This paper presents the methodology by which Global Positioning System and Geographic Information System technologies are being employed in the Texas counties of Frio and Karnes for the construction of an Enhanced 911 database.
GPS as the Foundation of an Enhanced Emergency 911 System
1. Juan Tobar
GIS Coordinator
Alamo Area Council of Governments
118 Broadway Suite 400
San Antonio, TX 78205
David Kruse, AICP
Regional Data Center Manager
Alamo Area Council of Governments
118 Broadway Suite 400
San Antonio, TX 78205
GPS as the Foundation of an Enhanced Emergency 911 System
Abstract: This paper presents the methodology by which Global Positioning System
(GPS) and Geographic Information System (GIS) technologies are being employed in the
Texas counties of Frio and Karnes for construction of an Enhanced 911 (E911) database.
Although not as spatially accurate as traditional aerial orthophotography, GPS provides a
viable and cost effective way to create an E911 database. In particular, GPS provides the
most cost effective way to acquire current road network elements while providing a
vehicle for database maintenance. In addition, GPS collected data in conjunction with
aerial photography has permitted the correction of private road network elements from
TIGER at significant savings as compared with traditional photogrammetric methods.
INTRODUCTION
Frio and Karnes County, Texas are part of a region wide effort to create an Enhanced 911
(E911) system using Geographic Information System (GIS) and Global Positioning
System (GPS) technologies. Located southwest and southeast of San Antonio
respectively, these two largely rural counties joined five other counties including
Atascosa, Bandera, Gillespie, Kendall, and Wilson as members of the E911 Regional
Plan administered by the Alamo Area Council of Governments (AACOG) . With a
combined population of 30,000 residents, Frio and Karnes counties experienced net
population losses though outmigration at the rates of 2.27% and 8.37% respectively,
between 1980 and 1990 (U.S. Bureau of the Census, 1990). With a declining tax base
and limited funds for the required local match of 25%, it was not surprising that by the
end of 1994 very little work had been done on E911. Through a partnership between
AACOG and Frio and Karnes counties, a state-of-the-art Enhanced 911 system is being
built in these counties.
2. OVERVIEW OF 911 SYSTEMS
There are three levels of nationally recognized 911 services which enable 3-digit dialing
into an emergency telecommunications networks. The first level of service is Basic and
it uses wire center defaults to enable callers to use the 3-digit number to reach a
centralized Public Safety Answering Point (PSAP). The second level of service involves
Automatic Number Identification (ANI) which is the automatic display of the telephone
number of the calling party at a PSAP. In addition, this level of service allows the routing
of calls based on calling number prefix. The downside to this level of service is that it
requires the caller to stay on the line to help the responding vehicle locate him or her if
needed. For this reason, the third level of service, Enhanced 911 (E911), is actively being
pursued in rural areas. This level of service is most common in urban areas and involves
Automatic Location Identification (ALI) which is the automatic display of the physical
street address (not a P.O. Box) associated with the telephone number (ANI) which is
displayed on a screen at a PSAP. In addition, E911 systems also route calls based on
predetermined Emergency Service Zones (ESZ). These features, when properly
implemented, provide a unique geographic location for each telephone call. This specific
location identification permits persons needing emergency services to be found even if
they are unable to respond further after dialing 911.
GIS AND E911 DATA MANAGEMENT
ESRI’s ARC/INFO GIS software was chosen for the Enhanced 911 databases being
created for Frio and Karnes Counties for the following reasons: first, E911 is geographic
by nature; second, a GIS allows the customization of its interface to be tailored for
Enhanced 911 data entry and feature manipulation; third, it permits the automation of
time consuming tasks; and fourth, with the increase in calls to PSAPs from wireless
communication devices GIS offers the best method to deal with this emerging technology
(Ozanich, 1996).
The geographic nature of an Enhanced 911 system means it is ideal for implementation
on a GIS platform. Enhanced 911 in a GIS environment requires at least three databases:
a Master Emergency Service Zone (MESZ), a Master Street Address Guide (MSAG), and
a Master Point Address Guide (MPAG) database. The MESZ database contains polygons
which represent Emergency Service Zones (ESZ) and data items which quantify and
qualify these polygons. Some of these items include: Emergency Service Number (ESN),
PSAP, law, fire, EMS, and Community (Table 1). The MSAG database contains arc
segments which represent roads and data items which quantify and qualify these roads.
Some of these items include: left and right from and to addresses, street name, and street
suffix (Table 2). Lastly, the MPAG database contains points which represent structures
and data items quantifying and qualifying these structures. Some of these items include:
address range, street name, phone number, and name of resident (Table 3).
Table 1: Master Emergency Service Zone (MESZ) Database
COL. ITEM NAME WIDTH OUTPUT TYPE N.DEC DESCRIPTION
3. 1 AREA 4 12 F 3 ARA OF POLYGON
5 PERIMETER 4 12 F 3 PERIMETER OF POLYGON
9 TB_ESN# 4 5 B - INTERNAL ID FOR ESZ POLYGON
13 TB_ESN-ID 4 5 B - UNIQUE ID FOR ESZ POLYGON
17 ESN 3 3 I - NUMBER OF ESZ
20 PSAP 4 4 C - NMAE OF PSAP
24 LAW 28 28 C - NAME OF LAW ENFORCEMENT
52 FIRE 20 20 C - NAME OF FIRE DEPT.
72 EMS 15 15 C - NAME OF EMS
87 COMM 32 32 C - NAME OF COMMUNITY
Table 2: Master Street Address Guide (MSAG) Database
COL ITEM NAME WDT OPUT TYPE N.DEC DISCRIPTION
H
1 FNODE# 4 5 B - FROM NODE ID
5 TNODE# 4 5 B - TO NODE ID
9 LPOLY# 4 5 B - LEFT POLYGON ID
13 RPOLY# 4 5 B - RIGHT POLYGON ID
17 LENGTH 4 18 F 3 LENGTH
25 COVER# 4 5 B - UNIQUE ARC ID
29 COVER-ID 4 5 B - UNIQUE ARC ID
33 L-ADD.FROM 7 7 I - LOWEST HOUSE NUMBER IN LEFT ADDRESS
RANGE
40 L-ADD.TO 7 7 I - HIGHEST HOUSE NUMBER IN LEFT ADDRESS
RANGE
47 R-ADD.FROM 7 7 I - LOWEST HOUSE NUMBER IN RIGHT ADDRESS
RANGE
54 R-ADD.TO 7 7 I - HIGHEST HOUSE NUMBER IN RIGHT ADDRESS
RANGE
61 PRFXDIR 2 2 C - STREET DIRECTION PREFIX;
N,S,E,W,NE,NW,SE,SW
63 STNAME 40 4 C - STREET NAME
103 STSUFFIX 4 4 C - STREET TYPE - AVENUE, ROAD POSTAL
ABBREVIATIONS
107 SUFDIR 2 2 C - STREET DIRECTION SUFFIX;
N,S,E,W.NE,NW,SE,SW
109 COMM 32 32 C - COMMUNITY NAME
141 PARITY 1 1 C - INDICATES IF RANGE IS ODD, EVEN OR BOTH
142 ESN 6 6 C - EMERGENCY SERVICE NUMBER TO BE ASSIGNED
TO THE STREET DEFINED WITHIN THE HOUSE
NUMBER RANGE SPECIFIED
148 PSAP 4 4 C - PUBLIC SAFETY ANSWERING POINT
152 EXCHANGE 3 3 C - TELEPHONE COMPANY EXCHANGE CODE
155 TAR 4 4 C - TELEPHONE COMPANY TAR CODE
159 DIRECT 50 50 C - DIRECTIONS ON HOW TO ARRIVE AT THE
BEGINNING OF THE ROAD
209 AKA 40 40 C - ALTERNATIVE ROAD NAME
249 TYPE 5 5 C - ROAD TYPES: R1, R2, R3, B, H1, H2, C1
254 SOURCE 5 5 C - GPS, TIGER, AERIAL PHOTOGRAPHY
259 MISC 50 50 C - RESERVED FOR FUTURE USE
4. Table 3: Master Point Address Guide (MPAG) Database
COL ITEM NAME WDTH OPUT TYP N.DEC DESCRIPTION
1 AREA 4 12 F 3 POLYGON AREA, EMPTY FOR POINTS
5 PERIMETER 4 12 F 3 POLYGON PERIMETER, EMPTY FOR POINTS
9 COVER 4 5 B - UNIQUE POINT ID
13 COVER-ID 4 5 B - UNIQUE POINT ID
17 RANGE 7 7 I - LOWEST HOUSE NUMBER IN SPECIFIC ADDRESS
RANGE
24 PRFXDIR 2 2 C - STREET DIRECTION PREFIX; N,S,E,W,NE,NW,SE,SW
26 STNAME 40 40 C - STREET NAME
66 STSUFFIX 4 4 C - STREET TYPE - AVENUE, ROAD - POSTAL
ABBREVIATION
70 SUFXDIR 2 2 C - STREET DIRECTION SUFFIX; N,S,E,W,NE,NW,SE,SW
72 UNIT 4 4 C - UNIT NUMBER OR APARTMENT NUMBER
76 COMM 32 32 C - COMMUNITY NAME
108 ZIP 5 5 C - ZIPCODE
113 ZIP4 4 4 C - ZIPCODE4
117 AREACODE 3 3 C - AREA CODE
120 EXCHANGE 3 3 C - EXCHANGE CODE
123 LINE 4 4 C - LINE CODE
127 PHONE 8 8 C - COMBINATION OF EXCHANGE-LINE
135 NAME 40 40 C - RESIDENTS NAME
175 ESN 6 6 C - EMERGENCY SERVICE NUMBER
181 MEDICAL 50 50 C - MEDICAL INFORMATION SPECIFIC TO ADDRESS
231 MADRS 59 59 C - MAILING ADDRESS
290 MCOMM 32 32 C - MAILING COMMUNITY
322 MZIP 5 5 C - MAILING ZIPCODE
327 OADRS 59 59 C - OLD ADDRESS
386 OCOMM 32 32 C - OLD MAILING ADDRESS
418 OZIP 5 5 C - OLD ZIPCODE
423 OMADRS 59 59 C - OLD MAILING ADDRESS
482 OCOMM 32 32 C - OLD MAILING COMMUNITY
514 OMZIP 5 5 C - OLD MAILING ZIPCODE
519 PARITY 1 1 C - INDICATES IF RANGE IS ODD, EVEN, OR BOTH
520 MISC 50 50 C - RESERVED FOR FUTURE USE
GPS AND E911 DATA MANAGMENT
GPS was chosen as the primary data acquisition technology for the Enhanced 911
databases being created for Frio and Karnes Counties for the following reasons. First, it
permits the construction of a reasonably accurate and current road network base map at a
reasonable cost. Second, it provides a simple and easy way to maintain both road and
structure databases. Third, it is a technology which may be needed to deal with the
increase of calls from wireless communication devices to PSAPs (Lucy, 1995). Lastly,
GPS is very cost and time effective when compared with other methods of data
acquisition.
DIFFERENTIAL GPS
Two types of systems exist for the transmission of differential corrections to GPS users:
traditional Differential GPS (DGPS) and Wide Area Differential GPS (WADGPS). In
evaluating these two methods we found that WADGPS provided a better solution to our
needs.
A traditional Differential GPS (DGPS) system maintains a reference receiver which
estimates pseudo-range measurement errors caused by variations in the satellite clocks,
orbital parameters, ionospheric delays, and atmospheric delays. GPS users in the vicinity
5. then pick up the transmitted estimated errors in the form of differential corrections which
are then used on their pseudo-range measurements in real time to produce one to five
meter accuracy in most situations.
Unfortunately, a number of factors may degrade the reliability of the differential
corrections received from DGPS. These factors include: the availability of the reference
station, multipath errors caused by the station’s surrounding environment, as well as the
satellite tracking capability of the station. In addition, these corrections are usually only
valid for local users, usually within a radius of about 200 km around the reference site
(Abousalem, et al, 1995).
Accuracy degrades as the distance between base station and rover increases. An estimate
of this degradation is 10 parts per million (ppm)/ kilometer (Aspen, 1995). This is
because errors estimated at the reference site become uncorrelated with those errors
experienced at the user’s location because of the spatial decorrelation between the error
sources.
WIDE AREA DIFFERENTIAL GPS
The Wide Area Differential GPS (WADGPS) solution chosen for this project was
ACCQPOINT, a North American WADGPS system developed to supply real time DGPS
corrections by Accqpoint Communications Corporation. This system relies on a limited
number of DGPS reference sites widely distributed across the country. Each site is
equipped with GPS and communications equipment.
Through the use of measurement domain algorithms and state-space domain algorithms
(Meuller, 1994) a WADGPS is able to combine the various DGPS corrections received
from reference stations and produce a locally valid single set of DGPS corrections. In this
manner, WADGPS produces a more accurate and reliable set of corrections resulting in a
more accurate and reliable position estimate at the GPS user’s end.
Since WADGPS is not completely dependent on any one reference site, the malfunction
of one or more of the contributing DGPS reference stations will not discontinue the
availability or significantly degraded the accuracy of the DGPS corrections supplied to
GPS users.
GPS SYSTEM ACCURACY
The GPS system chosen for this project was the Trimble Gold Card GPS receiver. The
Gold Card is a 3-channel GPS receiver capable of tracking up to 8 satellites. The stated
accuracy of the system with differential correction is 2-5 meters Circular Error Probable
(CEP), and for non-differentially corrected data 100 meters (CEP). Circular Error
Probable means that 50% of the positions are likely to be within a horizontal circle with a
radius equal to 2-5 meters and 100 meters respectively (Trimble, 1995).
6. There are two methods for acquiring differentially corrected data: real time differential
correction and post-processing differential correction. In real-time differential GPS the
base station calculates and broadcasts through radio telemetry the error for each satellite
as it receives the data. The rover receives this correction and applies it to the position it is
calculating. If the rover receiver loses contact with the base station, it stops computing
positions or computes positions with non-differential GPS accuracy. It is because of this
possible loss of contact and to correct those positions not corrected using real-time
differential operations that GPS post-processing techniques are often used in conjunction
with real time differential correction.
In post-processed GPS, the base station records the pseudo-ranges for each satellite
directly into a computer file. The rover also records its own positions in a computer file.
Pseudo-range is a distance measurement based on the correlation of a satellite transmitted
code and the local receiver’s reference code, that has not been corrected for errors in
synchronization between the transmitter’s clock and the receiver’s clock. After returning
from the field, the two files are processed and the output is a differentially corrected rover
file.
ROAD NETWORK DATA COLLECTION PROCEDURES
Field staff acquired road network data using three different methods. First, staff acquired
GPS road network data using two 100 MHz Pentium laptops. Each system had a Trimble
Gold Card GPS receiver running Aspen’s Field software. The second method was
digitizing 1992 aerial photography at a scale of 1:1,200. Finally, the third method
involved “rubber sheeting” the U.S. Census Bureau’s TIGER (Topologically Integrated
Geographic Encoding and Referencing) line files to the base acquired from the above
methods 1 and 2.
Of Karnes County’s total 1,667 road miles, GPS collected network elements accounted
for 891 miles or 54%, edited TIGER contributed 677 miles or 41%, and digitized aerial
photography made up 99 miles or 5% of the total. Of Frio County’s total 3,957 road
miles GPS accounted for 646 miles or 16%, TIGER contributed 3,279 miles or 83%, and
aerial photography made up 32 miles or 1% of the total.
It should be noted that the above statistics for TIGER will be reduced when private roads
that do not contain any valid structures are deleted from the database. Deletion of these
roads will occur once the structures have been digitizing from U.S. Department of
Agriculture aerial photography.
Table 4 presents a breakdown of the road mileage by acquisition method. The GPS
network elements consisted primarily of county, state, and federally maintained roads.
TIGER network elements consisted primarily of unnamed private roads. Lastly, aerial
photography contributed mostly road network elements inside city boundaries and un-
incorporated communities.
Table 4: Mileage by Acquisition Method
7. Karnes County Frio County
Mileage % of Total Mileage % of Total
GPS 891 54 646 16
TIGER 677 41 3279 83
PHOTOGRAPHY 99 5 32 1
TOTAL 1667 100 3957 100
GPS DATA COLLECTION
The first step in this effort was the porting of the MSAG database from UNIX
ARC/INFO v7.03 to the laptop platform by creation of an MSAG and MPAG data
dictionary with Aspen’s Pfinder software.
Data collection using GPS receivers proceeded from 15 January to 29 March 1996 for
Karnes County and from 1 April to 12 July 1996 for Frio County. Addressing
Technicians would begin their week by conducting mission planning with Aspens
QuickPlan software. This software permits the plotting of the number of satellites
available, their elevations or azimuths over time, and the Position Dilution of Precision
(PDOP).
PDOP is an indication of the current satellite geometry. It is the result of a calculation
that takes into account each satellite’s location relative to the other satellites in the
constellation. A low DOP indicates a higher probability of accuracy. A high DOP
indicates a lower probability of accuracy. PDOP consists of three components: HDOP or
the horizontal, latitude and longitude component; VDOP or the vertical, altitude
component; and TDOP or the temporal, clock offset component. A PDOP of 4 or below
gives excellent positions. A PDOP between 5 and 8 is acceptable. A PDOP of 9 or more
is poor. Figure 1 shows PDOP and its components for some of the dates during which
data was collected in Karnes County.
Figure 1: Karnes County DOP Measures
5
4.5
4
3.5
3 PDOP
DOP Masures
HDOP
2.5 VDOP
TDOP
2
1.5
1
0.5
0
8-Feb-96
25-Jan-96
29-Jan-96
31-Jan-96
15-Feb-96
16-Feb-96
22-Feb-96
26-Feb-96
28-Feb-96
4-Mar-96
6-Mar-96
8-Mar-96
13-Mar-96
18-Mar-96
26-Mar-96
10-Apr-96
1-Apr-96
8-Apr-96
Collection Dates
8. The actual collection procedure involved traveling the length of a road collecting network
and structure data with the receivers. As a technician began capturing road network data,
he or she codes a number of items into the database including: PRFXDIR, STNAME,
STSUFFIX, SUFDIR, COMM, AKA, TYPE, SOURCE, and MISC. When the technician
approaches a driveway entrance he/she codes PRFXDIR, STNAME, STSUFFIX,
SUFXDIR, and any other items that may be visible such as NAME. This methodology
was followed in Karnes County where a driver accompanied the Addressing Technician.
Unfortunately, in Frio County a driver was not available and so to expedite matters only
road network elements were collected. The Aspen field software then produced two
Standard Storage Format (SSF) rover files; one contained the coordinates for a road or a
series of roads and the second contained the coordinates for the structure points. Usually
field staff produced a number of these rover files each day. Once back at the field office,
post-processing differential correction was performed on this data.
As previously mentioned, GPS accounted for 646 miles or 16% of the total 3,957 miles in
Frio County. This distance is represented by 73,170 GPS points, of these 37,727 were
real-time differentially corrected, 35,680 where differentially corrected using post-
processing, 298 where not corrected because they were recorded after the base file ended,
and 5 were not corrected due to faulty base data. Data for Karnes County was not
available.
GPS DATA MANIPULATION
The differentially corrected data was then exported into ARC/INFO format and sent to
the home office in San Antonio where it was processed on a SUN SPARC IPX and 5
workstations running ARC/INFO v7.0.4. An Arc Macro Language (AML) program was
then used to generate the GPS road and structure data into line and point coverages. The
line covers were then further processed by establishing editing tolerances and then
CLEANing the coverage. Table 5 lists the tolerances used when processing the three
differently acquired coverages.
Table 5: Editing Tolerances in Feet
Tolerance GPS Aerial TIGER Description
Photo.
Fuzzy 1.64 0.02 1.64 An extremely small distance used to
resolve inexact intersection locations due
to limited arithmetic precision of
computers.
Dangle 410.00 0.0 410.00 Minimum length allowed for dangling
arcs.
Nodesnap 410.00 5.0 410.00 The minimum radial distance within which
two nodes will be joined.
Table 5: Continued
9. Tolerance GPS Aerial TIGER Description
Photo.
Weed 164.00 2.0 164.00 The minimum allowable distance between
any two vertices along an arc.
Grain 164.00 2.0 164.00 A parameter controlling the distance
between vertices on curves.
Snap 410.00 5.0 410.00 The distance within which a new arc will
be extended to intersect an existing arc.
Tic Match N/A 0.4 N/A The maximum distance allowed between
an existing tic and a tic being
digitized.
A buffer was then created around the line cover at 5 meters or 16.40 feet, the stated
minimum accuracy of our differentially corrected GPS data. The covers were then edited
to remove errors not eliminated by the CLEAN operation.
Figure 2 displays the 5 types of errors encountered after the above processes and their
corrections. In these figures the center lines are the actual GPS data while the two
straddling lines represent a 5 meter buffer. The majority of the arc segment errors
occurred at their beginning or end. These errors are attributable to vehicle acceleration,
deceleration, and stopping. These errors manifest themselves as: 1a-1b, extremely small
jagged and overlapping arc segments and associated dangling nodes; 2a-2b, undershoots
and overshoots and their associated dangling nodes. Additional errors include: 3a-3b,
errors caused by roads driven more than once which manifest themselves as braided arc
segments; 4a-4b, errors caused by disturbance to the GPS unit while data collection is in
process, these manifest themselves as small errors in the middle of otherwise straight arc
segments; and 5a-5b, errors caused by a loss of the GPS signal which manifest itself as a
gap of vertices along an arc segment.
Figure 2: Common GPS Errors and Their Corrections
DIGITIZING AERIAL PHOTOGRAPHY - ROAD NETWORK
10. GPS network data collection within the cities and communities was not as effective as in
the rural areas. This was primarily due to three considerations: first, the presence of
structures prevented the unobstructed view of the horizon and thus hampered satellite
acquisition; second, because of the number of structures, multipath became a real
concern; and third, a significant increase in the number of errors associated with
acceleration, deceleration, and stopping.
Through interaction with local electric power cooperatives, a number of aerial photos
were acquired for both counties. The 1992 photography at scales of 1:1,200 and 1:6,000
corresponded to those areas of most interest to the cooperatives, namely the incorporated
cities, and communities.
Of Karnes County’s total 1,667 road miles digitized aerial photography made up 99 miles
or 5% of the total. Of Frio County’s total 3,957 road miles aerial photography made up
32 miles or 1% of the total.
RUBBER SHEETING OF TIGER
Rubber sheeting is a procedure to adjust the features of a non-uniform coverage (data
layer). Rubber sheeting is performed by establishing links representing from- and to-
locations which are then used to define the adjustment.
The primary reason for the use of TIGER was the timely acquisition of private roads.
Since the GPS acquired network consisted of public rural roads and the digitized aerial
photography consisted of public urban roads then subtraction of these roads from TIGER
would leave the private road network. This private road network as already mentioned
accounted 677 miles or 41% of the total 1,667 miles in Karnes County. In Frio County
TIGER contributed 3279 miles or 83% of the total 3,957 miles in the county. Once
extracted this road network was “rubber sheeted” to our GPS and digitized base map.
In an attempt to quantify the accuracy of uncorrected TIGER, a series of buffers were
created from the acquired GPS road network. Three separate buffers at 5, 10, and 15
meters were generated. These buffers were then used to extract non-private TIGER road
lines within each buffer. Table 6 depicts the cumulative mileage and additional
percentages within each of the buffers.
Table 6: TIGER’s Network Accuracy
Karnes County Frio County
Buffer Miles % of Total Miles % of Total
5m 685 73 378 62
10m 155 17 80 13
15m 48 5 55 9
>15m 51 5 92 15
Totals 939 100 605 100
The TIGER non-private road network for Karnes County is distributed as follows 73%
was within 5m. An additional 17% lies within 10m, and 5% more within 15m. In Frio,
11. the distribution is as follows: 62% within 5 meters, an additional 13% within 10, and still
9% more is within 15 meters.
Assuming that these percentages can also be applied to the private road network which
was extracted from TIGER, one can predict that in Karnes County approximately 183
miles (677 extracted TIGER miles x 27%) and in Frio approximately 1,246 miles (3,279
extracted TIGER miles x 38%) will need additional editing to bring the entire database to
an error of no more that 16.40 feet or 5 meters. Please note that the 27% and 38% figures
come from adding the percentages found outside 5 meters using the above table.
DIGITIZING AERIAL PHOTOGRAPHY - STRUCTURE POINTS
Our structure point database will be created by digitizing un-rectified 1:7,920 aerial
photography purchased from the U.S. Department of Agriculture. These photographs will
be registered to our GPS and aerial photography road network through an affine
transformation function. An affine transformation is based on three or more control
points and calculates changes in scale, shift in the x-direction, shift in the y-direction and
any rotation for the output coverage.
Once these points have been digitized they will be used as a base map for correction of
the “rubber sheeted” TIGER road network. Using this point database, the topological
accuracy of the TIGER road network will be analyzed. Those road segments which do
not straddle their respective points will be digitized of the aerial photography, in addition,
those network elements that do not have corresponding structures will be eliminated from
the data set.
POPULATING THE DATABASES
As we have seen, GPS has permitted the automation of much of the E911 database
construction; likewise, GIS has significantly automated much of the E911 database
population tasks. Specifically, two Graphical User Interfaces (GUIs) were created to
address roads and structures. Traditional methods of addressing these features involves
the direct measurement of these features on maps or aerial photographs, calculation of
actual ground units, and then the manual entry of this data into a database.
The GUI which was created to address arcs permits a user to select an arc and then enter
a starting and ending left and right address when addresses have been predetermined such
as in block addressing in urban areas. Or, the user may choose not to enter any data and
allow an algorithm to calculate these values based on the previous ending addresses of
the adjacent arc, and an addressing unit (such as 5.28 feet). Thus, addressing becomes as
simple as pointing to a road element (Tobar, 1996).
Another GUI was created to address structures based on the National Emergency Number
Associations (NENA) RULE 16: Structures which states the following:
Method: When assigning numbers, the middle of the structure should
12. determine the number. Structures should always be numbered according
to the road they face, not where the driveway enters the road or where the
mailbox is. An exception to this is when the house can not be seen from the
road, then the driveway should be numbered and addressed on the road
from which it departs (Lucy, 1995).
The algorithm involves the following steps. Using an input point cover, calculate the
nearest point on the nearest arc. Split the arc at this point creating two arcs: one with
correct L-ADD.FROM and R-ADD.FROM items and a low ID, and a second with
incorrect L-ADD.FROM and R-ADD.FROM items and a high ID. Select the arc with the
lowest ID, extract its length, and check topology. Calculate the structure’s address by
dividing this arc length by an addressing unit (in this case 5.28 feet). Lastly, un-split the
arc to return it to its original state. This algorithm has been timed at 2 minutes per point
on a SUN Sparcstation 5 which is a considerable improvement over manual methods
which can take over 5 minutes per point. A rough estimate of the number of points which
will be coded is given by two databases provided to us from local electric cooperatives.
These databases identify all locations where the cooperatives maintain meters for
measuring electrical use. These locations include residential and non-residential
structures. The database for Karnes County includes some 3,425 points or 4.75 days of
processing, while the database for Frio County includes 3,117 points or 4.33 days of
processing.
CONCLUSION
The use of GPS on a laptop platform has significantly automated the Enhanced 911
database construction in Karnes and Frio Counties. The primary benefit is that data
collected in the field is logged directly into a computer along with GPS location data.
This information, through custom AMLs, can be loaded directly into our E911 GIS and
thus eliminates re-typing of the data. Also, since the GPS data is matched with the data
point, error has been reduced. Another benefit is that the GPS data collected for mapping
has been used to correct TIGER road topology. The use of these technique has eliminated
the need for expensive photogrammetry, which can also be inaccurate and still require
field validation. In addition, our GIS has served to automate many time consuming tasks.
In particular, the time usually taken to address roads and structures has been significantly
reduced through the use of custom GUIs.
REFERENCES
Abousalem, M. A., et. al.,” International Wide Area Differential GPS Networks”, ION
95, January 1995, Anaheim, CA.
ESRI, ARC/INFO Data Model, Concept, & Key Terms, Redlands, CA, 1991.
Hurn, J., GPS: A Guide to the Next Utility, Trimble Navigation, Sunyvale, CA, 1989.
Lucy, W. M., Addressing Systems: A Training Guide for 911, National Emergency
13. Number Association (NENA), Coshocton, OH, 1995.
Meuller, T., “Wide Area Differential GPS”. GPS World, Vol. 5, No. 6, June 1994, pp.
36-44.
Ozanich, B., E911 Data Base Guide: Building and Maintaining an E9-1-1 Data Base,
National Emergency Number Association (NENA), Coshocton, OH, 1996.
Ozanich, B., “Mapping Wireless 911 Calls”. Geo Info Systems, Vol. 6, No. 7, July 1996,
pp. 54-56.
State of Texas, Advisory Commission on State Emergency Communications (ACSEC),
Addressing Handbook for Local Governments, ACSEC, Austin, TX, August 1991.
Tobar, J., “Addressing via a menu”. Point Line Poly, Vol. 5, No. 3, 1996, pp. 7-13.
Trimble Navigation, Aspen-GPS: System Operation Manual, Sunyvale, CA.