See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/241809034

Using GIS to Measure the Effects of Service Area

and Frequency on Passenger Boardings at Bus
Stops

Article

CITATIONS READS

18 425

3 authors, including:

Kenneth J Dueker Ahmed M. El-Geneidy

Portland State University McGill University
79 PUBLICATIONS 1,567 CITATIONS 153 PUBLICATIONS 2,373 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Mobile Cities View project

All content following this page was uploaded by Ahmed M. El-Geneidy on 23 February 2014.

The user has requested enhancement of the downloaded file.

 Using GIS to Measure the Effect of Overlapping Service
Areas on Passenger Boardings at Bus Stops
Thomas J. Kimpel, Kenneth J. Dueker, and Ahmed M. El-Geneidy

Abstract: This study examines the effects of overlapping walking service areas of bus stops on the demand for bus transit. This
requires controlling for variation in potential transit demand as measured by the number of dwelling units and their loca-
tions. A model of passenger boardings for the morning peak hour of service is estimated. Boardings are modeled as a function
of potential transit demand at the level of the individual bus stop. To address overlapping bus stop service areas, a geographic
information system is used to measure the accessibility of each parcel to each bus stop relative to other accessible stops. A distance
decay function is empirically estimated and used to calculate walking accessibility from dwelling units to bus stops. This stop-
level boarding model is an improvement over methods in which ridership is typically related to potential transit demand using
one-quarter-mile service areas under the assumption of uniform density of demand, often with little or no consideration given
to double counting.

INTRODUCTION is related to potential transit demand by 1) intersecting census

This study examines the effects of overlapping bus stop service block groups with bus stop buffers and using areal interpolation to
areas on the demand for transit at the bus stop level. Potential calculate population or 2) counting the number of housing units
transit demand is measured at the most disaggregate level, in terms within stop buffers. These methods are based on the questionable
of the number of dwelling units per parcel of land. Transit supply assumption of uniform density of demand to allocate population
is measured spatially at the individual bus stop, and temporally or housing units to transit service areas. The approach used in
for the morning peak hour of service. A GIS-based approach is this study disaggregates potential transit demand to the stop level
used to measure accessibility of dwellings at the parcel level to and relates it to actual morning peak hour bus boardings at each
the nearest bus stop. The distance decay parameters of the acces- bus stop although the data are aggregated to average boardings
sibility function are empirically derived by varying intercept and per trip in the morning peak hour.
slope values systematically using ordinary least-squares regression.
Demand at the bus stop level, as measured by average morning BACKGROUND
peak hour boardings, is related by regression to a measure of ac- A review of the existing literature shows that stop-level transit
cessibility-weighted dwelling units that controls for competing demand is modeled from a spatial standpoint. Miller and Shaw
bus stops. (2003) stress the need for understanding the underlying spatial
This examination of walking distance to bus stops focuses assumptions as they relate to GIS transportation analysis. A
on potential transit demand from a residential standpoint using a number of researchers have empirically analyzed walking dis-
measure of integral accessibility (Makri and Folkesson 1999, Song tance to transit stops (Neilson and Fowler 1972, Levinson and
1996). The study focuses on inbound radial routes in the morn- Brown-West 1984, Hsiao et al. 1997, Zhao et al. 2003) based
ing peak time period serving close-in urban neighborhoods—a on information derived from passenger surveys. These studies
route type, service direction, and time period in which demand found that the relationship between transit demand and walk-
is primarily associated with residential boardings. ing distance is expressed as a negative exponential distance decay
A one-quarter-mile walking distance is a well-known rule of function. The findings from these studies suggest 1) that passenger
thumb in transit service planning. In most instances, bus stops demand decreases with respect to walking distance to stops and
are spaced closer than a quarter mile, creating overlapping bus 2) that a one-quarter-mile bus stop service area will not capture
stop service areas on the same route. In many areas, parallel bus all potential transit users while a larger service area will result in
routes are spaced at distances less than one-half mile, creating an overestimation of the number of potential riders if distance
overlapping service areas between routes that often operate at decay is not explicitly addressed.
different service frequencies. To control for these overlapping GIS techniques have been used to relax the assumption of
service areas, a geographic information system (GIS) is used to uniform density to prorate potential transit demand to transit
measure the accessibility of each parcel to bus stops within walk- service area buffers (Peng and Dueker 1998). Instead of uniform
ing distance and the integral accessibility of each bus stop to density, O’Neill et al. (1992) used street density, while Zhao
dwelling units within walking distance to the stop. Deriving and (1998) used dwelling units from a parcel database as the basis for
including distance decay parameters in the accessibility measure assignment. Also, Zhao addressed barriers to walking and used
is an improvement over traditional methods in which ridership network distance rather than straight-line distance to define transit

URISA Journal • Kimpel, Dueker, El-Geneidy 

service areas around bus stops. While these GIS approaches serve meier 1997). The relative accessibility to transit service using a
to more accurately measure potential transit demand, they are not gravity-based measure is obtained by weighting opportunities of
related to actual transit ridership. attraction for transit users (e.g., service frequency) and discounting
Rather than using ridership data based on passenger surveys, this attraction by a negative exponential or a Gaussian impedance
econometric models typically use a sampling of actual passenger measure based on distance. In this analysis, we use integral acces-
boardings. Most previous studies seeking to explain the determi- sibility to transit to address the overlap in service areas. Integral
nants of transit demand have been conducted at either the route accessibility is the sum of relative accessibility over all possible
(Kemp 1981, Horowitz 1984, Azar and Ferreira 1994, Hartgen destinations divided by the total attraction of the bus stop being
and Horner 1997) or route-segment (Peng and Dueker 1995, studied (Song 1996).
Kimpel 2001) levels. Stop-level transit demand has been discussed In addition to issues of overlapping service areas and distance
in the literature as being the most appropriate level of analysis decay in stop-level demand modeling, a third issue concerns ser-
(Peng and Dueker 1995, Kimpel 2001, Furth et al. 2003), and vice quantity. Besides spatial proximity to bus stops, passengers are
implemented in T-BEST (Chu, 2004). The use of automatic pas- also concerned with the availability of service across the temporal
senger counters at transit agencies increasingly supports this type dimension (Kittelson and Associates 2003) because it influences
of modeling because an abundance of high- quality ridership data wait times at transit stops. A measure of service quantity such as
can be collected at relatively low cost. (see Furth et al. [2003] for the number of buses per hour passing a given location is needed
a discussion of transit data collection technologies). to capture any variation in the level of service between stops on
Bus stops are typically located and spaced according to a transit the same route as well as between stops on competing routes. In
agency’s service standards. Ammons (2001) looked at bus stop the former case, certain bus stops will have higher service levels
spacing standards for a number of transit properties and found that compared to others because of varying service patterns (e.g., regu-
stop spacing typically ranges from 656 to 1,968 feet in urban areas. lar, limited, and express service). In the case of overlapping bus
Such small distances between stops leads to overlapping bus stop stop service areas on different routes serving the same destination,
service areas on the same route as well as with stops on adjacent choice riders would most likely walk to the bus stop associated
routes serving similar destinations. In prior research, competition with the greater service frequency certis paribus. The review of the
for choice riders was addressed at the route-segment level by Peng literature shows the strength of GIS-based methods, the need for
and Dueker (1995) and Kimpel (2001) through different means. a distance decay-weighted measure of potential transit demand
In the former study, competition was addressed in the modeling at the bus stop level, and the need to relate demand to automatic
stage using an explanatory variable based on the percent area of a passenger counter–generated passenger boardings. This research
buffer subject to overlap. In the latter study, competition was ad- builds on these developments and estimates a descriptive model at
dressed during the data-processing stage by proportionally assign- the disaggregate level—passenger boardings at bus stops averaged
ing potential demand in overlapping service areas using secondary over all trips in the morning peak hour. This is similar to planning
information derived from disaggregate data (tax parcel value) as models such as T-BEST, which is a stop-level model that also
the basis for allocation. One of the primary reasons that stop-level aggregates trips to time periods and identifies potential demand
demand models are lacking is because of the exceedingly complex using a buffering technique, but does not address distance decay.
difficulties associated with allocating potential transit demand in Our parcel-based accessibility measure incorporates the size effect
overlapping transit service areas to specific stops. Although the use (number of housing units), the likelihood of waiting at a bus stop
of a GIS to solve problems related to transit accessibility is now (scheduled headway), and a distance decay function.
fairly common, only a few researchers have adequately addressed
overlapping service areas in a manner consistent with theory and STUDY DESCRIPTION
only at spatial levels higher than the level of the bus stop. Also The study uses data from three sources. TriMet, the regional
notable is that none of the econometric studies have addressed the transit provider for the Portland metropolitan region has auto-
issue of distance decay but instead have relied on the assumption matic vehicle location (AVL) and automatic passenger counter
of a uniform density of demand within transit service areas. In the (APC) technologies on most of the fixed-route bus fleet collecting
present analysis, rather than using an arbitrary one-quarter-mile boarding and alighting information as well as service reliability
service area buffer, we use an initial distance of one-third mile and information at each bus stop. Metro, the regional transportation
then apply a distance decay function that is presented in more detail and land-use planning organization, distributes GIS data for bus
later. We utilize a network-based method for determining transit stops, bus routes, and tax parcels on a quarterly basis as part of
service areas using a GIS and undertake an analysis that addresses the Regional Land Information System. The Multnomah County
overlapping service areas through measurement of integral acces- tax assessment database was used to obtain information on the
sibility at the tax parcel level. number of units associated with multifamily parcels.
Accessibility is a measure of potential opportunities for Boardings associated with the morning peak hour of service
interaction (Hansen 1959). While accessibility can be calculated (7:30 A.M. to 8:30 A.M.) for two routes for 69 stops were ob-
in various ways, the gravity-based measure of accessibility is the tained from TriMet. The routes of interest are the 14 Hawthorne
most widely used measure in planning studies (Handy and Nie- and the 15 Belmont, two radial routes connecting southeast

 URISA Journal • Vol. 19, No. 1 • 2007

Figure 1. Study area Figure 2. Overlapping bus stop service areas

Portland with the Central Business District. The study area en- ONSXj = f {DWDUj}
compasses an inner-city area that is well served by bus transit that (1)
is well patronized. Nine months of data associated with weekday where:
service yielded approximately 126,000 data points. The study ONSXj = average passenger boardings per trip at stop j in the
stops were limited to those located between I-205 and S.E. 12th morning peak hour over all days;
Avenue. Stops that could attract patronage from other sources
such as transfer and park-and-ride locations rather than the sur- DWDUj = ∑i (exp(a – bdij)/(1+exp(a – bdij)) * DUi) = the sum
rounding neighborhoods were eliminated from consideration. of distance-weighted dwelling units associated with stop j ex-
The study area and the bus stop service areas within one- pressed as a probability using a negative logistic distance decay
third-mile walking distance along the street network are presented function;
in Figure 1. Note the prevalence of overlapping bus stop service
areas on the same route as well as between routes. The distribution where:
of dwelling units associated with parcels in relation to three bus dij = on-street distance in miles from parcel i to stop j; and
stops is presented in Figure 2. The different colored areas represent
locations where parcels have access to one or more stops. DUi = dwelling units at parcel i.

Distance Decay Function The estimated probabilities for several of the logistic func-
Zhao et al. (2003) fit a negative exponential function to survey tions exp(a – bd), Zhao et al.’s exponential function exp(-6.864d),
data of walking distance to transit stops. Others use an arbitrary and the uniform density of demand assumption (UDD) where p
one-quarter-mile service area buffer, in which the probability = 1 for d <= 0.25 miles and p = 0 for d > 0.25 miles are shown in
of demand falls from one to zero at exactly a one-quarter-mile Table 1. Figure 3 shows this information graphically.
distance. Similar to Vuchic (2005), we posit something in Parameters a = 2 and b = 15 were selected as the best repre-
between—that a negative logistic function of the form exp(a sentation of distance decay using the negative logistic function
– bdij)/(1+exp(a – bdij)) is better suited for distance decay of since this particular model provided the best fit of the data. This
transit demand to reflect a more gradual decline in transit demand parameter set depicts a steep distance decay prior to one-quarter
at short distances, a steeper decline as distance approaches one- mile. At short distances the probability of taking the bus is high,
quarter mile, and a more gradual tail. We estimated the distance while at distances approaching one-quarter mile the probability
decay function by empirically analyzing multiple sets of intercept is low.
(a) and slope (b) parameters in a series of ordinary least-squares Our approach to estimating the walking distance decay func-
regression models of transit demand allowing us to identify the tion is indirect. The direct approach requires information about
parameter set that maximizes goodness of fit. The estimation of where each transit rider lives and which particular stop he or she
the distance decay function utilized distance to the nearest stop accesses. This knowledge is often gained by means of an onboard
and does not include accessibility to more than one stop. The survey of transit riders; however, this technique normally yields
following model specification was used to empirically derive the sample sizes that are too small for subsystem analyses (e.g., stop,
parameters: corridor, or route level). Instead, our indirect approach involves

URISA Journal • Kimpel, Dueker, El-Geneidy 

Table 1. Estimated Probabilities for Various Distance Decay Functions
Negative Negative Uniform
Logistic Exponential Density
Parameters/
5-23d 4-21d 3-22d 2-22d 2-15d -6.864d UDD
Distance
d = 0.10 mile 0.9370 0.8699 0.6900 0.4502 0.6225 0.5034 1.0000
d = 0.20 mile 0.5987 0.4502 0.1978 0.0832 0.2689 0.2534 1.0000
d = 0.25 mile 0.3208 0.2227 0.0759 0.0293 0.1480 0.1798 1.0000
d = 0.30 mile 0.1301 0.0911 0.0266 0.0100 0.0759 0.1276 0.0000
d = 0.40 mile 0.0148 0.0121 0.0030 0.0011 0.0180 0.0642 0.0000

in GIS-based analysis of transit demand. Although people can

walk that distance, most transit riders do not. Thus, a quarter-mile
transit buffer overestimates the population thought to be served
by transit and lends support for bus stop spacing standards that
call for relatively short distances between stops.

Accessibility-Weighted
Demand Model
With the empirically estimated parameters for distance decay,
another demand model is estimated for the case of overlapping
bus stop service areas using a measure of integral accessibility.
The average number of passenger boardings per trip per bus
stop during the morning peak hour is modeled as a function of
potential transit demand at the level of the individual bus stop
controlling for overlapping bus stop service areas. Our model
controls for variation in potential transit demand as measured
by the number of dwelling units and their location (by distance
from all bus stops within walking access) as well as the amount of
scheduled service provided at stops. The following specification
was used for the model:

ONSXj = f {AWDUj} (2)

Figure 3. Estimated demand probabilities where:

ONSXj = average passenger boardings per trip at stop j in the
morning peak hour over all days;
estimating the distance decay function parameters by relating ac-
tual boardings to distance-weighted dwelling units by means of an AWDUj = ∑i ((Aij / ∑j Aij) * exp(a – bdij)/(1+exp(a – bdij)) * DUi
iterative fitting process using ordinary least-squares regression. = accessibility-weighted dwelling units around stop j;
While the model with the exponential function had the
highest R2, the results across the models did not vary that much where:
with values ranging from 0.285 to 0.315. The parameters a = 2 Aij / ∑j Aij = integral accessibility or proportion of accessibility at
and b = 15 yielded the best R2 value of all the negative logistic parcel i attributable to stop j;
functions; however, the low intercept value of a = 2 makes our
function similar to the exponential function estimated by Zhao et where:
al. Ridership is quite sensitive to distance, but the various measures Aij = accessibility of parcel i to bus stop j = exp(a – bdij)/(1+exp(a
of distance-weighted dwelling units were nearly indistinguishable, – bdij)) * BUSHRj * DUi;
perhaps because of the simplifying assumption of distance to
nearest stop. Nevertheless, our “best” distance-decay function is where:
consistent with prior research that shows distance decay that starts exp(a – bdij)/(1+exp(a – bdij)) = probability of taking transit
close, is steep, and has a long tail. Similarly, our estimation does based on the negative logistic distance decay function using the
not support the use of a quarter-mile buffer that is commonly used parameters a = 2 and b = 15;

 URISA Journal • Vol. 19, No. 1 • 2007

where:
dij = on-street distance in miles from parcel i to stop j;

BUSHRj = scheduled service measured by buses per hour at

stop j;

DUi = dwelling units at parcel i; and

∑j Aij = accessibility of parcel i to all stops j within 1/3 mile of

parcel i.

The integral accessibility (Aij / ∑j Aij) of parcel i to stop j is

a key concept in this research. It measures the share of parcel i
demand that is allocated to bus stop j, where the denominator
Figure 4. Measures of parcel accessibility
(∑j Aij) measures the accessibility of parcel i to all stops within
walking distance. The accessibility of parcels to all walking ac-
cessible stops is shown for parcels associated with bus stop 2606
in the first panel of Figure 4. The second panel shows walking
accessibility to stop number 2606 without considering overlap. are aggregated over all 69 study stops so considerable variation
More intense colors indicate a combination of nearness and in potential demand at any given stop may exist, depending on
density. The third panel of Figure 4 takes overlapping bus stop which particular distance decay function is used.
service areas into consideration. The third panel shows the effect Table 3 contains the descriptive statistics for the variables
of applying integral accessibility (Aij / ∑j Aij) of stop 2606 times used in the accessibility-weighted dwelling unit model and the
the accessibility (Aij) of stop 2606, the result of which we call other comparative models. Table 4 contains the results of the
accessibility-weighted dwelling units at parcel i attributable to regressions.
bus stop j (AWDUij). The results in Table 4 show that the accessibility-weighted
The number of distance-weighted dwelling units for the 69 dwelling unit (AWDU) model performs better than do the
study stops according to the uniform density of demand assump- comparison models.
tion, the negative exponential function derived by Zhao et al. The parameter for the number of dwelling units, control-
(2003), the negative logistic function using the parameters a = 2 ling for integral accessibility, 0.0147 boardings per accessibility-
and b = 15, and the same negative logistic function controlling weighted dwelling unit, is used to estimate morning peak hour
for integral accessibility are shown in Table 2. By incorporating boardings at stops on a per-trip basis for counts of accessibil-
distance decay, potential transit demand is shown to decrease ity-weighted dwelling units. The results of this simulation are
by a factor of approximately 2x using the negative exponential shown in Table 5.
function and the two negative logistic functions relative to the
traditional one-quarter-mile buffer method. Potential demand CONCLUSIONS
is higher relative to the negative exponential decay function for The research examined the determinants of transit boardings,
the negative logistic function using nearest stop criterion and taking advantage of automatically collected passenger data at
lower based on the notion of integral accessibility. These results bus stops. A tax parcel layer database was used as the basis
for calculating potential transit demand at each stop using the
Table 2. Distance-Weighted Dwelling Units
Decay Function Assumption Distance (feet) Units
UDD Nearest stop 1,320 10,854
DWDU (Neg. Exponential) Nearest stop 1,760 4,937
DWDU (Neg. Logistic) Nearest stop 1,760 5,601
AWDU (Neg. Logistic) Integral accessibility 1,760 4,266

Table 3. Descriptive Statistics for Accessibility-Weighted Dwelling Unit (AWDU) Model and Comparison Models
Name Mean Std. Dev. Var. Min. Max
ONSX 0.92 0.68 0.46 0.02 2.76
UDD 157.36 85.87 7374.40 17.00 391.00
DWDU (Neg. Exponential) 71.55 38.92 1515.00 10.19 194.53
DWDU (Neg. Logistic) 81.18 44.38 1969.10 11.33 210.36
AWDU (Neg. Logistic) 61.83 28.24 797.30 16.86 150.16

URISA Journal • Kimpel, Dueker, El-Geneidy 

Table 4. Model Results for Accessibility-Weighted Dwelling Unit (AWDU) Model and Comparison Models
Coef. Std. Err. T-ratio Adj. R2 Constant
UDD 0.0044 0.0008 5.5714 0.3064 0.2167
DWDU (Neg. Exponential) 0.0099 0.0018 5.6833 0.3152 0.2049
DWDU (Neg. Logistic) 0.0086 0.0015 5.5939 0.3082 0.2159
AWDU (Neg. Logistic) 0.0147 0.0069 6.3350 0.3652 0.0069

Table 5. Simulation of Stop-Level Boardings Using Accessibility-Weighted Dwelling Units

Accessibility-Weighted Dwelling Units Per Stop Estimated Stop-Level Boardings Per Trip During Morning Peak Hour
25 0.368
75 1.104
100 1.472
150 2.207

measure of integral accessibility that takes into consideration About The Authors
distance-weighted accessibility and competing stops. The analysis
was confined to the morning peak hour, when transit demand is Thomas J. Kimpel is a research associate in the Center for Ur-
most directly related to dwelling units. ban Studies, Portland State University, where he has been
Data preparation required the use of a GIS, which consisted employed since 1996. His areas of interest include GIS
of snapping dwelling units from parcel centroids to abutting analysis, transportation and land-use planning, and bus
streets, computing distance on the street network to all bus stops transit performance monitoring.
within one-third-mile distance, computing integral accessibility
Corresponding Address:
of dwelling units to those stops, and summing the integral acces-
Thomas J. Kimpel
sibility of dwellings for each bus stop.
Center for Urban Studies
Distance decay parameters of the accessibility function were
Portland State University
empirically derived from ordinary least-squares regression models
506 S.W. Mill Street, Room 350
by varying intercept and slope values. These parameters were then
Portland, OR 97201
used to estimate a stop-level bus boarding model using accessibil-
Phone: (503) 725-8207, Fax: (503) 725-8480
ity-weighted dwelling units. The number of accessibility-weighted
E-mail: [email protected]
dwelling units is positively related to the number of boarding pas-
sengers. The parameter on this variable can be used to estimate
Kenneth J. Dueker, Professor Emeritus of Urban Studies and Plan-
morning peak hour transit ridership at the bus stop level.
ning, Portland State University, is an experienced educator
This research illustrates the power of analysis using detailed
and researcher in transportation. He directed the Center
disaggregate data, boardings at the bus stop level, and for parcel-
for Urban Studies at PSU from 1979 to 1998. His areas of
level counts of dwelling units. A GIS analysis was needed to relate
interest include transportation and land-use interactions,
dwelling units to the street network and to calculate distances
travel and parking behavior, and GIS transportation.
to bus stops. A distance decay function was derived and used
to compute an accessibility measure to account for overlapping Corresponding Address:
bus stop service areas for an improved estimation of stop-level Kenneth J. Dueker
transit demand. Center for Urban Studies
It is important to note that distance decay parameters may Portland State University
not be constant; they may vary by trip purpose and access mode. 506 S.W. Mill Street, Room 350
In the future, it is recommended that more reliable distance Portland, OR 97201
decay parameters be estimated from passenger intercept surveys Phone: (503) 725-4040, Fax: (503) 725-8480
conducted at bus stops. These surveys can ask transit users about E-mail: [email protected]
their point of origin, trip purpose, destination, access mode, and
whether they will undertake a transfer. It is expected that decay
curve parameters will vary based on these factors. Accordingly,
a better transit demand model can be generated.

10 URISA Journal • Vol. 19, No. 1 • 2007

 Ahmed M. El-Geneidy is a postdoctoral research fellow in the Hsiao, S., J. Lu, J. Sterling, and M. Weatherford. 1997. Use of
Department of Civil Engineering and the Humphrey Insti- geographic information system for analysis of transit pedes-
tute of Public Affairs, University of Minnesota. He formerly trian access. Transportation Research Record 1604: 50-59.
worked under the auspices of the Center for Urban Studies Kemp, M. A. 1981. A simultaneous equations analysis of route
as a graduate research assistant. His areas of interest include demand and supply, and its application to the San Diego
GIS analysis in transportation planning and bus transit bus system. Washington, D.C.: UMTA, Report DTUM-
performance monitoring. 60-80-71001.
Corresponding Address: Kimpel, T. J. 2001. Time point-level analysis of transit service
Ahmed M. El-Geneidy reliability and passenger demand. Portland, OR: Unpub-
Department of Civil Engineering and lished Doctor of Philosophy in Urban Studies, Portland
Humphrey Institute of Public Affairs State University.
University of Minnesota Kittelson and Associates. 2003. Transit capacity and quality of
500 Pillsbury Drive S.E. service manual. Washington, D.C.: U.S. Department of
Minneapolis, MN 55455 Transportation.
E-mail: [email protected] Levinson, H. S., and O. Brown-West. 1984. Estimating bus rider-
ship. Transportation Research Record 994: 8-12.
Makri, M., and C. Folkesson. 1999. Accessibility measures for
References analyses of land-use and travelling with geographical infor-
mation systems. Department of Technology and Society,
Ammons, D. N. 2001. Municipal benchmarks: assessing local Lund Institute of Technology, Sweden, 1-17.
performance and establishing community standards. Thou- Miller, H., and S. Shaw. 2001. Geographic information systems
sand Oaks, CA: Sage Publications. for transportation: principles and applications. New York:
Azar, K. T., and J. F. Ferreira. 1994. Integrating geographic Oxford University Press.
information systems into transit ridership forecast models. Neilson, G., and W. Fowler. 1972. Relation between transit
Journal of Advanced Transportation 29(3): 263-79. ridership and walking distances in a low-density Florida
Chu, X. 2004. Ridership models at the stop level. Tampa, FL: retirement area. Highway Research Record 403: 26-34.
Center for Urban Transportation Research, University of O’Neill, W. A., R. D. Ramsey, and J. Chou. 1992. Analysis of
South Florida, Report No. BC137-31. transit service areas using geographic information systems.
Furth, P. G., B. Hemily, T. H. Mueller, and J. G. Strathman. Transportation Research Record 1364: 131-38.
2003. Uses of archived AVL-APC data to improve transit Peng, Z., and K. J. Dueker. 1995. Spatial data integration in route-
performance and management: review and potential. Wash- level transit demand modeling. URISA Journal 7: 26-37.
ington, D.C.: Transportation Research Board. TRCP Report Song, S. 1996. Some tests of alternative accessibility measures:
No. 23 (Project H-28), http://gulliver.trb.org/publications/ A population density approach. Land Economics 72(4):
tcrp/tcrp_webdoc_23.pdf. 474-82.
Handy, S. L., and D. A. Niemeier. 1997. Measuring accessibility: Vuchic, V. 2005. Urban transit: operations, planning and econom-
An exploration of issues and alternatives. Environment and ics. Hoboken, NJ: John Wiley and Sons, Inc.
Planning A 29(7): 1,175-94. Zhao, F. 1998. GIS analysis of the impact of community design
Hansen, W. 1959. How accessibility shapes land use. Journal of on transit accessibility. Proceedings of the ASCE South
the American Institute of Planners 25(2): 73-76. Florida Section 1998 Annual Meeting, Sanibel Island,
Hartgen, D., and M. W. Horner. 1997. A route-level transit Florida, 1-12.
ridership forecasting model for lane transit district: Eugene, Zhao, F., L. Chow, M. Li, I. Ubaka, and A. Gan. 2003. Forecasting
Oregon. Charlotte, NC: Center for Interdisciplinary Trans- transit walk accessibility: regression model alternative to buf-
portation Studies, Report No. 170. fer method. Transportation Research Record 1835: 34-41.
Horowitz, A. J. 1984. Simplifications for single-route transit
ridership forecasts. Transportation 12: 261-75.

URISA Journal • Kimpel, Dueker, El-Geneidy 11

View publication stats