. "Humanitarian Logistics Network Design under
Mixed Uncertainty." [In English]. European Journal of Operational Research 250, no. 1
(Apr 1, 2016): 239-50.
UNHCR. "Unhcr's Role in Support of the Return and Reintegration of Displaced Populations:
Policy Framework and Implementation Strategy." edited by UNHCR, 17, 2008.
———. "Unhcr Figures at a Glance." http://www.unhcr.org/figures-at-a-glance.html.
Vaz, E., K. Lee, V. Moonilal, and K. Pereira. "Potential of Geographic Information Systems for
Refugee Crisis: Syrian Refugee Relocation in Urban Habitats." [In English]. Habitat
International 72 (Feb 2018): 39-47.
Williams, H., and E. Downes. "Development of a Course on Complex Humanitarian
Emergencies: Preparation for the Impact of Climate Change." J Nurs Scholarsh 49, no. 6
(Nov 2017): 661-69.
Xu, K. C., R. A. Davidson, L. K. Nozick, T. Wachtendorf, and S. E. DeYoung. "Hurricane
Evacuation Demand Models with a Focus on Use for Prediction in Future Events." [In
English]. Transportation Research Part a-Policy and Practice 87 (May 2016): 90-101.
Xu, Ying, Carleen Maitland, and Brian Tomaszewski. "Promoting Participatory Community
Building in Refugee Camps with Mapping Technology." Paper presented at the Seventh
International Conference on Information and Communication Technologies and
Development, Singapore, Singapore, 2015.
Yin, W. H., P. Murray-Tuite, S. V. Ukkusuri, and H. Gladwin. "An Agent-Based Modeling
System for Travel Demand Simulation for Hurricane Evacuation." [In English].
Transportation Research Part C-Emerging Technologies 42 (May 2014): 44-59.
Yoo, Eunae, William Rand, Mahyar Eftekhar, and Elliot Rabinovich. "Evaluating Information
Diffusion Speed and Its Determinants in Social Media Networks During Humanitarian
Crises." Journal of Operations Management 45 (2016): 123-33.
Yuan, H. S., and C. Q. Zhu. "Shock and Roam: Migratory Responses to Natural Disasters." [In
English]. Economics Letters 148 (Nov 2016): 37-40.
Zhang, Y., W. Z. Wu, Q. Wang, and F. Z. Su. "A Geo-Event-Based Geospatial Information
Service: A Case Study of Typhoon Hazard." [In English]. Sustainability 9, no. 4 (Apr
2017).

42

Appendix

Code
import arcpy

Irmischer = r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\Irmischer"
LandCoverage =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\LandCoverage"
DistanceToCamp =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\RCDistanceToCamp"
ColorMapFile = r"C:\Users\Colleen\Desktop\ThesisData\Colleen\TravelLikelyhood.clr"

EventInput =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\Harasta_20170618"
WeatherInput =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\Weather_20170618"

EventOutput =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\HarastaMovement"

Join_output_temp =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\EventJoin"
Output_temp =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\OneEventBufferDay"
OutputRaster_temp =
r"C:\Users\Colleen\Documents\Colleen_Thesis\Colleen_Thesis.gdb\RasterDay"

fields =['violence', 'death', 'MaxTemp', 'MinTemp', 'Precip']

arcpy.SpatialJoin_analysis(EventInput,WeatherInput,Join_output_temp,"JOIN_ONE_TO_ONE"
,"KEEP_ALL","#","CLOSEST")

for x in range(1,8):

DayOutput = Output_temp + str(x)
DayRaster = OutputRaster_temp + str(x)
ColormapOutput = EventOutput + str(x)
ColormapRange = 0
ViolenceDistance = 18 * x
ViolenceText = str(ViolenceDistance) + " KILOMETERS"
DeathDistance = 20 * x
DeathText = str(DeathDistance) + " KILOMETERS"
DefaultDistance = 16 * x
43

DefaultText = str(DefaultDistance) + " KILOMETERS"
MaxTempDistance = 12 * x
MaxTempText = str(MaxTempDistance) + " KILOMETERS"
MinTempDistance = 10 * x
MinTempText = str(MinTempDistance) + " KILOMETERS"
PrecipDistance = 8 * x
PrecipText = str(PrecipDistance) + " KILOMETERS"

with arcpy.da.SearchCursor(Join_output_temp, fields) as cursor:
for row in cursor:
Violence = row[0]
Death = row[1]
MaxTemp = row[2]
MinTemp = row[3]
Precip = row[4]

if Violence is not None:
if Violence == "yes":
BufferDistance = ViolenceText
ColormapRange = 20
else:
BufferDistance = DefaultText

if Death is not None:
if Death == "yes":
BufferDistance = DeathText
ColormapRange = 40
else:
BufferDistance = DefaultText

if MaxTemp is not None:
if MaxTemp > 95:
BufferDistance = MaxTempText
else:
BufferDistance = DefaultText

if MinTemp is not None:
if MinTemp < 32:
BufferDistance = MinTempText
else:
BufferDistance = DefaultText

if Precip is not None:
if Precip > 0:
BufferDistance = PrecipText
else:
44

BufferDistance = DefaultText

arcpy.Buffer_analysis(EventInput,DayOutput,BufferDistance)

arcpy.PolygonToRaster_conversion(DayOutput,"Day",DayRaster,"CELL_CENTER","D
ay",0.0058)

Colormaptemp = arcpy.Raster(DayRaster)/arcpy.Raster(DayRaster) * ColormapRange

Eventtemp = Colormaptemp + arcpy.Raster(Irmischer) + arcpy.Raster(DistanceToCamp)
+ arcpy.Raster(LandCoverage)

arcpy.Int_3d(Eventtemp,ColormapOutput)

arcpy.AddColormap_management(ColormapOutput,"#",ColorMapFile)

Abstract (if available)

Abstract There are Complex Humanitarian Emergencies (CHE) happening worldwide. Currently, 68.5 million displaced persons exist around the world resulting from these CHEs. This project seeks to develop an ArcPy script for predicting movements within these crises to help government and non-government agencies manage resources. This project is intended to predict large-scale flows of people to allow agencies to prepare for an influx of displaced persons and organize shelters, food, clothing, medical personnel, and other necessities. ArcPy within ArcGIS Pro will give non-trained users an easy format to view spatially relevant processes and information about the CHE that they may be unable to share with the public. Additionally, this project looks at theoretical models of human movement that are most relevant to the circumstances of the CHE and should help speed communication between GIS and non-GIS professionals. This project uses the Syrian refugee crisis as a test case because of the large-scale movement and long duration involved.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Using GIS to perform a risk assessment for air-transmitted bioterrorism within San Diego County

PDF

Integrating GIS into farm operations at the Homer C. Thompson Research Farm in Freeville, New York

PDF

Classifying the alpines: developing a methodology to track environmental changes in the alpines utilizing remote sensing and ecozone vegetation patterns

PDF

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

PDF

Development of a Web GIS application to aid marathon runners in the race selection and planning process

PDF

Harnessing GIST-enabled resources in the classroom: developing a Story Map for use with secondary students

PDF

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

PDF

Wetland mapping and restoration decision making using remote sensing and spatial analysis: a case study at the Kawainui Marsh

PDF

A spatial narrative of alternative fueled vehicles in California: a GIS story map

PDF

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

PDF

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

PDF

Radar horizon estimation from monoscopic shadow photogrammetry of radar structures: a case study in the South China Sea

PDF

Evaluating surface casing depths of oil & gas operations in an effort to protect local groundwater: a GIS enabled process

PDF

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

PDF

Developing art-based cultural experiences in North Kohala: A community engagement project with OneIsland

PDF

Using aerial imagery to assess tropical forest cover surrounding restoration sites in Costa Rica

PDF

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

PDF

Collecting and managing VGI infrastructure assessments in support of stability operations

PDF

Use of least-cost path analysis to identify potential movement corridors for jaguars across the US-Mexico border

Asset Metadata

Creator Campbell, Colleen Marie (author)

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

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 04/04/2019

Defense Date 04/04/2019

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

Tag complex humanitarian event,GIS,migration,OAI-PMH Harvest,refugee,Syria

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Fleming, Steven Douglas (committee chair), Bernstein, Jennifer Moore (committee member), Marx, Andrew (committee member)

Creator Email cmcampbe@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-137287

Unique identifier UC11675448

Identifier etd-CampbellCo-7182.pdf (filename),usctheses-c89-137287 (legacy record id)

Legacy Identifier etd-CampbellCo-7182-1.pdf

Dmrecord 137287

Document Type Thesis

Format application/pdf (imt)

Rights Campbell, Colleen Marie

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