University of Southern California Dissertations and Theses

A comparison of address point and street geocoding techniques in a computer aided dispatch environment

(USC Thesis Other)

A comparison of address point and street geocoding techniques in a computer aided dispatch environment

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content A COMPARISON OF ADDRESS POINT AND STREET GEOCODING TECHNIQUES
IN A COMPUTER AIDED DISPATCH ENVIRONMENT

by

Jimmy Tuan Dao

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)

August 2015

Copyright 2015 Jimmy Tuan Dao

ii

DEDICATION
I dedicate this document to my mother, sister, and Marie Knudsen who have inspired and
motivated me throughout this process. The encouragement that I received from my mother and
sister (Kim and Vanna) give me the motivation to pursue this master's degree, so I can better
myself. I also owe much of this success to my sweetheart and life-partner, Marie who is always
available to help review my papers and offer support when I was frustrated and wanting to quit.
Thank you and I love you!

iii

ACKNOWLEDGMENTS
I will be forever grateful to the faculty and classmates at the Spatial Science Institute for their
support throughout my master’s program, which has been a wonderful period of my life. Thank
you to my thesis committee members: Professors Darren Ruddell, Jennifer Swift, and Daniel
Warshawsky for their assistance and guidance throughout this process. I also want to thank the
City of Brea, my family, friends, and Marie Knudsen without whom I could not have made it this
far.
Thank you!

i

TABLE OF CONTENTS
DEDICATION ii
ACKNOWLEDGMENTS iii
LIST OF TABLES iii
LIST OF FIGURES iv
LIST OF ABBREVIATIONS vi
ABSTRACT vii
CHAPTER 1: INTRODUCTION 1
1.1 Brea, California 2
1.2 Motivation 5
1.3 Research Questions and Objectives 6
1.4 Thesis Organization 6
CHAPTER 2: BACKGROUND AND LITERATURE REVIEW 8
2.1 Data Accuracy 15
CHAPTER 3: METHODOLOGY 18
3.1 Data Analysis 19
3.2 Street Ranges 20
3.3 Georeferencing 22
CHAPTER 4: RESULTS 32
4.1 Calls-for-Service 32
4.2 Geocoding 32
4.3 Single House Locator 33

ii

4.4 Dual Ranges Locator 35
4.5 Accuracy Comparisons 39
CHAPTER 5: DISCUSSION AND CONCLUSIONS 43
5.1 Next Steps 45
REFERENCES 48
APPENDIX A: Brea GIS Layers for CAD 53
APPENDIX B: Software for Building Brea CAD System 54
APPENDIX C: Brea Street Type Field Definition 54
APPENDIX D: Study Area: City of Brea Boundary 55
APPENDIX E: City of Brea’s Computer Aided Dispatch 56

iii

LIST OF TABLES
Table 1. Brea Street/Centerline Field Definition Table from ArcMap 21
Table 2. Brea Address Point Attribute Table and Field Definitions 30
Table 3. Monthly Response Times from January to November, 2014 (Provided by Brea Police
2014) 46

iv

LIST OF FIGURES
Figure 1. City Boundary Map of Brea, California .......................................................................... 3
Figure 2. Freeway Complex Fire, provided by City of Brea, 2008 ................................................ 9
Figure 3. Diagram of a geocoder process (Esri 2014) .................................................................. 11
Figure 4. Diagram of a geocoding service in progress (Esri 2015) .............................................. 12
Figure 5. Map showing TIGER/Line segments and address ranges (U.S. Census 1997)............. 14
Figure 6. Project Workflow Diagram ........................................................................................... 19
Figure 7. Street Layer Review Process for CAD Conversion ...................................................... 20
Figure 8. Georeferencing Procedures for Digitizing Address Points and Street Layers .............. 23
Figure 9. Proposed New Subdivision, Using Brea 3” Aerial Imagery ......................................... 24
Figure 10. Georeferenced Building Plan of New Subdivision ..................................................... 25
Figure 11. Digitized results of the New Subdivision .................................................................... 26
Figure 12. Workflow Process for Address Points Creation to Support Brea’s CAD System ...... 29
Figure 13. Create Common Places for CAD System ................................................................... 31
Figure 14. Revised Address Table for Geocoding ........................................................................ 33
Figure 15. Single House Address Locator Setup for Geocoding Address Points ......................... 34
Figure 16. Geocoding Results from Single House Locator .......................................................... 34
Figure 17. Geocoded Results (8,148 out of 9,671) from Single House Address Locator ............ 35
Figure 18. Example of the reference data used for a dual ranges address locator (Esri 2014) ..... 36
Figure 19. Street Reference Layer for Dual Ranges Locator Table in ArcMap ........................... 37
Figure 20. Dual Ranges Address Locator and Street/Centerline Feature Class ........................... 38
Figure 21. Geocoded Results from Dual Ranges Address Locator .............................................. 38
Figure 22. Geocoded Results (9,250 out of 9,671) from Dual Ranges Address Locator ............. 39

v

Figure 23. Comparisons of Geocoded Results .............................................................................. 40
Figure 24. Analysis of Geocoded Results ..................................................................................... 41

vi

LIST OF ABBREVIATIONS
AVE Avenue
BLVD Boulevard
BPD Brea Police Department
CAD Computer Aided Dispatch
CIR Circle
CT Court
DR Drive
EOC Emergency Operations Center
Esri Environmental Systems Research Institute
GIS Geographic Information System
IT Information Technology
LN Lane
LP Loop
PKWY Parkway
PL Place
POI Point of Interest
PDF Portable Document File
RD Road
ST Street
TER Terrace
TIFF Tagged Image File Format
TRL Trail

vii

ABSTRACT
Understanding address points and street ranges is critical for providing information quickly and
accurately to emergency responders. This thesis investigates the process of updating address
points and street ranges in a computer aided dispatch (CAD) environment to help improve
response time for emergency services while developing a more reliable geocoder for CAD. In a
geographic information system (GIS), addresses verify through a process called geocoding, a
topic that is currently being studied and tested in many CAD environments. Geocoding is one of
the most critical components in CAD because Dispatchers depend on it to accurately confirm the
location and relay the information to first responders. Based on the applied work experience and
lessons learned in supporting CAD, an exact match to the property, or calls-for-service locations,
are critical and can potentially save lives. Using street ranges for address verification is not as
accurate as address points because street ranges only provide an approximation of location,
which can require additional efforts to locate the caller and increases response time. Ideally,
Dispatchers require each call point be provided as an exact physical location. This investigation
examines the City of Brea, California as a case study on GIS administration in the capacity of
maintaining and updating GIS data for CAD use. Verifying emergency call requests is one of the
most important functions, allowing Dispatchers to send appropriate aid expeditiously. Therefore,
accurate and current address point and street range information are critical in the performance of
CAD functions. The results of the research inform the fitness of use and accuracy of address
points versus street ranges in a CAD environment for the City of Brea, California. Moreover, this
research aims to promote greater data sharing and interagency cooperation among local, county,
and state agencies in the United States.

1

CHAPTER 1: INTRODUCTION
This research aims to inform the fitness of use and accuracy of address points versus street
ranges in a CAD environment, and to promote greater data sharing and interagency cooperation
among local, county, and state agencies. In addition, results of this research were used to validate
the superiority of address points versus street ranges in the City of Brea, California CAD system.
The City of Brea continues to embrace GIS technologies because its functionality is currently
integral to the success and deployment of CAD. The main goal of geocoding technology for the
City’s CAD system is to pinpoint the locations of 911 calls as precisely as possible, resulting in
faster response times.
There are various ways in which geocoding techniques are applied, such as locating and
responding to 911 calls, or to the investigation on Dengue fever in many parts of Africa
(Goldberg 2011). As street addresses help individuals to locate one another in this complex
world (Zanbergen 2008), in GIS, geocoding, which is the process of verifying address locations,
is a topic that dates back to the early 1960s when a team of researchers from the United States
Census Bureau studied how to better validate, locate, and store the nation’s address information
(O’Reagan and Saalfeld 1987).
In a CAD environment, one of the most critical components for Dispatchers is to
accurately verify the location of a property, or calls-for-service, resulting in faster emergency
response time. Herein, calls-for-service refers to 911 calls initiated by the public that requires
immediate response from public safety personnel for aid. In this matter, an exact match to the
property is preferred. Using street ranges for address verification is not as accurate as address
points. This is because street ranges provide an approximation of the location, resulting in greater
effort to find exact locations and thus reduced response time. Therefore, Dispatchers prefer that

2

each call points to the exact physical location. Additional discussions on what defines an exact or
approximate match will be provided herein.
One of the significant challenges facing the GIS community is to maintain and update
geospatial data for CAD. The City of Brea’s CAD system, which is operated under the Police
Department (PD) and supported by their in-house Information Technology Division (IT), is fully
dependent on their GIS data. These data layers include address points, street ranges (or address
ranges), parcel information, property lines, reporting districts, beat boundaries, city boundary,
surrounding cities, fire stations, police stations, parks, schools, and retail locations. The city
boundary, beat boundaries, and reporting districts do not require frequent-updates, since their
spatial references seldom change. On the other hand, address points and street ranges do require
monthly updates. As verifying emergency call requests is one of the most important functions of
the City of Brea’s CAD system, allowing Dispatchers to send appropriate aid expeditiously,
accurate, current address points, and street range information are critical in the performance of
CAD.
1.1 Brea, California
The City of Brea, once a small oil producing town surrounded by wildland areas in north Orange
County, is now a bustling city, close to reaching its maximum building limits, based on the
City’s general plan studies (see Figure 1).

3

Figure 1. City Boundary Map of Brea, California
Brea is approximately 30 miles east of Los Angeles, with a population of over 42,000, and
daytime population of over 100,000 that has been increasing steadily since the 1970s, with the
opening of the Orange Freeway (57) and Brea Mall (City of Brea 2015 ). As mentioned
previously, Brea has its own police department, an outstanding public school system, as well as
fire services, and active commercial and residential sectors that make it one of the most desired
places to live and work (Brea Olinda Unified School District 2015). As the continued growth of
the City further intensifies the demand to provide public safety, the CAD system needs to
geocode addresses correctly to ensure faster response times to 911 calls.
There are many benefits for an interagency CAD system, such as shared deployment
costs, better management of GIS datasets, consistency, and more knowledge sharing opportunity
among the GIS community. Unfortunately, at this time, similar to other agencies in Southern
California, Brea’s proprietary CAD system is an obstacle for sharing resources with neighboring
agencies. Another motivation for this study, it was becoming more difficult for Dispatchers to
verify calls-for-service locations, resulting in longer response times.
Prior to this study, common CAD issues were becoming difficult to resolve within the
City’s existing IT system, including: (1) problems with updating and removing existing address

4

points, common place names, and contact information; (2) the inability to verify calls-for-service
and the challenges posed by street information that was out of range; and (3) addresses did not
match police reporting districts. As these issues progressed, moving to a GIS based CAD system
provided dispatch the optimal tools necessary to respond to emergency calls as a means of
improving public safety for Brea residents.
Over the last several decades, advancement in GIS functionality has made it an integral
part of supporting any CAD systems (Babinski 2009). For example, improved location-based
data access, advanced software, and hardware are providing many more opportunities in the uses
of GIS technologies for supporting CAD (Newcombe 1994). In 2007, Ray Drlik, former Brea
Information Technology Manager, hired the City’s first full-time GIS Analyst to support city
staff and their new GIS-based CAD system. In the public safety sector, GIS has been used for
decades to identify, record, and respond to emergency requests (Nesbary 2001). Many
emergency coordinators, Dispatchers, and first responders rely on GIS to pinpoint emergency
call locations and provide appropriate responses quickly (Newcombe 1994). Today’s CAD
systems nationwide typically depend on GIS to identify and respond to emergency calls.
Improved alternative solutions in the management of a CAD system are more possible
today due to the continued advancement in GIS software, hardware, data, and more qualified
GIS professionals (Mehrotra et al 2013). Today’s CAD systems have moved far beyond the
Microsoft Disk Operating System (MS-DOS) base that used command lines and identified
emergency service calls on a monochromatic computer screen (Mehrotra et al 2013). Current
CAD systems are visually impactful, with a map-based graphical user interface that supports
quick identification of the caller, their location, and response to emergency calls.

5

The integration of GIS and spatial data in CAD continues to grow (Dvorak 1997). In the
early 2000’s CAD administrators in Brea demanded a better visualization or user interface as
part of their dispatch system. In 2007, the City of Brea began the process of replacing their
existing CAD with a graphical user interface (GUI) and map-based system. After a lengthy
selection process, Intergraph Corporation (Intergraph 2015), a premier geospatial and CAD
vendor, was awarded the contract to develop Brea’s next generation CAD 911 system. Moving to
an Intergraph based CAD system required extensive data migration. Intergraph uses proprietary
software, customized interfaces, and data formats that are different than the City’s Esri based
GIS infrastructure. The software and data challenges during this Intergraph data migration
involved consolidating Esri’s feature classes into Intergraph’s data format for its CAD mapping
system. Since Brea uses Esri software and data architecture, it was a challenge to migrate the
City’s data to work in Intergraph, and took over a year to accomplish.
1.2 Motivation
The role of local government is to serve its residents. There are many services local government
provides (Babinski 2009), such as home inspections, trash services, road maintenance, building
permits, business licenses, meal services for low-income seniors, and police and fire services for
the community. As the demand for these services increases, particularly in public safety, cities
like Brea are motivated to find better technology to facilitate services provided to their
constituents. Advancement in GIS software, computer hardware, and sensor technologies make it
more plausible for a local agency to be more effective in providing public safety for its citizens.
Without these advances and lowered deployment costs, a GIS based CAD system would not be
possible in Brea.

6

1.3 Research Questions and Objectives
There were two questions in this research. First, which address locator is more accurate to use in
a CAD environment? Second, does this research promote greater interagency cooperation and
data sharing within the GIS community who supports CAD systems nationwide?
In order to answer each of these questions, four objectives were set for this thesis. This
thesis aimed to: (1) analyze over 16 thousands calls-for-service locations in Brea between
January 1 to December 15, 2014; (2) geocode the calls-for-service locations and compare the
accuracy rates between the Single-House Locator and Dual-Ranges Locator; (3) compare the
geocoded results through field inspection to determine its proximity to the physical locations;
and (4) determine the fitness of use and accuracy of address points versus street ranges in a CAD
environment for the City of Brea. Several techniques were utilized in order to accomplish the
study’s research objectives, including relevant work experience in the administration of GIS data
for CAD.
1.4 Thesis Organization
Chapter one begins with a background of current work in the maintenance of CAD within the
GIS community and review of relevant research articles on the process of updating address
points and address ranges for CAD.
Chapter two examines the literature reviews incorporated in this study, including an
overview discussion on geocoding techniques and geospatial data accuracy of the study area, and
how the GIS data was prepared.
Chapter three discusses the methodology applied in this study, including an overview of
the study area, analysis of the CAD data sources utilized, and how the GIS input data was
prepared. In addition, the procedure used to complete this study is explained: a comparison

7

analysis of street ranges and address points, georeferencing techniques, address point creation,
assigned common places, and project workflow in a CAD environment.
The results, Chapter four, provides an analysis of 911 call locations for the year 2014 in
the City of Brea, including how the calls compared when geocoded against multiple address
locators. Two address locators are included in the analysis: Single House Address Locator and
Dual Street Ranges Address Locator, which produced different matched rates to assess overall
accuracy.
Chapter 5, the final chapter, discusses the conclusions determined based on the results,
the validity of the results, the challenges within the GIS community in managing a CAD system,
and the future work to be considered.

8

CHAPTER 2: BACKGROUND AND LITERATURE REVIEW
For some time now, geocoding has been a term only understood by individuals and organizations
in disciplines such as geography and spatial sciences. At the same time, advancement in
computer hardware and mapping software has been crucial in the uses of geocoding in today’s
globally connected world. For example, Google Maps, Apple’s iPhones, and the navigational
system in many vehicles use geocoding technology to provide driving directions to the nearest
steakhouses or recommend alternate routes during unexpected road closures.
Moreover, GIS information is widely used in many government agencies for emergency
preparedness. For example, during a natural disaster, access to pertinent information quickly can
mitigate potential loss of lives and property. In the early morning of November 15, 2008, a brush
fire from Corona, California, quickly spread to Brea, due to the extreme dried vegetation and
changing wind conditions (Freeway Complex Fire 2013). All designated City personnel
(including GIS and Dispatch) reported to the Emergency Operations Center (EOC) at Brea City
Hall to assist with information gathering and logistics. As shown in Figure 2, City staff worked
with county and state agencies to gather pertinent information and produce an assessment map
showing the affected burn areas, size, and agencies involved in fighting the wildfire.

9

Figure 2. Freeway Complex Fire, provided by City of Brea, 2008
Using existing GIS data from the City of Brea, information provided by the Orange County Fire
Authority and CAL Fire (statewide fire agency), the Freeway Complex Fire showed that
interagency cooperation for emergency response is possible.
The term geocoding may represent different things to different individuals, but, in
essence, it means coding the Earth with geographic reference information (or data layers) that
can be reconciled for computer mapping uses (Harries 1999). The history of geocoding goes
back to the early 1960s and the U.S. Census Bureau, to identify methods of mapping data
collected in the field across the country, address by address (Harries 1999). For example, the
major development of automating the geographic coding of postal addresses was accomplished
by the Census Bureau in 1963 (O’Reagan and Saalfeld 1987), leading to the evolution of today’s
geocoding technology. For Brea’s CAD system, the geocoding of 911 calls is supported by

10

Intergraph mapping software that relies on the underlying data layers (address points and street
ranges) to be as accurate as possible. Today, there are numerous geocoding algorithms described
in the literature and used in open source and proprietary GIS software packages. Detailed
information regarding the mathematics and computational technology underlying the science of
geocoding can be found in Zhan and et al (2006) article, thus is not presented in this thesis. A
synopsis of this technology and the geocoding process utilized in this thesis work is provided in
Figures 3 and 4 below (Esri 2014).

11

Figure 3. Diagram of a geocoder process (Esri 2014)

12

Figure 4. Diagram of a geocoding service in progress (Esri 2015)
When configured in a GIS application (desktop or web-based), a geocoder is one of the most
powerful geoprocessing tools, able to quickly and accurately validate addresses. As shown in
Figures 3 and 4, geocoding requires multiple processes and calculations to take contextual
information (127 West Point Drive) and reconcile with the reference information to produce GIS

13

features for uses in a computer mapping application. In addition, these diagrams offered a
framework to the methodologies associated with this thesis and its findings.
The Esri ArcGIS geocoder offers different types of address locator styles which can be
configured by the user. An address locator is the main tool for geocoding in Esri’s ArcGIS
software; it converts textual descriptions of information into geographic features (Esri 2013). On
the other hand, the address locator style, is the skeleton of the address locator, it determines the
types of addresses that can be geocoded, communicates with the reference dataset, and
determines the output information (Esri 2013). The critical issue is that the choice of an address
locator can be extremely difficult for GIS users. Structure of Esri ArcGIS software allows an in-
depth comparison of address locators, which has not been previously documented in the
literature.
In a GIS application (desktop or web-based), many complex spatial data layers sit on top
of one another, displayed on the computer screen often in multiple colors and symbols intended
to present a full picture of the overall subject matter. GIS maps can display complex information
in a way that is simple to understand. A geocoder can be a GIS user’s best friend, helping to
retrieve location information quickly.
For instance, the Topologically Integrated Geographic Encoding and Referencing
(TIGER) data, developed by the U.S. Census Bureau, is an example of a reference dataset used
to geocode addresses nationwide (Klosterman and Lew 1992). The TIGER file is a polyline
based file, with address ranges tied to each polyline or street segment as shown in Figure 5.

14

Figure 5. Map showing TIGER/Line segments and address ranges (U.S. Census 1997)
Each street segment, or line, has low and high values at the line’s endpoint, such as 200 and 299
S Brea Blvd (Kellison 2012). In this case, 250 S Brea Blvd would be assigned by the geocoder to
the middle of the street segment.
Conversely, a failed geocoding result will occur if an address is entered outside of the
given street range provided in the reference dataset, because the geocoder only recognizes textual
information that is within its given range. Therefore, the success of a GIS application requires
that the reference dataset, like the TIGER data, be fully vetted to assure that it successfully
searches for addresses. Depending on the data quality, each street segment should include
attribute fields that are used to identify the street name, street type, directional prefix, suffix, and
address ranges. In a street range geocoder, the geocoded information, based on the reference
dataset, is often an estimation of the physical location (Davis Jr. and Fonseca 2007). On the other
hand, in an address point-based locator, each result should be matched exactly to the reference
data (Brosowsky and Ekdahl 2012). For a consistent geocoder, frequent updates to the address
range dataset is mandatory. In addition, the user also needs to provide the information accurately,
such as filling in the address completely (i.e. “140 S Brea Blvd, Brea, CA, 92821”), without
grammatical errors, correct spelling of street name, abbreviation of a street type, and avoid
formatting errors (Yildirim et al. 2013).

15

Goldberg and Cockburn (2010) developed a best-match criterion to describe the choices
used by a geocoder to determine the best output from a set of candidate results. The methodology
includes several approaches to enhance the efficiency of a geocoder. They devised a hierarchy-
based approach where reference datasets are placed into a qualitative and arbitrary ranking and
the address locator listed first will be used first. This method assumes relative accuracy between
the reference layers. Another is an uncertainty-based approach where the reference data sources
are first sorted from low to high based on their position within the hierarchy.
2.1 Data Accuracy
Great strides have been developed in the last twenty years to improve the accuracy of digitized
data in GIS to facilitate applications like CAD, because data accuracy is critically important for
the myriad of uses of GIS data. In the case of geocodes, the more accurate the results, the more
lives can be saved.
Accuracy in a digitization environment is affected by the characterization of positional
error, where the error is contributed by the operator or media type (Bolstad et al 1990). Bolstad
and et al (1990) examined the findings of manually-digitized point data sampled by four
operators from Mylar to paper maps. The methodology used in the study included sample
location points from United States Geological Survey (USGS) 1:24,000 scale maps. USGS maps
were commonly used as reference maps for creating and digitizing GIS data types like points,
lines, and polygons. Bolstad and et al (1990) also discussed the concepts of accuracy and
precision as useful in explaining the positional irregularity of GIS coordinate data. Bolstad
(1990) sees accuracy as a measurement of the nearness of quantities to their true value, and
precision measures the degree of conformity of measurements among themselves. For instance,

16

in GIS, digital data accuracy can be reflected in errors on the map attribute database and
variation of errors that are related to the reference datasets.
Another important method for evaluating accuracy and precision of digital map data
includes the comparison of field survey data and coordinates derived from a GIS data layer.
However, when converting paper maps to digital data, some map sources contain errors that were
derived from the original paper maps. If map features were not correctly mitigated the errors may
be transferred to the digital dataset. The process of collecting GIS data, whether in the field or
over the internet requires advanced and careful planning. To convert analog to digital map data
includes manual input of vector data, hands-on digitizing, topographic maps or high-resolution
aerial imagery. For data collection and improved accuracy, manual digitization is still a common
form of collecting spatial data to convert paper maps to vector layers in GIS or other types of
environmental and engineering design software. In past decades, preparation for collecting
manually-digitized GIS data entailed setting up map sheets to a flat table surface. A digitizing
board and cursor store reference points on an analog map and transfers the control points to a
workstation. As a result, systemic errors from the operator or hardware can cause degradation to
the digitized data and uncertainty associated with manual digitization is small compared to other
sources within this study (Bolstad et al. 1990).
GIS data collection techniques have come a long way in the last twenty years. Vast
improvements in collecting control points for digitizing improve the data accuracy. In addition,
with the right software and data sources, today’s methods of digitizing data are vastly more
efficient than twenty years ago. A digitizing board is no longer necessary to capture control
points. Today, GIS software can be used to easily import images in many digital formats from
scanned paper maps. Digitizing is still considered “manual” when GIS-based drawing tools are

17

used to trace features from georeferenced images by hand using a mouse. For instance, ArcGIS
desktop software from Esri allows users to create control points and digitize plan maps without
the need of sophisticated tools and hardware (Esri 2013). In addition, today there are many high
accuracy digital base maps, including topography, satellite imagery, and high-resolution
orthophotography (3-inch aerial imagery), available to be integrated within a GIS that can be
used to improve the collection of control points and manually digitized data layers. Access to
high resolution imagery data from the private and public sectors have vastly improved data
collection techniques in today’s GIS. Nevertheless, common digitization errors that can affect the
accuracy of the resulting data include digitizing data at the wrong scale and again not mitigating
errors on the original maps, transferring them to the digital version.
Although there are many geocoders available, Esri geocoding tools provide the most
consistent and accurate methods in the comparisons of address points versus street ranges for the
City of Brea, because the City’s GIS infrastructure (datasets and software) is based on Esri’s file
types. Data accuracy is important in this study because it affects response times. The more time
that it takes Dispatchers to verify, or geocode 911 calls, the longer it takes for emergency service
personnel to respond. Therefore, ongoing updates improve the accuracy of Brea’s GIS datasets.
The City’s collection and update techniques include manual digitization of paper maps that are
georeferenced and incorporated into existing data references, such as streets, parcels, city
boundaries, and 3-inch aerial imagery. In the next chapter the methodology for testing the
hypothesis stated in Chapter 1 will be implemented using the Esri geocoding styles: Single
House Address Locator and Dual Ranges Locator.

18

CHAPTER 3: METHODOLOGY
Address points-based locator is the most appropriate for geocoding calls-for-service locations in
a CAD environment. This study utilizes research techniques and skillsets developed through
more than a decade of professional work experience within the public sector. CAD is a consistent
work-in-progress system, requiring the cooperation between Dispatch and GIS personnel
working together to resolve geocoding issues. When a 911 call fails to geocode from the address
points-based locator, the CAD system then uses the street ranges locator to geocode the location.
Development of a secondary geocoder provides Dispatch a temporary resolution to verify 911
calls. 911 call locations that were not successfully geocoded from the address points-based
locator are evaluated and added to the data reference to be updated to the CAD system.
The research for this study included field work and data collection of addresses. Common
place names and new street names were incorporated into the input address dataset. In addition,
non-physical research techniques include digitizing and georeferencing building plans of new
developments. The methodology (Figure 6) used in this thesis is to compare and evaluate the
accuracy of a single house address locator versus a dual ranges address locator, using the
following steps: (1) identify an address point locator style to reference with the address point
feature class; (2) identify the dual ranges street locator style and assign a reference street layer;
(3) assign both address locators with similar geocoding options, such as spelling sensitivity and
matched rate criteria; (4) assign address table to be geocoded by both address locators; (5)
review the geocoded results and overall matched rates from both locators; (6) create a visual
inspection of the geocode address points using ArcMap and measuring its physical accuracy; and
(7) assign ground-truthing methodologies through testing from the CAD conversion process and
confirmation from Dispatch.

19

Figure 6. Project Workflow Diagram
3.1 Data Analysis
Since most government agencies today have access to GIS technologies, a GIS based CAD
system is more feasible (Chatterton 1987). At the same time, each city in the United States
incorporates GIS differently. In some cities, GIS serves a critical need to city functions, fully
supported by a team of trained GIS professionals. For other cities, GIS is minimally utilized,
usually due to a lack of awareness, untrained staff, and limited infrastructure resources.
Today, the base of building a successful CAD system rests on access to reliable GIS data,
free of topology errors, such as overlapping segments and slivers. Prior to the data migration to
Intergraph, a full data assessment of the City of Brea’s GIS revealed that significant data
cleaning efforts were needed. Fourteen critical feature layers from the City’s GIS inventory,
including address points and streets/centerlines layers, were insufficient in regards to topological
errors, overlapping segments, missing attribute fields, and required a complete overhaul. For
CAD to work correctly, the address point and street layers must accurately verify emergency
calls. Additional GIS layers, such as the city boundary, parcels, police reporting
districts/emergency service zones, police beat areas, and surrounding cities were reviewed to see

20

if additional revisions were required for these to be effective CAD supporting data layers. More
layers were added later, such as evacuation centers, parks, schools, shopping centers, city
facilities, and hazard areas to create a richer map experience.
The data migration to an Intergraph based CAD system in Brea uncovered many
challenges. Early in the data assessment, the existing city GIS data required significant revisions,
or cleaning. Cleaning tasks included fixing geometry errors in the streets layer, removing slivers
from the police reporting districts, updating the police beat boundaries, and adding attribute
information to the address point layer. Updated parcels, address points, streets/centerlines, and
reporting districts were identified as critical operational layers. Organizing and cleaning the data
layers helped to successfully migrate the City’s CAD to Intergraph, completed in 2008. Since
2008, Brea staff has taken over the responsibilities for supporting CAD, including ongoing
software and GIS data updates to facilitate Dispatch’s needs.
3.2 Street Ranges
The process for reviewing street segments for conversion to the CAD system updates is further
illustrated in Figure 7, showing the complexity and efforts required during each update process.

Figure 7. Street Layer Review Process for CAD Conversion
Step 1
•Street name fully parsed into individual fields.
Step 2
•Assigned to local projection: State Plane, NAD 1983, CA Zone6 US Feet.
Step 3
•EntityID Field (Brea_ID), with unique value tied to each address point.
Step 4
•No Topology issues: overlapping and incomplete segments.
Step 5
•No duplicate segments.
Step 6
•No Duplicate number values in the "BREA_ID" (EntityID) field.

21

In addition, the data layers mentioned in the previous section are provided by the City of Brea
and will be used to develop a workflow model in building an Intergraph based CAD system.
In the City of Brea, a streets/centerline layer that provides accurate coverage for the area
that the Brea Dispatch is responsible for was critical to the success of the system. At the time,
Brea provided police and dispatch services to the residents of Brea and Yorba Linda, California
(neighboring city). The existing street layer did not provide sufficient attribute information to
meet the demand by Dispatch. As such, the City purchased a new street/centerline dataset from
Tele Atlas, a geospatial company that provides routing information for GIS applications
(desktop, web-based) and other navigational devices, and was acquired by TomTom in 2007
(Hoef and Kanner 2007). The Tele Atlas street dataset included updated address ranges and
routing information, single and dual-line street segments for Brea and the surrounding cities
(TomTom 2015). The street layer has fields to identify speed limits, one way streets, address
ranges, overpasses, underpasses, and freeway ramps, detailed in Table 1. This street/centerline
layer gives CAD the capability to track, identify, respond to emergency calls, and send the
appropriate help more effectively than the City’s legacy system.
Table 1. Brea Street/Centerline Field Definition Table from ArcMap
FIELD NAME DESCRIPTION
BREA_ID Unique record number for each street segment/feature
L_F_ADD Left from address
L_T_ADD Left to address
R_F_ADD Right from address
R_T_ADD Right to address
PREFIX Street prefix
NAME Street name
TYPE Street type
SUFFIX Feature direction suffix
FCC Feature Class Code
POSTAL_L Postal code (ZIP or FSA) left

22

FIELD NAME DESCRIPTION
POSTAL_R Postal code (ZIP or FSA) right
ACC Arterial Classification Code
NAME_TYPE “R” (always PRN for this product)
SHIELD Shield (“I,” “U,” “S,” “T,” “A” or blank) and shield subtype
HWY_NUM #, # with letter, or blank
SEG_LEN Segment length in miles
SPEED Speed in miles per hour
ONE_WAY One‐way indicator
F_ZLEV From node elevation
T_ZLEV To node elevation
FIELD NAME DESCRIPTION
FT_COST From‐To Impedance in minutes
TF_COST To‐From Impedance in minutes
FT_DIR From_To Direction
TF_DIR To_From Direction
NAME_FLAG Name metadata flag
STATUS Street category
FULL_NAME Combined address fields (PREFIX, NAME, TYPE, SUFFIX)
ALIAS1_NAM Street name alias 1
ALIAS2_NAM Street name alias 2
ALIAS3_NAM Street name alias 3
ALIAS1_PRE Street prefix alias 1
ALIAS2_PRE Street prefix alias 2
ALIAS3_PRE Street prefix alias 3
ALIAS1_TYP Street type alias 1
ALIAS2_TYP Street type alias 2
ALIAS3_TYP Street type alias 3
ALIAS1_SUF Street feature direction suffix alias 1
ALIAS2_SUF Street feature direction suffix alias 2
ALIAS3_SUF Street feature direction suffix alias 3

3.3 Georeferencing
Georeferencing is a technique used to geoenable images without spatial metadata so that the
information on the images can be transferred from the rasters to new vector data through
digitizing, such as tracing building plans as points, lines, and polygons. The process of

23

incorporating georeferencing for updating new streets and address information in Brea is further
illustrated in Figures 8 to 11. When a new development is approved, the Building Division
emails city staff, including Dispatch and the GIS Administrator of the changes. The email
includes information on new addresses, street information, and an attached portable document
file (PDF) of scanned plan(s) of the area. The GIS Administrator then converts the attached
scanned plans to a tagged image file format (TIFF). The TIFF map file then gets added into
ArcMap and overlays existing layers such as streets, parcels (if available), and imagery as
reference layers for georeferencing. Georeferencing includes adding control points obtained from
the TIFF map file and linking these points to the spatial georeferenced GIS layer(s). Then the
digitized address points and street information are added to the existing reference layers.

Figure 8. Georeferencing Procedures for Digitizing Address Points and Street Layers
Step 1
•Open ArcMap and add imagery, streets, address points and parcel (if available) layers.
Step 2
•Zoom to the areas where the new homes will be built.
Step 3
•Add the TIFF image file to ArcMap.
Step 4
•Confirm that the Georeferencing toolbar is activated and available in ArcMap.
Step 5
•Select the TIFF image file from the Georeferencing toolbar.
Step 6
•In Georeferencing, select the "Fit to Display" option.
Step 7
•Add Control Points and align to the referenced GIS layer.
Step 8
•Begin an editing session of the Address Point layer and add new points and attribute information.
Step 9
•Follow step 8, but this time digitize new streets and include address ranges.
Step 10
•Save the edit session.

24

Figure 9. Proposed New Subdivision, Using Brea 3” Aerial Imagery

25

Figure 10. Georeferenced Building Plan of New Subdivision

26

Figure 11. Digitized results of the New Subdivision
For Dispatch, address points are the preferred data reference for geocoding calls-for-
service. However, when a 911 call fails to geocode from the address point layer, the CAD system
mitigated that problem by using the streets/centerlines as a secondary reference data layer. The
street layer provides Dispatchers the flexibility to geocode emergency calls when CAD is unable
to verify from the address points dataset. Therefore, a consistent update to the street layer is
vitally important. However, further discussion in this document will show that address ranges
alone are not sufficient to produce the most accurate address verification (Zandbergen 2008)
within a CAD environment. To better geocode 911 calls, an accurate address point layer is
required.

27

3.4 Address Points
The address points will be updated as new construction and address change requests are made.
Unlike verifying a location by the street layer or address ranges, address points provide the
geocoding application (desktop or web-based) a reference to search for the exact location,
without estimating where the call is coming from. When the geocoding application fails to verify
a call location from the address point reference layer, the address ranges will assume the
responsibility to geocode the location. With address points, Dispatch can create common places
to search for locations without typing the full address. For example, if a dispatcher received a
distress call for medical aid at a McDonald’s on the corner of Imperial Highway and Laurel
Avenue in Brea, the dispatcher can type the business name (or one of the alias names) which can
be used to facilitate determining the location, which can save valuable time in sending help
(Couret 1999). Combined, both streets and address point layers work extremely well in CAD to
geocode calls-for-service locations.
As Brea continues to grow, more housing and services are needed. New streets, homes,
business locations, and change of businesses need to be regularly added or revised. CAD is a
“work-in-progress” system that requires continuous changes and updates. Therefore, using
ArcGIS’s ModelBuillder and georeferencing techniques can be used to build workflows that can
help streamline the process of updating address points and street ranges.
3.5 Address Points Creation
Since Brea did not have its own address point layer, this provided an opportunity for the City to
create one using the countywide parcel layer from the Orange County Assessor Office. Using
ArcMap and the geoprocessing tool, the ownership property information table was joined to the
parcel layer, to be later converted into an address point feature. For a simplified address point

28

table, the attribute information from the joined parcel layer was truncated, as illustrated in Figure
12. The revised table now includes site address information, such as house number, pre-
direction, street name, street type, street suffix, and unit number. See Table 2 for field
descriptions of the address point table attribute.

29

Figure 12. Workflow Process for Address Points Creation to Support Brea’s CAD System
Step
1
•Download Parcel and property information data from Orange County Assessor.
Step
2
•Add downloaded data to ArcMap for further analysis.
Step
3
•Join Parcel and property data using APN fields.
Step
4
•Export the joined parcel layer to a new feature class.
Step
5
•Add the joined Parcel layer (w/property ownership information) back to ArcMap.
Step
6
•Convert polygon parcel layer to address points.
Step
7
•Use the Feature to Point (Data Management) tool.
Step
8
•Add the new Address Point layer to ArcMap.
Step
9
•Add Brea city boundary layer to ArcMap.
Step
10
•Use the Selection Tool to select only the address points within the city boundary.
Step
11
•Export the selected address points (within city boundary) as a new layer.
Step
12
•Add new address point to ArcMap.
Step
13
•Identify address points without site address information for further research, including field work.
Step
14
•Add new fields to address points layer, based on approved specs per Intergraph.
Step
15
•ADDRESS, HOUSENUM, ADDR_PD, ADDR_SN, ADDR_ST, UNIT, SUFFIX, CITY, STATE, POSTAL, NAME, BREA_ID, LON,
LAT, ALIAS1, ALIAS2, ALIAS3, ALIAS4, ALIAS5 GROUP, TELEPHONE.
Step
16
•Use "Calculate" tool to populate attribute information for the above fields.
Step
17
•Remove the joined fields and duplicates.
Step
18
•Use a number sequencing script to populate BREAID field with unique values to each record.
Step
19
•If necessary, populate NAME, ALIAS(S), GROUP and TELEPHONE field with relevant information per dispatch's requests.
Step
20
•Use the "Calculate Geometry" tool to update LAT and LON fields with local projected coordinates
(NAD_1983_StatePlane_California_VI_FIPS_0406_Feet).
Step
21
•Save revised address point layer!

30

Table 2. Brea Address Point Attribute Table and Field Definitions
FIELD NAME DESCRIPTION
HOUSENUM Address number
ADDR_PD Prefix direction for the address
ADDR_SN Street name for the address
ADDR_ST Street type for the address
SUFFIX Street suffix direction
UNIT Unit or suite designator
CITY City in which the address is located
STATE State in which the address is located
POSTAL Zip Code in which the address is located
NAME Common place name (Starbucks)
ADDRESS
Combined address fields (HOUSENUM, ADDR_PD, ADDR_SN, ADDR_ST
and UNIT)
BREA_ID Unique identification number for the address
LON Longitude coordinate position for the address
LAT Latitude coordinate position for the address
ALIAS1 First alias name for the address
ALIAS2 Second alias name for the address
ALIAS3 Third alias name for the address
ALIAS4 Fourth alias name for the address
ALIAS5 Fifth alias name for the address
GROUP Group of category assigned to the address (Retail, Bank, Market)
TELEPHONE Contact information tied to each address

Next, the truncated parcel data is exported to a new feature class. This new parcel feature
class will be converted from a polygon to address points based feature class. The conversion of
the parcel polygon layer involves using the Feature to Point (Data Management) tool in ArcGIS.
In ArcMap, the new address point gets “clipped” to eliminate features outside of the City limits.
This process uses the City boundary as a reference layer, using the ArcMap Selection tool (or the
Clipped geoprocessing tool). The address point layer required specific fields and data formats.
New fields including house number, pre-direction, street name, street type, street suffix, unit
number, common places, aliases, telephone, and unique identification number (now known as

31

Brea_ID) were then added to the attribute table. The fields were populated using the “Calculate”
tool in ArcMap. Prior to adding the address points to CAD, three additional quality control steps
were required: (1) manual checking of the data table to confirm that each point has a fully
qualified address, unit information, common place names, and contact information, with no null
values; (2) each point must include latitude and longitude coordinates, based on the local defined
projection, State Place, NAD 1983, CA Zone 6, US Feet; and (3) each address point must have a
unique ID number in the “BREA_ID” field, this field cannot contain any duplicate values.
3.5 Common Places
The City’s CAD system is an ever evolving desktop mapping application requiring frequent
updates to the address point and street layers. These updates are important for Dispatchers to
better verify and geocode calls-for-service locations. The process for updating common places is
illustrated in Figure 13, namely the steps to update existing address points, common places, and
street ranges without georeferncing.

Figure 13. Create Common Places for CAD System
For example, updating an existing common place name from McDonald’s to Burger King
involved identifying the feature and updating its attribute information in ArcMap. Several checks
currently exist for updating address points and streets. These include adjusting street ranges,
updating contact information to existing address points, or removing features.
Step 1
•In ArcMap, bring in the Address Point and Street layers.
Step 2
•Start an editing session in Editor.
Step 3
•For address points, use the Search tool to identify the address point requesting update.
Step 4
•In "Editor" mode, update the attribute information, i.e., change McDonald's to Burger King in the NAME field.
Step 5
•To update street ranges, select the street name and adjust the four fields with address ranges.
Step 6
•Save edits...

32

CHAPTER 4: RESULTS
This chapter documents the comparisons of address points and street ranges in a computer aided
dispatch environment. The analysis results from the geocoding matched rates show that there are
advantages and disadvantages of using address points versus street layer for verifying or
geocoding calls-for-service.
4.1 Calls-for-Service
As of February 18, 2015, the City of Brea has a population of 42,393 (City of Brea 2015). The
Brea Police Department received 16,383 calls-for-service from January 01, 2014 to December
15, 2014 (City of Brea 2014). Calls-for-service are 911 calls that have been received by the Brea
Dispatch, including emergency calls from outside city limits, due to the City’s mutual aid
agreement with neighboring agencies. Using the “Select by Attributes” query tool in ArcMap
10.2.2 and visual inspection, it was determined that 15,268 out of 16,383 calls-for-service
originated in Brea. After further analysis and review, 9,671 calls-for-service have been identified
with a fully qualified address (e.g. 1717 E Birch St) for geocoding comparisons. These 9,671
records were then geocoded against multiple address locators, such as Esri’s US Address—
Single House and Dual Ranges locators, in order to compare the levels of accuracies of each
address locator.
4.2 Geocoding
A geocoding service, sometimes referred to as a geocoder, is a geoprocessing tool that provides a
better user experience within a map application, desktop, or web-based. A geocoding service is a
location, or address finder that can be used by multiple mapping technology companies,
including Esri. Google Maps and Microsoft’s Bing Maps would be irrelevant without a
dependable, accurate geocoding service to help users find what they are looking for. A

33

geocoding service uses customized algorithmic formulas, or a geocoding engine to display
address results from a set of reference feature datasets (Zanbergen 2008). Not every geocoding
service is the same, and each service requires a different data reference. For this research, two
address locators and two separate reference layers were used to geocode against a table of
addresses (Figure 14) to determine the strengths and weaknesses of each address locator. This
process was performed using Esri’s ArcGIS 10.2.2, ArcCatalog, ArcMap, and Microsoft Excel
2012.

Figure 14. Revised Address Table for Geocoding
4.3 Single House Locator
The Esri’s US Address—Single House (herein referred to as Single House) locator style creates
address locators for United States addresses (Esri 2014). This locator uses an address point
feature class with over 53,000 points maintained to support Brea’s CAD system. The City’s
address point layer represents a feature point in the reference data that corresponds to a single
fully qualified or verified address from the address table (see Figure 15). In this case, each
address that is being geocoded must be represented within the reference data layer; otherwise the
result is an unmatched address. This style of locator does not accept a range of addresses to

34

verify; rather it provides an exact one-to-one match criteria. Therefore, the Single House address
locator style requires that each feature in the reference data correspond to a single address value,
such as address points.

Figure 15. Single House Address Locator Setup for Geocoding Address Points
Using the above parameters, the Single House locator, with address points as the reference layer,
successfully geocoded 8,155 locations from the revised table of 9,671 records; producing a
matched rate of 84 percent, as shown in Figure 16.

Figure 16. Geocoding Results from Single House Locator

35

Whereas Figure 17 shows that 8,148 addresses geocoded within the city limits, with seven
locations verified outside Brea’s city boundary. Results from Figure 17 show that unmatched
addresses derived from the Single House locator could be mitigated in the future by adding the
unmatched addresses to the address point reference data to improve the match rate.

Figure 17. Geocoded Results (8,148 out of 9,671) from Single House Address Locator
4.4 Dual Ranges Locator
The Esri’s US Address—Dual Ranges (will be referred to as Dual Ranges) locator style is
commonly used to geocode addresses. Benefits of a dual ranges address locator include more
flexibility in verifying addresses, compared to a single house locator. For instance, in a dual
ranges locator, addresses are geocoded and verified as long as they fall within the assigned range

36

of values from the left and right (odds/evens) sides of a given street segment (Esri 2014), shown
in the example in Figure 18.

Figure 18. Example of the reference data used for a dual ranges address locator (Esri 2014)
Also, the dual ranges address locator can designate the side of the street segment where the
address is located, resulting in less ambiguity for first responders. For Brea, the dual ranges
address locator uses a polyline based street segment feature class, with values for left and right
sides for each segment, or feature. For instance, each feature in the reference data for the dual
ranges locator denotes a street segment with two ranges of addresses that fall along that street
segment, one for each side of the street (Esri 2014).
In this analysis a street/centerline feature class is assigned as a reference layer for the
Dual Ranges address locator style. Each street segment has a beginning and an end address
number range for each side of the street, as well as street name information. In addition, the
reference layer also includes fields that contain the street's prefix direction, prefix type, street
type, suffix direction, ZIP Code, and municipal code information for each side of the street, as
shown in Figure 19. Further, each street segment also supports normal block ranges,
alphanumeric addresses, and contains cross-street information for searching by intersections
(Esri 2014).

37

Figure 19. Street Reference Layer for Dual Ranges Locator Table in ArcMap
The Dual Ranges locator also was assigned similar parameters to the Single House
locator for comparison purposes. Spelling sensitivity, minimum candidate score, and minimum
match score are designated with values similar to the Single House Locator, shown in Figure 20.
Using the Dual Ranges address locator and the street reference layer, resulted in a matched rate
of 96 percent. A total of 9,250 out of 9,671 addresses geocoded successfully, as shown in Figure
21. In addition, all of the geocoded locations are spatially verified by using ArcMap to analyze
that all the geocoded points are within the Brea City limits, depicted in Figure 22.

38

Figure 20. Dual Ranges Address Locator and Street/Centerline Feature Class

Figure 21. Geocoded Results from Dual Ranges Address Locator

39

Figure 22. Geocoded Results (9,250 out of 9,671) from Dual Ranges Address Locator
4.5 Accuracy Comparisons
The geocoded results produced by the Single House and Dual Ranges locators show that there
are many options to verify calls-for-service locations. From the geocoded results, it seems that
the Dual Ranges address locator style, with the street/centerline reference data, performed much
better than the address point based Single House locator. In most instances, the criterion assigned
to the Dual Ranges locator results in a higher match rate than the Single House locator. For
example, in the reference dataset for the Dual Ranges locator, the street segment for Laurel
Avenue has a range from 200 to 299; therefore, any addresses geocoded in that given range will
result in a match. On the other side, the Single House locator only produces a geocoded address

40

if there is an exact match to the reference data (i.e. 216 Laurel Avenue). In most cases, the
flexibility of the Dual Ranges locator produces a higher geocoding matched rate than the Single
House locator. However, in some instances, the geocoded addresses from the Dual Ranges
locator are 40 to 80 (or more) feet off from the actual locations, as shown in Figure 23.
Therefore, in the case of Brea, geocoding by address points is the preferred method because
Dispatch demands that each location be verified to exactly match the call location, resulting in
faster response time to the 911 call.

Figure 23. Comparisons of Geocoded Results
As an example of performance, from Figure 23, 204 Laurel Ave and 216 Laurel Ave are
44 to 84 feet off, when compared to the geocoded results from the Single House locator. The

41

difference in Figure 23 might not seem catastrophic, but when medical aid is urgently needed,
every second counts. Moreover, as shown in Figure 24, the Dual Ranges address locator
geocoded 131 Kraemer Blvd to a location that is over 700 feet away from the actual property.
Overall, the geocoded results shown in Figure 24 are off between 300 to 700 feet from the
physical locations. In contrast, the address points based Single House address locator geocoded
the location to the exact building.

Figure 24. Analysis of Geocoded Results
Ideally, Brea Dispatch requires that all calls-for-service geocode to the exact location. The more
time first responders take to arrive at the scene of the incident, the less they have to attend and
help. In Brea, precision is at the utmost importance when it comes to responding to calls-for-

42

service. Therefore, in Brea, geocoding by address points is the preferred method because
Dispatch requires that all calls-for-service location verified exactly to the call locations, resulting
in faster response time and enhanced services provided to the community.

43

CHAPTER 5: DISCUSSION AND CONCLUSIONS
The purpose of this thesis is to present an objective analysis for comparing how different
reference datasets and address locators perform, to determine the geocoder that works best within
a given organization. The geocoding system in Brea City CAD is closely integrated into other
core operational and workflow systems, and tailored to the type of data it encounters. This has
produced results of sufficient quality for a range of users, including public safety, Police
Dispatchers, and first responders (Goldberg et al 2013). For instance, the ongoing maintenance
of a map-based CAD system is critically important for public safety, since it can provide
accurate and on-demand information to first responders. The process of updating an existing
CAD system with GIS layers, such as address points and street ranges are laborious, expensive,
and time intensive (as demonstrated in the workflow model). Nevertheless this effort is deemed
extremely worthwhile since the outcome potentially saves lives.
In addition, this research aims to promote the interagency cooperation and data sharing
within the GIS community who supports CAD systems for local government agencies
throughout the United States. There are many benefits for an interagency CAD system, such as
shared deployment costs, better management of GIS datasets, consistency, and more knowledge
sharing opportunity among the GIS community. Unfortunately, at this time, similar to other
agencies in Southern California, Brea’s proprietary CAD system is an obstacle for sharing
resources with neighboring agencies.
Addresses help individuals to locate one another in this challenging world (Zanbergen
2008). Geocoding is a topic that has been greatly studied and examined in many fields, including
local government and public safety agencies. The process of geocoding provided an extended
understanding on its many uses, for example the investigation of Dengue fever in Africa

44

(Goldberg 2011). For CAD, one of the most critical components for Dispatchers is to accurately
and quickly verify the location of a distress call to provide timely response. For Dispatch, this
involves geocoding each call to the exact location of the caller. Therefore, Dispatch prefers an
exact match to reduce confusion for first responders and to minimize potential errors, such as
driving to a wrong location.
The transition to a GIS centric CAD system proved to be full of challenges, with
extensive data conversion issues, learning new software, and unfamiliar graphical user interfaces.
Intergraph’s computer aided dispatch system uses customized graphical user interfaces and tools,
which have a different look and feel compared to Brea’s previous map-less CAD system.
However, in the end, these challenges provided a great learning opportunity for the Dispatchers
to become more comfortable using the system.
In the City of Brea, Dispatch received over 16 thousands calls-for-service in 2014 (Brea
Police Department). This thesis examined and tested the accuracies of address locators used to
geocode, or verify 911 calls. Verifying calls-for-service to its exact location is critically
important, resulting in faster response time. Therefore, Dispatch demands that its CAD system
geocode each call to the physical location. This is only possible using an address point data
reference with the geocoder. On the other hand, while the street range data reference helped to
locate many 911 calls, in some cases it resulted in greater ambiguities than address points. When
a location fails to verify the exact physical location, it required more time and effort by first
responders to offer assistance.
A total of 9,671 addresses were used to compare the different levels of accuracy and
matched rates from multiple address locators. The geocoding results show that when addresses
are geocoded using the street range address locator, the matched rate is 96 percent, compared to

45

84 percent matched rate in the address point locator. However, a closer examination of the
geocoded features from the street range locator showed some discrepancies. Some of the features
are off by 30 to 700 feet from the actual calls-for-service locations. On the other hand, when
using the address point locator, each geocoded feature matches correctly to the physical location,
without any ambiguities. Validating 911 calls correctly to serve the community is an important
and integral part within public safety. The ability to dispatch calls efficiently and timely could
result in a greater chance to saves lives in life-threatening situations, where faster response time
is most critical. Therefore, Brea’s Dispatch preferred methodology for addressing verification is
through address points.
The methodologies and discussions described in this thesis show that an accurate, well-
tested address point locator is a favored solution to street range address locator for geocoding
911 calls. An address point based locator, maintained by city staff on a consistent basis, should
also improve the overall match rate. A street range address locator is useful in a situation where
the address point locator failed to geocode a location. Furthermore, matched rates from the
geocoding results do not always verify to the physical call location. When it comes to geocoding
addresses, or the need to confirm a location for billing purposes, maintaining an accurate,
consistent address point layer is not only useful for Dispatch, but citywide. In addition, this work
could represent a model for future interagency cooperation among neighboring cities and
counties who may want to deploy a sustainable GIS-based CAD system, with shared costs and
resources.
5.1 Next Steps
Updating spatial data for a mission-critical system like CAD is a daunting task within the GIS
community. Results of the research show the superiority of address points when compared to

46

street address ranges in a CAD environment for Brea’s Dispatch, in particular when it comes to
geocoding accurate, precise calls-for-service locations. However, future work in this subject
could include adding a third address locator. For example, a composite address locator, which
includes multiple address locators for geocoding locations may be the preferred tool, since it
could include both address points and street ranges, resulting in more matched locations.
The next step in this research is to examine how response time is calculated because
some 911 calls are more urgent than others (Table 3). Furthermore, Dispatch determines the
severity of each call, based on the information received from the caller. For example, a Code 3
caller with a life-threatening emergency will receive faster response time than a Priority 3 caller
who is requesting assistance due to a minor fender-bender.
Table 3. Monthly Response Times from January to November, 2014 (Provided by Brea
Police 2014)
MONTHLY
AVERAGE
CODE 3
(MINUTES)
PRIORITY 1
(MINUTES)
PRIORITY 2
(MINUTES)
PRIORITY 3
(MINUTES)
JANUARY 2.59 6.12 7.21 11.31
FEBRUARY 3.31 5.27 7.28 11.23
MARCH 3.37 6.00 8.45 12.50
APRIL 3.25 3.44 7.48 11.00
MAY 3.48 5.48 7.42 11.14
JUNE 3.31 5.26 7.03 11.25
JULY 3.35 5.28 7.33 12.58
AUGUST 4.11 6.06 7.04 12.02
SEPTEMBER 3.39 5.46 7.30 12.10
OCTOBER 3.29 6.54 7.28 11.55
NOVEMBER 3.30 3.43 7.39 10.56

47

In addition, measuring response time is also a challenging task (sometimes impossible). Each
response varies based on multiple environmental and human influences: weather, time of day,
traffic conditions and geocoding results.
Moving forward, the data, examples, and methodologies used may provide some
potential usefulness for other government agencies who are considering migrating to a GIS based
CAD system. Though it may seem simple to assign address points as the main reference data
source for geocoding addresses, agencies with minimal funding sources and GIS staff may not
have the capabilities to maintain such a labor intensive dataset. Rather, it might make more
economic sense to use a street range data layer to geocode calls-for-service. Overall, each agency
needs to determine its preferred levels of accuracy, when validating calls-for-service locations.

48

REFERENCES
Babinski, Greg. 2009. GIS as an Enterprise Municipal System. Government Finance Review
25(1): 22-30.
Bolstad, Paul V., Gessler, Paul and Lillesand, Thomas M. (1990) 'Positional uncertainty in
manually digitized map data', International Journal of Geographical Information Science,
4: 4, 399 — 412.
Brea Olinda Unified School District. District Profile. 2015. Accessed April 28, 2015.
http://www.bousd.k12.ca.us/cms/page_view?d=x&piid=&vpid=1264862003493.
Brosowsky, John and Nathan Ekdahl. 2012. Why You Should Invest in an Address Point Layer:
Identifying 9-1-1 Caller Location. A GeoComm White Paper. Accessed March 18, 2015.
http://www.geo-comm.com/wp-content/uploads/2012/11/Address-Point-Layer-
Development-GeoComm.pdf.
Chatterton, M. (1987). Assessing police effectiveness: Future prospects. British Journal of
Criminology, 27, 80-86.
City of Brea. 2015. Planning Division. Demographics. Accessed April 28, 2015.
http://www.ci.brea.ca.us/index.aspx?NID=492.
City of Brea. 2014, Brea Police Department. Accessed December 12, 2014.
http://www.ci.brea.ca.us/index.aspx?NID=336.
Couret, C. (1999). Police and technology: The silent partnership. American City and County,
114(9), 31-32.
Davis Jr., Clodoveu A. and Frederico T. Fonseca. 2007. Assessing the Certainty of Locations
Produced by an Address Geocoding System. Geoinformatica 11: 103–129.

49

Dvorak, M. (1997). Counties, cities cooperate on public safety. American City and County,
112(8), 8.
Enders, Alexandra; Brandt, Zachary (2007). Using Geographic Information System Technology
to Improve Emergency Management and Disaster Response for People with Disabilities.
Journal of Disability Policy Studies 17, 223-229.
Esri. 2015. Geocode Services. Accessed April 7, 2015.
http://resources.esri.com/help/9.3/arcgisserver/apis/soap/SOAP_Geocode_Overview.htm.
Esri. 2014. The geocoding process. Accessed April 9, 2015.
http://resources.arcgis.com/en/help/main/10.2/index.html#//00250000000r000000.
Esri. 2014. ArcGIS Help 10.2, 10.2.1, and 10.2.2. Accessed February 24, 2015.
http://resources.arcgis.com/en/help/main/10.2/index.html#//00250000000v000000.
Esri. 2013. ArcGIS Help 10.1. Accessed April 29, 2015.
http://resources.arcgis.com/en/help/main/10.1/index.html#/Essential_geocoding_vocabul
ary/002500000004000000/.
Esri. 2013. ArcGIS Help 10.2, 10.2.1, and 10.2.2. Accessed April 29, 2015.
http://resources.arcgis.com/en/help/main/10.2/index.html#//01m500000003000000
Freeway Complex Fire. 2013. Freeway Complex Fire. Accessed May 1, 2015.
http://en.wikipedia.org/wiki/Freeway_Complex_Fire.
Geographic Information System Siting of Emergency Vehicles Improves Response Time.
(2012). US Fed News Service, Including US State News Retrieved. Accessed October 6,
2014. http://search.proquest.com/docview/925963521?accountid=14749.
Goldberg, Daniel, Morven Ballard, James H Boyd, Narelle Mullan, Carol Garfield, Diana
Rosman, Anna M Ferrante and James B Semmens. 2013. An evaluation framework for

50

comparing geocoding systems. Accessed April 7, 2015. http://www.ij-
healthgeographics.com/content/12/1/50.
Goldberg, Daniel. 2011. Advances in Geocoding Research and Practice. Transactions in GIS
15(6): 727-733.
Goldberg, Daniel A. and Myles G. Cockburn. 2010. Improving Geocoding Accuracy with
Candidate Selection Criteria. Transactions in GIS, 2010, 14(s1): 149–176.
Harada, Yutaka and Takahiro Shimada. 2006. Examining the impact of the precision of address
geocoding on estimated density of crime locations. Computers & Geosciences: 32:1096–
1107.
Harries, Keith. 1999. Mapping Crime: Principle and Practice. Accessed April 8, 2015.
https://www.ncjrs.gov/html/nij/mapping/ch4_3.html.
Hoef, Marcel van de. and Joram Kanner. 2007. TomTom Agrees to Acquire Tele Atlas for EU2
Billion (Update10). Accessed April 30, 2015.
http://www.bloomberg.com/apps/news?pid=newsarchive&sid=agT1Po33faG4&refer=ho
me.
Intergraph. 2015. Accessed April 28, 2015. http://www.intergraph.com/about_us/default.aspx
Kellison, Michael T. 2012. Address Points and a Master Address File: Improving Efficiency in
the City of Chino. Master of Science thesis, University of Southern California, CA, USA.
Klosterman, Richard, and Alan Lew. 1992. TIGER products for planning. Journal of the
American Planning Association 58 (3) (Summer): 379-385.
Mehrotra, Sharad, X. Qiu, Z. Cao and A. Tate. 2013. Technological Challenges in Emergency
Response. IEEE Computer Society July/August: 5-8.

51

Nesbary, D. (2001). The Acquisition of Computer-Aided Dispatch Systems. Social Science
Computer Review, 19, 3, 348-356.
Newcombe, T. (1995). The 911 revolution. Advancing technology is making for vastly more
capable emergency-responsesystems. Can there be too much emphasis on rapid response?
The Business of Government, 7, 56-59.
O’Reagan, R. Thomas and Alan Saalfeld. (1987). Geocoding Theory and Practice at the Bureau
of the Census. Accessed April 8, 2015. https://www.census.gov/srd/papers/pdf/rr87-
29.pdf.
Putnam, R.D. (1995). Bowling alone: America’s declining social capital. Journal of Democracy,
6(1), 65-78.
Schootman, Mario, David A. Sterling, James Struthers, Yan Yan, Ted Laboube, Brett Emo and
Gary Higgs. 2007. Positional Accuracy and Geographic Bias of Four Methods of
Geocoding in Epidemiologic Research. Elsevier Inc.: 17:464–470.
Tele Atlas. 2007. Dynamap/Transportation. Version 9.1. Retrieved from City of Brea. GIS
Department. 1 Civic Center Circle. Brea, CA 92821.
TomTom. 2015. MultiNet. Accessed April 30, 2015.
http://www.tomtom.com/en_gb/licensing/products/maps/multinet/index.jsp?WT.Click_Li
nk=home_quick_link.
U.S. Census. 1997. Accessed April 28, 2015.
https://www.ncjrs.gov/html/nij/mapping/ch4_3.html
Yang, Duck-Hye, Lucy Mackey Bilaver, Oscar Hayes, and Robert Goerge. 2004. Improving
Geocoding Practices: Evaluation of Geocoding Tools. Journal of Medical Systems 28(4)
(August): 361-370.

52

Yildirim, Volkan, Tahsin Yomralioglu, Recep Nisanci and Halilibrahim Inan. 2013. Turkish
street addressing system and geocoding challenges. Proceedings of the ICE - Municipal
Engineer, 167(2) (December): 99 –107.
Zandbergen, Paul A. 2008. A comparison of address point, parcel and street geocoding
techniques. Computers, Environment and Urban Systems 32: 214-232.
Zhan, Benjamin F., Jean D. Brender, Ionara D. Lima, Lucina Suarez and Peter H. Langlois.
2006. Match Rate and Positional Accuracy of Two Geocoding Methods for
Epidemiologic Research. Elsevier Inc.: 16:842–849.

53

APPENDIX A: Brea GIS Layers for CAD
LAYER NAME FORMAT ATTRIBUTES SOURCES
Address Points Point Feature Class ADDRESS, HOUSENUM, ADDR_PD,
ADDR_SN, ADDR_ST, UNIT, SUFFIX,
CITY, STATE, POSTAL, NAME,
BREA_ID, LAT, LON, ALIAS1, ALIAS2,
ALIAS3, ALIAS4, ALIAS5, GROUP,
TELEPHONE
City of Brea
Streets Line Feature Class L_F_ADD, L_T_ADD, R_F_ADD,
R_T_ADD, PREFIX, NAME, TYPE,
SUFFIX, FCC , POSTAL_L, POSTAL_R,
SEG_LEN, SPEED, ONE_WAY,
F_ZLEV, T_ZLEV, CITY_L, CITY_R,
BREA_ID, FULL_NAME,
ALIAS1_NAM, ALIAS2_NAM,
ALIAS3_NAM, ALIAS1_PRE,
ALIAS2_PRE, ALIAS3_PRE,
ALIAS1_TYP, ALIAS2_TYP,
ALIAS3_TYP, ALIAS1_SUF,
ALIAS2_SUF, ALIAS1_SUF3
City of Brea
Parcels Polygon Feature Class APN, SITEADDR, HOUSENUMBR,
OWNER_NAME
City of Brea
Police ESZs Polygon Feature Class ESZ, VALUE City of Brea
Police Beats Polygon Feature Class BEAT, NAME City of Brea
Fire ESZs Polygon Feature Class ESZ, VALUE City of Brea
City Boundary Polygon Feature Class AREA, PERIMETER, CITY_NAME City of Brea
Surrounding
Cities
Polygon Feature Class AREA, PERIMETER, CITY_NAME City of Brea
Police Stations Point Feature Class Name, ADDRESS, Zipcode, City,
Agency
City of Brea
Fire Stations Point Feature Class Name, ADDRESS, Zipcode, City,
Agency
City of Brea
POI Point Feature Class NAME, ADDRESS City of Brea

54

LAYER NAME FORMAT ATTRIBUTES SOURCES
Parks Polygon Feature Class NAME, ADDRESS City of Brea
Schools Polygon Feature Class NAME, ADDRESS City of Brea
3” Imagery Raster Dataset Cell Size (X,Y), 3” Resolution City of Brea
APPENDIX B: Software for Building Brea CAD System
SOFTWARE SKILLS EXPERIENCE
Esri ArcGIS (ArcMap) Data editing and georeferencing 16 years
Esri ArcGIS (ArcCatalog) Geodatabase design, feature class
management, Geocoding and
creating Address Locator
16 years
Intergraph GeoMedia Pro
w/iMapEditor
CAD data conversion 7 years
Intergraph iDispatcher Test converted CAD data after map
roll
7 years
APPENDIX C: Brea Street Type Field Definition
STREET TYPE NAME GIS
AVENUE AVE
BOULEVARD BLVD
CIRCLE CIR
COURT CT
DRIVE DR
LANE LN
LOOP LP
PARKWAY PKWY
PLACE PL
ROAD RD
STREET TYPE NAME GIS
STREET ST
TERRACE TER
TRAIL TRL
WAY WY

55

APPENDIX D: Study Area: City of Brea Boundary

Projection: Projected Local State Plane: CA Zone 6 (US Feet)

56

APPENDIX E: City of Brea’s Computer Aided Dispatch

Source: City of Brea. 2014. Intergraph’s CAD Mapping System.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Address points and a maser address file: improving efficiency in the city of Chino

PDF

A fire insurance map geocoder for pre-earthquake San Francisco

PDF

Finding food deserts: a study of food access measures in the Phoenix-Mesa urban area

PDF

Development of a Web GIS for urban sustainability indicators of Oakland, California

PDF

Peoples of Washington historical geographic information system: geocoding culture using archival standards

PDF

Closed landfills to solar energy power plants: estimating the solar potential of closed landfills in California

PDF

Use of GIS for analysis of community health worker patient registries from Chongwe district, a rural low-resource setting, in Lusaka Province, Zambia

PDF

Guiding business oriented volunteered geographic information through geotigger services: a case study of CrossFit affiliates

PDF

Visualizing historic space through the integration of geographic information science in secondary school curriculums: a comparison of static versus dynamic methods

PDF

A comparison of urban land cover change: a study of Pasadena and Inglewood, California, 1992‐2011

PDF

Geospatial analysis of unintended casualties during combat training: Fort Drum, New York

PDF

Spatial analysis of urban built environments and vehicle transit behavior

PDF

The role of GIS in asset management: County of Kauai Department of Parks and Recreation a need for an asset management program

PDF

Integration of topographic and bathymetric digital elevation model using ArcGIS interpolation methods: a case study of the Klamath River Estuary

PDF

Identification and analysis of future land-use conflict in Mecklenburg County, North Carolina

PDF

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

PDF

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

PDF

Surface representations of rainfall at small extents: a study of rainfall mapping based on volunteered geographic information in Kona, Hawaii

PDF

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

PDF

Distribution and correlates of feral cat trapping permits in Los Angeles, California

Asset Metadata

Creator Dao, Jimmy Tuan (author)

Core Title A comparison of address point and street geocoding techniques in a computer aided dispatch environment

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 06/18/2015

Defense Date 04/27/2015

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

Tag address points,computer aided dispatch,geocoding,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Ruddell, Darren M. (committee chair), Swift, Jennifer N. (committee member), Warshawsky, Daniel N. (committee member)

Creator Email daoucla@gmail.com,jimmytda@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-571741

Unique identifier UC11301282

Identifier etd-DaoJimmyTu-3480.pdf (filename),usctheses-c3-571741 (legacy record id)

Legacy Identifier etd-DaoJimmyTu-3480.pdf

Dmrecord 571741

Document Type Thesis

Format application/pdf (imt)

Rights Dao, Jimmy Tuan

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